An Item in Burkett's Hypothesis

Delbert Burkett, in his book Rethinking the Gospel Sources: From Proto-Mark to Mark, argues for a hybrid synoptic hypothesis, which posits two similar versions of Mark that were blended by the canonical author into one text.  Thus he tries to use the best arguments for the Griesbach hypothesis, but avoids its weaknesses by keeping the Q material out of Mark's sources; he thinks the Q material was added to these earlier versions of Mark separately by Matthew and Luke.

I admire Burkett's creativity, and his work has done a lot to spur my own thinking on the synoptic problem.  However, Burkett's case rests on one important piece of evidence (among others I won't be discussing today): a statistical improbability he claims to find in the distribution of Markan dual expressions between GMatt and GLuke.

Burkett's discussion can be found on pp. 121-129 of his book, with an important in-depth detailed discussion in his Appendix B.  In that appendix, he explains how he takes Neiryck's category 10 and revises it to create a list of "two temporal or local expressions in the same context".  He then eliminates expressions not in both GMatt and GLuke to create a good test.  He divides the remaining expressions into categories summarized on p. 123.  He then compares the category of dual Markan expressions split between GMatt and GLuke, and those where GMatt and GLuke contain the same half.  He finds out of the total of 17 expressions in both categories, 14 are split, and 3 have one shared item.  Assuming random choices by Matthew and Luke, the odds of this happening are improbable (testing well beyond the .05p significance level).

I admire Burkett's work here, but I must disagree with his findings.  I note that if we can find a distribution whose likelihood is greater than .05p, Burkett's hypothesis fails significance.  This could be done in two ways, either by removing items from the first list (of split expressions), or by adding items to the second list (of one-shared-items).  We will show that enough items can be removed from the first list to remove significance for Burkett's hypothesis.

There are five expressions whose place on Burkett's list is dubious.  We will consider each in turn.

(Please note that when romanizing Greek, I prefer to use "X" for chi, and "C" for xi, in contrast to some conventions.)

Mark 1:28 

Burkett thinks this is a single two-fold example, where Matthew writes EIS OLHN THN SURIAN on the one hand, and Luke writes EIS PANTA TOPON TES PERIXWROU on the other.  But instead, isn't this actually an example where Matthew uses one half, and Luke both halves, of Mark's duplicate expression "everywhere/throughout the whole region of Galilee"?

Matthew does use EIS OLHN where Luke uses EIS PANTA TOPON.  But when Matthew adds TEN SURIAN, surely he is refering to Mark's "region of Galilee"?  Thus, Luke refers to the same place when he uses TES PERIXWROU.  What Burkett tries to do is pair SURIAN strictly with Mark's "Galilee", since both are proper names, but this won't work--Syria and Galilee are not the same, though Syria and the "region of Galilee" are.  So Matthew effectively uses just "throughout the whole region of Galilee", and Luke effectively uses both "everywhere" and "throughout the whole region [of Galilee]".

Furthermore, PANTAXOU is only found in the Alexandrian texts.  What if Luke had a version with the Alexandrian reading, and Matthew didn't?  In which case, Matthew wouldn't have read a duplicate expression at all, and it clearly wouldn't belong on Burkett's list. 

Perhaps Burkett is also suggesting a nestled duplicate in "the whole/region".  (He seems to be on his chart.)  However, this is hard to read in the grammar, and at any rate Matthew would then be the one to refer to both, by using both EIS OLHN and TEN SURIAN--again, the latter to refer to the "region of Galilee"--and Luke only TES PERIXWROU--to refer to "the region [of Galilee]".  This would likewise eliminate it from Burkett's list.

The above considerations show that Mk 1:28 does not obviously serve as an example of opposed Matthean/Lukan readings of a Markan pleonasm (of time/space or any other type).  It should be dropped from Burkett's analysis.

Mark 2:19bc

Matthew writes EF OSON and Luke writes EN W.  But couldn't Matthew be read as using both halves of Mark's "while...as long as", trying his best to blend together the meaning of Mark's verse?  (EN W...OSON XRONON).  Thus Mk 2:19, too, can be dropped from Burkett's list, as least as easily as it can be kept on it.

Mark 3:8cd

This one is fairly simple to dispense with: "beyond the Jordan" and "around Tyre and Sidon" are not duplicate expressions.  We drop it from the list without need for further discussion.

Mark 5:12

Mark's "into the pigs/into them" (EIS TOUS XOIROUS EIS AUTOUS) is not obviously a local expression, an expression of place.  Pigs are not a location in the same way Mark's other locations are; pigs are mobile creatures, who themselves have a location.  The non-obviousness of this example also lets us drop it from Burkett's list.  This example also shows how subjective the category of "local/temporal expressions" is.

Mark 10:27

This is perhaps the easiest case: "with God/with [the] God" (PARA QEW PARA TW QEW) is not a local or temporal expression at all.  Dropping this one from the list is a no-brainer.


What does this leave us with?  Twelve expressions, nine of which are opposed readings of Markan pleonasms, three of which are shared readings of one item.  At this pont one can calclulate the binomial odds of this distribution, and find that while still interesting, they now fail significance at the conventional .05p level.  They are still worth explaining (I would simply say that Luke is interested in including Markan material that Matthew did not include, and is less interested in making sure all of GMatt is included) but they do not disprove alternative hypotheses that assume Markan priority, without assuming Markan blending of separate sources.