Corpus-Paul

["David C. Hindley" <DHindley AT compuserve.com>]

On 05/03/99, ""Sheila E. McGinn, Ph.D." <smcginn AT jcu.edu>" wrote:

> "Stephen C. Carlson" wrote:
> 
> > At 08:53 PM 5/2/99 -0400, John C. Hurd wrote:
> > >    And closer to home is Kenneth J. Neumann's, "The Authenticity of the 
> > > Pauline
> > Epistles in the Light of Stylostatistical Analysis"  ("SBL dissertation 
> > series, 120;
> > Atlanta, GA:  Scholar's Press, 1990).  He was
> > >my student, and I was his director and am his friend.
> >
> > Would you (or anyone else) care to summarize Neumann's conclusions?

> I did a critical review of this piece a few years back (when it first came 
> out).  I
> will be happy to send it to the list as a WP attachment.  However,  I want 
> to warn
> folks in advance, for the sake of those who are careful with such things 
> and may want
> to delete the message rather than reading it.  I will use the same subject 
> heading as
> on this message.  (By way of reassurance, I have just run my virus-checker 
> and my PC
> is clean.)  John Hurd may then want to respond to the review, or send his 
> own
> comments.
> 
> Sheila
> 

I had a little trouble retreiving the message, as I get my posts via the 
digest, and Lyris seems to convert some data weirdly. However, I cleaned it 
up in MS Word and found the review very interesting.

Years ago I picked up a copy of A. Q. Morton & James McLeman's _Paul, The 
Man and the Myth_ (1966). The analysis was relatively crude, even by the 
standards of the 60's, and I did not quite follow the methods they employed 
to determine the statistical significance of the relationships they 
detected.

I agree with you that "Neumann's use of multileveled multivariate analysis 
is a tremendous methodological advance over previous attempts at stylistic 
analysis of the Biblical material." When I encountered the book (early 
1980's, I think) I wondered why something like multileveled multivariate 
analysis had not been applied to the data. However, when I encountered this 
analytic tool (1978) it was so new that my professor could not even find a 
textbook that dealt with it (substituting photocopies of a published 
example study that employed it), so perhaps it was not available to Morton 
& McLeman.

My comments on your review of Neumann's book are these:

>> Neumann employs the population model, which entails applying various 
indices of style to six of the undisputed letters (the Hauptbriefe, 1 
Thess, Phil; Phlm is omitted because of its brevity), and then to a group 
of texts which definitely are not Pauline (initially Heb, 1 Clem, and 
letters of Ignatius, Philo, Epictetus, and Josephus). <<

>> Finally the disputed letters are measured "by those indices which 
clearly distinguish the various authors, especially those setting off the  
Pauline from the non Pauline literature. These indices combined by the 
multivariate technique of discriminant analysis should indicate the 
probability that the letters belong with the Pauline corpus or with the 
writings of other authors." <<

It appears to me that Neumann is setting up the analysis to prove what he 
has already determined. By using pre-defined sets of undisputed and 
disputed letters, isn't he really comparing the sets rather than 
identifying authorship?

As for the level of statistical significance, the F-test is definitely the 
proper tool to use for multileveled multivariate analysis. Whether we have 
a valid samlple size, well that is another. We need to know the size of the 
universe population (that is, the totality of everything written by Paul), 
and of course we do not know that at all. Consequently, we can only 
determine the significance between samples.

Dave Hindley
Cleveland, OH
Dhindley AT compuserve.com