testing the synoptics
- I repeat - I am calling that material common only to Matthew and Luke q, and
that material found only in Matthew M and that common to Matthew & Mark m and
then testing to see whether there are, or are not, significant stylistic
differences once allowance is made for genre. If any of the blocks is
itself diverse that may well show up - as it seems to in the M parables.
The problem of m q overlaps might tend to reduce any differences between some
of the samples - samples containing overlap materials could be looked at to
see if they are closer than usual to the other block.
I had considered the possible auxiliary hypothesis that Matthew had made his
sources irrecoverable. This is often said about the 4th Gospel but some
stylometric work in progress elsewhere seems to suggest that that is not so,
as does Fortna's second book on this topic. I am not very happy with
untestable auxiliaries. If the result really suggested this I would rather
say other evidence suggests conclusion x but this evidence conclusion y and
that we would need to devise some other test to resolve this. But it is too
easy to dream up scenarios which may not happen. It is better to go and look.
What is not quite so easy is to decide how diverse is diverse. At present I
am tending towards this measure. Let us partition each of the three sets of
data into a training set and a test set where we have enough samples. If an
analysis of the training set can then correctly assign at least 75% of the
test samples to the correct blocks then we have a real
difference. (There are also some other rather more technical tests e.g. the
p-values on the distances between the group centroids.) But I think the
assignment test will be more readily seen as valid in literary circles.
My figure of 75% is based on comparative success rates of 200 word samples in
other recent studies on literary assignment.
Incidentally NT scholars are very keen on vocabulary statistics or rather
vocabulary counts - most such tests never offer any tests of significance
(Gaston is a rare and honourable exception). It would be interesting to see
how well such tests fared on say correctly assigning portions of Mark or
Matthew to the right gospel.
Mail from .com etc. may receive temporary diversion here
David L. Mealand * E-mail: D.Mealand@...
University of Edinburgh * Office tel.:(+44)-131-650-8917 or 8921
Scotland, U.K. EH1 2LX * Office fax.:(+44)-131-650-6579
In a message dated 23/3/98 11:07:12, you comment:
<< But M is simply the set of material that only Matthew has - not
necessarily a single source. >>
Is not Q simply the set of material that Matthew and Luke share, but not Mark,
and again not necessarily a single source?
- Synoptic-L was established to stimulate a broad discussion of the
relationships among the gospels, with emphasis on high-quality,
academic contributions to criticial analysis of the synoptic problem
that might benefit other scholars in the field. Accordingly, this
list is designed for contributions to collegial research, not as a
platform for interminable, irreconcilable debate between well-
entrenched, opposing positions. To this end, we have formulated
protocols for this list, whose observance we feel will foster this
atmosphere. A full copy of the list protocols is found on-line at:
There have been some questions lately about how the protocols would be
applied in practice. In general, we hope that everyone will use common
sense. We offer the following commentary as guidance:
Before replying to another post, it is always wise to ask yourself
whether your point will (a) contribute to resolving a problem in the
prior post and (b) benefit other subscribers. If not, the debate is
better pursued through private e-mail.
> 1. Please keep messages as concise and as relevant as possible.For example, if the message you wish to reply to contains many points,
>Remember that recipients may be less likely to pay attention to an
>unnecessarily lengthy message.
please pick out the two or three most important points and leave the
rest aside. A full, point-by-point rebuttal is likely to cause everyone
else to lose interest.
> 2. Please quote only the relevant parts of messages that you areA rule of thumb, found in other forums, is that if the amount of quoted
>responding to. Recipients may pay less attention to your message if they
>have to re-read irrelevant parts of previous, lengthy messages.
material exceeds your contribution, you probably are quoting too much of
Stephen Carlson, Thomas Longstaff, Mahlon Smith, and Mark Goodacre
- David Mealand wrote:
> Incidentally NT scholars areI think most studies based on vocabulary statistics would manage to
> very keen on vocabulary statistics or rather vocabulary counts -
> most such tests never offer any tests of significance (Gaston is a
> rare and honourable exception). It would be interesting to see how
> well such tests fared on say correctly assigning portions of Mark or
> Matthew to the right gospel.
assign passages to the right Gospel because usually the statistics
are comparative. To take Matthean vocabulary, for example, the lists
of Hawkins, Goulder, Luz, Gundry and Davies-Allison all make their
decisions about what is and isn't characteristic or distinctive by
means of comparison with the other Synoptics.
This is, in fact one of the potential problems with the lists. If
they are not properly used, there can be some circularity, as when
Davies and Allison, for example, say that their list of Matthean
vocabulary is "helpful for determining to what extent a given passage
has been moulded by the evangelist", in spite of their claim that
many of the words on their list are there because they "must . . . be
regarded as very often editorial" (*A Critical and Exegetical
Commentary on the Gospel According to Matthew*, Vol. 1 (ICC;
Edinburgh: T & T Clark, 1988), pp. 75-9).
It is for these kinds of reason that I would be interested to see
some successful stylometric study on the Synoptics. I was
struck by Lloyd Gaston's conclusion to his own *Horae Synopticae
Electronicae: Word Statistics of the Synoptic Gospels* (Montana: SBL,
"Perhaps the very preciseness of the tables will encourage us to
rely more on considerations of style and content. These latter
criteria, while superficially not as "objective", are really
much more important" (p. 12).
All the best
Dr Mark Goodacre M.S.Goodacre@...
Dept. of Theology, University of Birmingham