Tuesday, 15 July 2014

The Problem with Mark-Q Overlap

I've recently read „The case against Q“ by Mark Goodacre, who makes a case for the Farrer theory as an alternative to the more popular two source hypothesis: He argues that instead of Matthew and Luke independently redacting Mark using the hypothetical source Q, Luke used Matthew as well as Mark, and no such document as Q existed.

I found his argument very convincing in parts, less so in others, but mostly, I was once again irritated by the notion of Mark-Q overlap, a concept whose apparently wide acceptance has baffled me for quite some time. In this post I will try to explain why, starting with an explanation of what Mark-Q overlap is.


Minor and major agreements


At the center of the issue is the realization that there is a significant number of triple tradition passages where Matthew and Luke agree against Mark. This is a problem for the original two source hypothesis, which assumes that Mark and Q are two documents with no literary connection, and Luke and Matthew independently used both to write their own gospels. In this scenario, there is no way to explain how, for example, Luke and Matthew could both have the same expanded version of an originally Markan scene. The additional material obviously cannot be from Mark, and it can hardly stem from Q, because the probability that Q included a phrase or phrases which just so happen to fit perfectly into an unrelated Markan pericope, Matthew and Luke both recognized this and inserted the material from Q in the same place in the same way is very low, expecially considering that would have had to happen in a number of places in the synoptic gospels.

These agreements therefore cannot be explained by the original two source hypothesis, which limits the hypothesis' ability to explain the evidence we have and thereby constitutes evidence against it. In order to solve this problem, defenders of Q have divided these pericopes where Matthew and Luke agree against Mark into two broad categories, the major agreements and the minor agreements.

The major agreements are in fact so major that there seems to be no reasonable way to explain them without some kind of literary relation between Luke's changes and Matthew's changes. The way the two source hypothesis usually explains material common to Luke and Matthew is by positing that it stems from their common source, Q. A simple thought process therefore lead to the invention of the category of "Mark-Q overlap":
Since Luke and Matthew share material in these passages, it must have been in Q. Since Mark has different, but obviously related versions of the same pericopes, different versions of the same pericope must have been in Mark and Q.

This complicates the relationship between Mark and Q quite a bit; where before, there were two documents which were basically unrelated except for the common theme of being-about-Jesus (and even that can be disputed in the case of Q), now there has to be some relationship between the two, be it direct or indirect, literary or oral. But well, if that is the theory that explains the evidence, then so be it.

Now to the category of minor agreements, which differs from the major agreements only in the size of the agreement: Where major agreements have entire phrases which are present in Matthew and Luke but absent in Mark, minor ones usually only have single words. It seems obvious that the same mechanism that is used to explain major changes both Luke and Matthew made to their Markan source material would also be suitable to explain smaller changes. There is no problem in principle with the theory that in pericopes which have minor agreements, Mark and Q simply had very similar, in fact almost identical versions of the same story, and the version in Q just differed in a few single words. Luke and Matthew then both used Q's version of the material, and voila, we have a minor agreement.

Yet this is not what Q theorists have done, otherwise, there would have been no need to create the two categories of minor and major agreements in the first place. Instead they basically admitted that they do not know the exact way the minor agreements came into existence and only offered a few possible explanations, like later scribes taking over Matthew's changes to Mark into manuscripts of Luke, or vice versa.

Generally, I would find this absolutely acceptable, and even admirable, for two reasons:
  1. Admitting that there is just no way to know, in cases where there really is just no way to know, either because one does not have any explanation, or because there are so many plausible explanations that it seems impossible and unnecessary to choose one, is just basic intellectual honesty. This should generally be encouraged, especially in New Testament studies, where so much of the evidence is suspect and/or contradictory, so many basic facts are disputed, yet almost noone admits the enormous amount of uncertainty, and almost everybody seems to just pick a number of facts they take for granted, either without justifying their choice at all or with only a short and obviously deficient justification, and builds their theory on that basis.
  2. The amount of trust that is placed in the presence of single words or phrases in the original gospel texts has always baffled me. Everyone knows the incredible amount of changes that we know have been made to the original manuscripts. Logic dictates that one always keeps in mind that any given phrase in the NT may not originally have been there, and could simply be the result of a later change which we do not have the means to detect. In my opinion, this makes any theory that is based solely on a few single words and phrases suspect from the beginning.

Yet in this case, one cannot help but notice that Q theorists offered a perfectly fine theory for explaining all agreements of Luke and Matthew against Mark in general, yet chose to only cite it for some of the existing agreements, and state that another group (which simply has the property of being "minor") must have come to be via a different mechanism.

Why did they do this?

An artificial distinction


The only justification for this that I know of, and I would be happy to be corrected in this case, is that, whereas major agreements occur only in pericopes concerning Jesus' ministry, there are minor agreements in the passion narrative as well. And existing scholarship on Q had already established that Q did not contain a passion narrative, which rules our the explanation of Mark-Q overlap at least for those passages.

This reasoning is obviously problematic if not downright fallacious, for several reasons.

  • First of all, it holds an entire group of agreements hostage because some of them do not fit the established theory for explaining them. Why not just state that Mark-Q overlap explains all agreements in general, and then state that those which cannot be explained by it must result from textual corruptions, be accidental changes made by both evangelists, or some other unknown reason? No, instead, all agreements considered to be "minor" are taken out of the Mark-Q overlap category, because some of them are problematic for the theory. This has three interesting consequences:
    • It obscures the true amount and significance of the agreements by introducing two arbitrary and unnecessary additional categories.
    • It hides the reason why there is a distinction between major and minor agreements in the first place.
    • It obscures the fact that any non-passion minor agreement may very well result from Mark-Q overlap as well.
  • The argument is incredibly circular. It basically states that, based on past reconstructions of Q, we have concluded that Q did not contain a passion narrative. Now we have evidence that it might actually have done so, which would change the reconstruction of Q on which the initial argument was based. Instead of admitting this, one now takes the conclusion that Q has no passion narrative as a premise, tries to explain away any evidence pointing towards the opposite directon, and approaches future reconstructions on this basis, thereby of course reinforcing the original conclusion.
  • An even bigger, but related, point is this: The argument for the validity of the two source hypothesis, and therefore the existence of Q, is mixed up with the question about the nature of Q. As long as at least the major agreements cannot be properly explained, the two source hypothesis simply does not account all too well for the evidence we have, which means that the argument for the very existence of Q is weak. However, the existence of Q must be established first before one can reason about its content and nature in any detail. Letting a result of the latter process interfere with the former, which is a prerequisite for the latter, is just bad logic.

The shape and content of Q


So the handling of the minor/major issue is more than problematic. However, I mentioned before that I actually find the very idea that there was an overlap between Mark and Q even more spurious, because it means that we cannot possibly establish any certainty at all about the content of Q.

Let me elaborate:

Once we admit that Mark and Q sometimes contained slightly different versions of the same story, we must admit that it is quite plausible that they sometimes contained exactly the same version. One could of course dispute this by postulating that the relationship between Mark and Q is merely an oral one, which would indicate that it is unlikely that a pericope found its way from one document to the other without any changes. To this I would respond that we know that a number of pericopes must have been in both documents in very similar shape, simply because the versions in Matthew/Luke in overlap material have a lot of verbatim agreements with the Markan versions. Any mechanism which leads to the exact transmission of entire phrases from one document to another is also able to transfer entire pericopes.

If, however, Mark and Q share some pericopes in (virtually) identical form, this means that Matthew and Luke both only had this one version to use in their own gospels. We would therefore see pericopes like this in identical form in all three synoptic gospels and call them triple tradition.
This, however, means that any given triple tradition material which does not feature any agreements of Luke and Matthew against Mark may well have been in Q as well.

Note that I am not arguing that it is possible that pure triple tradition pericopes also appeared in Q, but that it is probable. Strictly seen, it is possible that Q contained the first half of the Magna Carta, both both Luke and Matthew simply (and probably correctly) decided that this material did not belong into a gospel about Jesus Christ and did not use it. However, this is obviously highly improbable. Triple tradition passages being in Q as well as Mark, however, is not improbable at all once we postulate a relationship between Mark and Q; in fact, we just stated that in the case of the major agreements, this is exactly the case. For the reasons I just spelled out, we just have no way to know how much pure triple tradition material was in Q and how much wasn't; all documents we have today would look exactly the same either way. Indeed, it would seem rather unlikely that all passages which were present in both Mark and Q left exactly those traces in Luke and Matthew that we would need to classify them as Mark-Q overlap.

But it gets worse. We have established that some passages were in Mark and Q in different versions, but here, too, it is impossible to say how many. Where Q had a different version of a story than Mark, it is completely plausible that Matthew or Luke sometimes used Mark's version and sometimes used Q's version. In fact, based on what we know about their redactional habits when adapting Mark, this picking and choosing is exactly what we would expect them to do. But if either Luke or Matthew chooses to pick Q's version while the other stays with Mark, which in this scenario will almost certainly occur on many occasions, then Q's version will appear to us like a Lukan or Matthean addition. This means that for any given Lukan or Matthean addition to Mark, we must admit that it is plausible (i.e. definitely possible and not improbable) that it was in Q as well.

In addition, even in the absence of any Markan context, any pericope exclusive to either Matthew or Luke may have been in Q as well; the other evangelist may simply not have used it for whatever reason.

I am obviously generalising here; for any single pericope, there may of course be reasons based on its content for why it seems more or less likely that they originated with Q, Mark, Luke or Matthew. For the broad outline, however, my argument stands. Moreover, given the fact that these reasons are usually of the form „Matthew/Luke/... would/would not have written/used/changed X because of his tendency to do Y“, and these statements are, again, based on what we think Matthew/Luke had at their disposal as sources and what they did with them, many if not most such reasons should be regarded as speculation.

What this means for Q


On the assumption that Q existed and overlapped, to some degree, with Mark, this brings us to the following point:
  • We know for a number of pericopes that they must have been in both Q and Mark because of the existence of major agreements.
  • We have evidence that some others have been in both Mark and Q because of the existence of minor agreements.
  • We know that pure double tradition material must have been in Q.
  • We know that for absolutely anything written in Matthew and Luke, including pure triple tradition material, Matthean and Lukan additions to Mark, as well as pericopes exclusive to Matthew and Luke, they may plausibly have been in Q as well.
Current reconstructions of Q embrace the first and third point, sometimes acknowledge the second point to some degree, and completely ignore the fourth one. Yet there is no reason to do so. If Mark and Q shared some material, and we simply do now know how they came to do so, then it is absolutely plausible that they shared some more material as well. Based on the amount of material one wants to have Mark and Q share, one can now reconstruct Q in any way one wants, either sharing as few material as possible (which is what most reconstructions do) or sharing a lot more.

In the latter case, Q could have included all triple tradition material (including, of course the passion narrative) as well as, for example, all material exclusive to Matthew (and, of course, the double tradition, which has to be included in any case). If you find this idea unlikely and speculative, observe the following two points:

Logically and probability-wise, the currently dominant reconstructions where Mark and Q share as few material as possible are in no way preferrable to the case I sketched above (or anything between those two extremes).

The version I just sketched, where Q includes triple tradition, double tradition, and Matthean additions, would mean that Q is identical to the Gospel of Matthew. This would mean that Luke simply used Matthew in addition to Mark, a view that is the most cited alternative to the two source hypothesis, and defended by reputable scholars like Goodacre and Goulder.

The latter point shows that the Farrer theory is actually just a special case of (and not really an alternative to) the two source hypothesis, where the Gospel of Matthew is Q.

Even if you find all this unpersuasive (please let me know where you think I went off the track), you will have to admit that the content and character of Q is a lot less certain than most scholars think or pretend it is, and that statements like "Q did not containx X or Y" are pure speculation.


The two source hypothesis as a theory


In science, a hypothesis has to be falsifiable to qualify as a theory, which means that it must make predictions which, if they turn out not to be true, can falsify the theory. This is a standard that can obviously not be met by the two source hypothesis, but then that is to be expected, if only because there is no way to conduct any meaningful experiments on creating-the-New-Testament, and new documents relevant to the question are not found all too often. For the same reasons, most other hypotheses about the creation of the New Testament are not falsifiable either.

Note, however, that the original two source hypothesis which regarded Mark and Q as unrelated, was falsifiable. It could be falsified by finding agreements of Luke and Matthew against Mark in Markan material, and this is exactly what happened; these were simply not expected and not explained by the theory.

Once one allows for Mark-Q overlap, however, there is nothing new one could find out about the present texts that would actually contradict the two souce hypothesis. In fact, I cannot even think of any document one could find that would contradict it.

But again, in New Testament scholarship, falsifiability is rare and not essential. Explanatory power, however, is. Let's see what we have in this department: What we attempt to explain is the shared content of three documents whose original shape we kind of know. The basic assumption of Markan priority is not really up for dispute, so what we really want to explain is the content of Matthew and Luke.

Our explanation is a postulated document of which there is no record, which must have contained some passages and likely contained some others, but we have no way to know which or how many, as I argued above. Essentially, we can only speculate about the nature of the document. Additionally, this document has some kind of relationship with the gospel of Mark, but we do not know of which kind. With this document, we can account for the bulk of the material in both Matthew and Luke, though we have no way to know if M/L/triple tradition material comes from the document or not, and we may not be able or willing to explain at least some minor agreements.

To me, this does not look convincing at all. The theory replaces one big question mark with at least two other big question mark, and is not falsifiable in any way. It also cannot help us too much in understanding other texts of the time, simply because, as argued before, the contents of the document at the heart of the theory cannot be determined with any certainty. In addition, the postulation of Mark-Q overlap has always felt artificial and ad-hoc to me, like something that nobody would have even thought of if it weren't necessary to safe the theory.

I mentioned the Farrer theory in the beginning as an alternative to, or at least a more concrete version of the two source hypothesis. I do believe this theory has its own problems, but I would probably prefer it over a vague theory with no predictive power at all whose very core feels artificial to me. I am, however, convinced that New Testament scholarship would do well to acknowledge the amount of uncertainty there is about the content and nature of Q, and stop taking assumptions for granted which are speculative at best.