Date: Sun, 29 Mar 1998 16:50:38 +0000
From: Stevan Davies <miser17@epix.net>
To: crosstalk@info.harpercollins.com
Subject: Re: Original GThomas


Mike Grondin wrote:
> Kevin Johnson has been so admirably precise in the wording of his
> response to Paul Miller, that I can hardly find anything with which to
> disagree.

Yes indeed. I will be nice to hear more from Kevin.

> (1) Keyword associations: I don't deny that one can find plenty of
> "keywords" in the text which seemingly tie adjacent sayings together
> (and this has been done). But there are _also_ plenty of "keywords"
> which are shared by _non-adjacent_ sayings. The Coptic verb for
> 'amazed', for example, is found in sayings number 2 and 29, and in them
> alone. If necessary, I think I could furnish quite a lengthy list of
> such examples, but the point is this: given that there are many
> "keywords" shared by adjacent sayings, and many others shared by
> NON-adjacent sayings, has the frequency of the former been adjusted for
> the frequency of the latter? If not, how is one entitled to cite the
> (non-adjusted) frequency of the former as indicative of anything?

Wow. That's a brilliant comment. Or, should I say, that's something
I've never quite figured out on my own. I've known there was 
something wrong with the "catchword" business but I never could
put my finger on it. But what you say is just exactly right. The 
catchword theory is a statistical argument. As such the theory
needs to be set out as you would have it. Questions like  "how do
we know this isn't happening by chance?" are ANSWERABLE 
questions and should be answered rather than ignored.

Alas, I once did this sort of thing with the question "how do we know
that Thomas' sayings are randomly distributed vis a vis the
synoptics?" more than to say "it looks that way to me." Used
a statistical package and checked it out with colleagues who know
statistics. Sent the thing to CBQ and it was rejected on the grounds 
that the editors couldn't think of who could even review it 
for possible publication. (It's on my website if anybody cares.)

Vis a vis Arnal's argument (more on which will follow) it occurred
to me that IF the catchword organization in Thomas is real, and 
there are two strata, one would not expect catchword connections
to cross the strata. But they do in 20/21, 26/27, 61/63, and
multiply in 107/108/109/110/111 (cf. Patterson's list).

Steve