How similar are the Synoptics, and how do we represent it?

I have enjoyed the feedback, especially on twitter, from lots of people since I dusted down the blogging machine and reignited it earlier this week. I should blog more often! I had forgotten how much fun it is.

Anyway, this is the third post in the current series; cf. post one and post two.

Before continuing with critical reflections on Matthew Larsen's Gospels Before the Book, I'd like to pause to think a bit more about how we represent degrees of similarity between the Synoptic Gospels. This has been one of the most enjoyable take-aways from reflecting over the last couple of days. Larsen produced a proportional Venn diagram (p. 104) of the degree of similarity between Matthew and Mark in a bid to show just how similar these works are. He uses the pericope divisions in Aland's Synopsis. Here's my coloured version of his diagram, using my Synoptic colouring scheme: [2]:

Key:

Matthew's non-Marcan material (blue)
Mark's non-Matthean material (red)
Material shared by Matthew and Mark (purple)

The numbers:

Mark: 115 Aland pericopae
Matthew: 178 Aland pericopae
Overlapping: 107 Aland pericopae

93% of Mark is paralleled in Matthew.
60% of Matthew is paralleled in Mark.


The  diagram, though rough and ready, provides one metric for seeing how closely related Matthew is to Mark. The question then arises: what about Luke's relation to Mark? How close is it? Here's the proportional Venn diagram, again using my colouring scheme:

Key:

Luke's non-Marcan material (yellow)
Mark's non-Lucan material (red)
Material shared by Mark and Luke (orange)

The numbers:

Mark: 115 Aland pericopae
Luke: 185 Aland pericopae
Overlapping: 101 Aland pericopae

88% of Mark is paralleled in Luke.
55% of Luke is paralleled in Mark.

These figures are surprisingly similar to the figures for Matthew, surprising as Matthew is often regarded as so much closer to Mark, a kind of "second edition" of Mark. To some extent, this is a result of using the Aland pericopae rather than the traditional verse parallels, but I look forward to running some more precise numbers in due course.

The thing that got me thinking afresh about this whole question was Larsen's comment that "there are no two works from the ancient world more similar to each other" than Matthew and Mark. As I mentioned the other day, I am not sure if this is right. Matthew and Luke are much more similar overall, but we tend to miss this because of classic Two-Source Theory thinking that minimizes their macro-similarities, and projects their close non-Marcan agreements onto a non-extant source, with a view to maintaining  their independence from one another. Here's the proportional Venn diagram, again using my colouring scheme, and again using the Aland pericopae:

Key:

Matthew's non-Lucan material (blue)
Luke's non-Matthean material (yellow)
Material shared by Matthew and Luke (green)

The numbers:

Matthew: 178 Aland pericopae
Luke: 185 Aland pericopae
Overlapping: 137 Aland pericopae

74% of Matthew is paralleled in Luke.
77% of Luke is paralleled in Matthew.

I am grateful to Joe Weaks on The Macintosh Biblioblog (this is all wonderfully nostalgic!) for raising the question about the utility of traditional Venn diagrams like this. His suggestion is to work instead with rectangles:

Graphically Displaying Synoptic Data

This post shows how we can attempt to represent Matthew // Mark, Mark // Luke, and Matthew // Luke using coloured rectangles. I must admit that I really like Weaks's proposal, and not just because we use the same colour scheme. The only thing I'd say is that I think it would be harder to do the rectangle thing in black and white because it would be less clear that we are dealing with overlapping rectangles, whereas with circles, it is obvious even in black and white where one work ends and another begins.

Weaks continues by asking the next major question: can one represent the overlaps between all three Synoptics using the coloured rectangle approach? Weaks shows that it is possible here:

Graphically Displaying Three Synoptic Gospel Data

This diagram is also excellent. My only qualm is that on first sight, it can look like the double tradition (green) is also part of the triple tradition (brown), which is of course not the case, and one simply has to discipline one's mind not to see it that way, though of course in a more complex version one could at least attempt to depict with shades of colour pericopae that are pure triple (with only a handful of minor agreements), pure double (with no Marcan agreements), and everything in between. But life is probably too short for that.

Larsen's Challenge to Studying Synoptic Relations

This is a second post on Matthew Larsen, Gospels Before the Book (Oxford: Oxford University Press, 2018) [First post here] in which I'd like to comment on Larsen's challenge to the study of Synoptic relations in Chapter 6.

Larsen's concern is that scholars of the Synoptic Problem tend to see the gospels as separate, discrete books, each with a unique author, [1] rather than seeing the gospels as different instantiations of the same fluid textual tradition. He illustrates the point by noting the way that various Synoptic theories are diagrammed. He is talking about diagrams like these, and he gives his own versions of them (p. 102), and writes:

In all these graphic depictions, each constellation of textualized gospel tradition is represented as its own discrete unit, bounded by lines within a box or a circle or some other shape, with arrows indicating the direction of source relationship and redaction. All of this, however, as should be clear by now, serves to reinforce the third-century and subsequent gospel textuality and authorship discourse, reifying each gospel as an enclosed, separate text with its own unique author. How might we rethink the data? (p. 102).

It's actually not always the case that these diagrams are "bounded by lines within a box or a circle or some other shape"; my own preference has been to avoid the boundary lines, e.g. here in my book The Synoptic Problem: A Way through the Maze (p. 22):

In fact the irony of Larsen's concern about "clear black lines separating out discrete gospels from one another" (Larsen, p. 104) is that sometimes entities in these diagrams are placed in a box in order to show uncertainty about their existence or tangibility, as here in my diagram of the Two-Source Theory (Way Through the Maze, p. 20):

Nevertheless, Larsen's broader point is worth thinking about. Is it fair to say that diagrams like these tend to make us think too rigidly in terms of discrete, separate gospels, with different authors, and to ignore the overwhelming similarity between the Synoptics? Larsen's suggestion is to represent the "degree of overlap" between Matthew and Mark by means of a "Proportional Venn diagram of Overlap Between the Gospels of Mark and Matthew" (p. 104). So he counts the number of parallel "stories" in the index of Aland's Synopsis (see further yesterday's post):

Mark: 115 “stories”.
Matthew: 178 “stories".
Overlapping: 107 “stories”.

He then plugs these numbers into a Proportional Venn diagram, which I have adapted here in a coloured version (using my Synoptic colouring scheme) [2]:

Key:
Matthew's non-Marcan material (blue)
Mark's non-Matthean material (red)
Material shared by Matthew and Mark (purple)

It's a great idea to represent the data in this way, and I'm grateful to Larsen for thinking of it. There are precedents, e.g. the nice Wikipedia coloured diagram, but I don't recall having seen a proportional Venn diagram like this.

There is a point that needs making, though. The proportional Venn diagram is doing something completely different from the theory diagrams. The proportional Venn diagram is illustrating some of the data, while the other diagrams are illustrating theories of Gospel relationships. In other words, the Venn diagram is illustrating (an element in) the Synoptic Problem while the other diagrams are illustrating solutions to it.

It's a basic point, I know, but it is an important one. I have argued that one of the difficulties with the way that the Synoptic Problem is studied is that a theory is presented (usually Two-Source) and the data is then refracted through it. As Jason Staples says, it's "solution to plight" thinking.

In a sense, Larsen's preference for the proportional Venn diagram could be seen to forward this aim -- we might think of it as a way of encouraging people first to take the data seriously, and to get a sense of the problem before proceeding to solution. The difficulty, though, with the way that Larsen discusses the issue is that the Venn diagram is presented as an alternative to the theory diagrams, contrasting their bounded, discrete entities, with his overlapping materials. But both are necessary -- finding ways to represent the data as accurately and as clearly as possible as well as representing the theories as clearly as possible.

And with respect to those theory diagrams, everyone discussing the issue realizes that there are massive overlaps between Matthew and Mark. That's the beginning point of the discussion. If there were only differences, there would be no Synoptic Problem. Placing an arrow from Mark to Matthew (and to Luke) only expresses a model of textual relationships. One can still, like Burkitt and others, see Matthew as a "fresh edition of Mark" (see yesterday's post), and use a classic diagram to show that relationship, the new edition being subsequent to and incorporating the previous edition. Or, to use Larsen's language, if we "think of the textual tradition we call the Gospel according to Matthew as continuing the same unfinished textual tradition of “the gospel” more broadly understood", there is nothing to stop us illustrating that in a theory diagram, a diagram that would be attempting a solution to the problem, which is a quite different thing from a diagram that attempts to depict the data.

[1] Larsen regularly uses the term "human" author, though I am not sure why the adjective is necessary given that no one is arguing for animal or alien authors.

[2] Generated using the Venn Diagram Generator.

How Similar is Luke to Matthew? Reflections Stimulated by Larsen

I have recently been enjoying reading Matthew Larsen's Gospels Before the Book (Oxford: Oxford University Press, 2018) and have been reflecting on elements in its thesis. In one sense, I was predisposed to find the book appealing since I have myself flirted with the idea of Mark as an "unfinished" gospel (NT Pod 71), though I am less certain about some of Larsen's broader claims. In due course I hope to comment on his scepticism about our ability to do source- and redaction-criticism, but first, in this post, a couple of positive observations.

In Chapter 6 of Gospels Before the Book, Larsen reflects on how his thesis impacts on synoptic relations, which is a topic of interest to me. Larsen argues that we should not see Mark and Matthew as distinct "books", each with their own author. Each is an instantiation of a fluid textual tradition. To develop this point, he writes:

Viewed from within a different framework, we begin to see another picture. If one assumes the texts we now call the Gospels according to Matthew and Mark are not both part of the same fluid textual tradition, then to my knowledge there are no two works from the ancient world more similar to each other than the Gospel according to Mark and the Gospel according to Matthew, a fact often overlooked (101).

This perspective reminded me of the strong 20th century (mainly British) scholarly tradition of seeing Matthew as a kind of "second edition" of Mark. The tradition goes back, I think, to F. C. Burkitt in 1910, who described Matthew as "a fresh edition of Mark, revised, rearranged, and enriched with new material” (The Earliest Sources for the Life of Jesus (Boston & New York: Houghton Mifflin, 1910). Streeter echoed the judgement in his famous Four Gospels, and there's a fairly strong continuing tradition of seeing Matthew this way, e.g. by Graham Stanton, James D. G. Dunn, and more recently Francis Watson.

But a further thought on reading Larsen here occurred to me, and that thought was, "What about Luke?" Larsen is arguing that Mark is so similar to Matthew that there are no two works in the ancient world that are anything like as close as these two. But the point becomes stronger if one draws in Luke too. If Matthew and Mark are two of the most similar works from antiquity, surely Matthew and Luke are even more so.

The difficulty here is that decades of two-source thinking, with its insistence on Luke's independence from Matthew, have tended to immunize us against noticing the extent of the similarity between these two gospels. We allow Q to mediate their non-Marcan similarities, and then we stress their differences in attempting to underline their independence. But the similarities between Matthew and Luke are not limited to the two-hundred or so verses of double tradition. It is a question of their entire gospel projects.

I have been attempting press the point about the macro-similarities between the two works, in addition to the micro-similarities, for some years. If Matthew is effectively a kind of fresh edition of Mark, could Luke be seen still more as a fresh edition of Matthew? I don't know if I want to go that far, but I do think it worth pointing out once again just how similar these two works are. Unlike Mark, both begin with Infancy Narratives; both end with resurrection appearances & commission to "the eleven"; both feature a lot of additional identical sayings material, frequently with very close verbatim agreement.

It turns out that we can quantify the similarity between the two in a rough-and-ready way. Larsen does an interesting experiment in quantifying the degree of agreement between Matthew and Mark by using the index of Aland's Synopsis, and it's something we can extend to Matthew and Luke. Larsen's figures are as follows (pp. 103-4):

Mark: 115 “stories”.
Matthew: 178 “stories".
Overlapping: 107 “stories”. Thus, Larsen says:

93% of Mark is paralleled in Matthew.
60% of Matthew is paralleled in Mark.

I checked Larsen's numbers and they came out the same way for me. I then did a count on the Lucan parallels, and they come out like this:

Luke: 185 "stories"
Overlapping with Matthew: 137. Thus:

74% of Matthew is paralleled in Luke.
77% of Luke is paralleled in Matthew.

It is of course a clunky and imprecise way of doing things, and my own preference would be to do it on the basis of sentences or verses rather than Aland units, but it is interesting nevertheless to see just how quantifiably "similar" Matthew and Luke are to one another, at least according to this metric.