Loading ...
Sorry, an error occurred while loading the content.

Re: Synoptic Methodology

Expand Messages
  • Stephen C. Carlson
    ... I like the basic contours of your approach better this time around, and we may be closer to some kind of agreement. If we are to investigate the Synoptic
    Message 1 of 3 , Aug 14 1:26 AM
    • 0 Attachment
      At 05:31 AM 8/13/98 -0400, E. Bruce Brooks wrote:
      >There are six possible sequences of ABC, the other five being ACB, BAC,
      >BCA, CAB, CBA. I shall consider the ABC sequence for convenience. Between
      >its first two members, the relation can be dependent (A > B) or independent
      >(A, B). The options for C are wider: it can be dependent on none, or one,
      >or the other, or both, of the previous two.

      I like the basic contours of your approach better this time around, and we may
      be closer to some kind of agreement. If we are to investigate the Synoptic
      Problem *ab initio*, then the following approach would seem to be useful:

      (1) Define various equivalence classes of candidate solutions
      (2) Devise tests to eliminate preliminary solutions
      (3) Repeat (2) until done.

      1. Candidate Solutions

      It is clear that even from your example of 48 relationships, the number of
      candidate solutions increases so rapidly that some amount of pruning ought
      to be done to maintain tractability. Therefore, I propose that certain
      related candidate solutions ought to be grouped into the same equivalence
      class. I offer the following:

      (A) If two documents or two set of documents are independent, then
      their relative chronological order is irrelevant for purposes of source
      criticism. For example, the order K;L;M is equivalent to L;K;M (K=Mark,
      L=Luke, M=Matt). This would coalesce the 48 of your candidates down to
      25 equivalent classes of candidates.

      (B) Hypothetical documents should only be proposed for indirect
      relationships between at least two documents. For example, the order
      M>x>L (see Chapman) is equivalent to M>L. In other words, a hypothetical
      document that is a source to a single document is considered to be
      indistinguishable from an editorial policy. This eliminates intermediate
      texts, as you've suggested.

      At this point, we'll slowly consider some candidates.

      Zeroeth-Order Relationships
      Between any documents, they are independent. Therefore, all are
      independent, leaving only one candidate: [K;L;M] (Oral Hypotheses).
      This possibility, however, does not account for the demonstrable existence
      of a literary relationship between and among all three Synoptics.

      First-Order Relationships
      Between any two documents, they may be independent, or one is directly
      dependent on the other. Pairwise, there are 3 possibilities each, for
      3*3*3 - 2 (cycles) = 25 candidates. Of the 25 candidates, 17 require at
      least two to be independent, leaving only 6 that can account for the fact
      of a literary relationship between any two of the Synoptics. First-order
      candidates, however, do not include the dominant Two Source Hypothesis
      (Weisse I), suggesting that more work is needed.

      Second-Order Relationships:
      Second-order relationships allow for a hypothetical document as a
      source for any two or more extant documents. It is tempting to expand the
      pairwise possibilities to four choices: independence, left and right direct
      dependence, and indirect dependence, for a total of (4*4*4 - 2) = 62 candidates.
      However, this does not include the Simons/Holtzmann II/Gundry Three Source
      Hypothesis (Mark-Q with subsidiary dependence of Matthew for Luke) and an
      Ur-Gospel (Lessing) or an Ur-Markus (Weisse II, Holtzmann I) with indirect
      dependence of all three. The first possibility allows for there to be both
      a direct and indirect relationship for any two goespels, for 2*2*2*(3*3*3 - 2)
      = 8*25 = 200 possibilities and the indirect, triple dependence (yes or no):
      200*2 = 400.
      So, when we between to consider hypothetical documents, the number quickly
      mushrooms to 400. Even here not all proposed solutions are enumerated. For
      example, Abbott's deutero-Mark Hypothesis makes an extant Synoptic the source
      of a hypothetical document. As another example, Brian Wilson's double indirect
      triple dependence model is not listed, nor are literary relationships between
      two hypothetical documents are considered.

      2. Devising tests to filter out candidate solutions

      You have argued for an initial "two-body" approach as a starting point, while
      I prefer to tackle the Griesbach Hypothesis straight on, examining all three
      gospels synoptically. Here is an illustration why I think that the pairwise,
      two-body approach is fraught with difficulty. Kümmel, Neirynck and Fitzmyer
      have argued that, considering only Mark and Luke, an examination of the order
      of pericopes, i.e. the five or so transpositions of pericopes between Mark
      and Luke, leads one to the conclusion that Mark is prior to Luke. However,
      for each of the five examples, the order of Matthew and Mark is identical and
      therefore consistent with the Griesbach Hypothesis! [I believe Farmer noticed
      this, but I can't find the cite]. Therefore, this two-body test that apparently
      concluded Luke's dependence on Mark was in fact incapable of eliminating the
      Greisbach Hypothesis, even though the Griesbach holds to Mark's dependence on
      Luke (and Matthew). Maybe you can do better, but it seems difficult to avoid
      this trap without considering the third synoptic in the first place.

      Stephen Carlson
      --
      Stephen C. Carlson mailto:scarlson@...
      Synoptic Problem Home Page http://www.mindspring.com/~scarlson/synopt/
      "Poetry speaks of aspirations, and songs chant the words." Shujing 2.35
    • E. Bruce Brooks
      Topic: Synoptic Methodology From: Bruce In Response To: Stephen Carlson I appreciate Stephen s prompt and rational response to my expanded suggestion for a
      Message 2 of 3 , Aug 14 6:45 AM
      • 0 Attachment
        Topic: Synoptic Methodology
        From: Bruce
        In Response To: Stephen Carlson

        I appreciate Stephen's prompt and rational response to my expanded
        suggestion for a Synoptic strategy. Herewith a few answers to his
        counterprosal:

        Stephen remarks, of my 48 schematic possibilities for the 3 Synoptics:
        ----------
        1. Candidate Solutions

        It is clear that even from your example of 48 relationships, the number of
        candidate solutions increases so rapidly that some amount of pruning ought
        to be done to maintain tractability.
        -----------
        BRUCE: The number 48 is indeed too many to attack one at a time. But I
        tried to show (and I still think it is correct) that whether or not we
        eliminate some of them by preliminary test or mutual agreement, it will
        still be both necessary and sufficient to apply the 2-body test three times
        (once in each of the possible 2-body pairings) to reduce the 48 to 1.
        Unless of course the results are indeterminate or multivalent, in which
        case we have more work cut out for us.

        Stephen again:
        -----------
        Therefore, I propose that certain
        related candidate solutions ought to be grouped into the same equivalence
        class. I offer the following:

        (A) If two documents or two set of documents are independent, then
        their relative chronological order is irrelevant for purposes of source
        criticism. For example, the order K;L;M is equivalent to L;K;M (K=Mark,
        L=Luke, M=Matt). This would coalesce the 48 of your candidates down to
        25 equivalent classes of candidates.
        ------------
        BRUCE: We can symbolize order-irrelevant cases by a slash. If Luke (OK, I
        can use L for it) uses both Mark (K) and Matthew (M), but Mark and Matthew
        are unrelated, then indeed, for purposes of the problem (if not for the
        history of early Christianity), K,M>>L is indistinguishable from M,K>>L,
        and those options can be written as a single form K/M>>L. (Notice that the
        order question is not equivalent, at least for formula purposes, if only
        one of the first two texts is used by the third; thus, K,M > L is *not*
        equivalent to M,K > L). These functionally equivalent forms would tend to
        come out at the same stage in the successive application of the two-body
        test. It's harmless, for present purposes, to lump them. But it's handy,
        for other purposes, to distinguish them. Since (as I see it) the two
        equivalent solutions come out in the experimental wash with the same total
        effort (three two-body passes), it seems to me better to keep all 48 as
        distinct. They make a nice pattern.

        Stephen on the complication of conjectural sources:
        -------------
        (B) Hypothetical documents should only be proposed for indirect
        relationships between at least two documents. For example, the order
        M>x>L (see Chapman) is equivalent to M>L. In other words, a hypothetical
        document that is a source to a single document is considered to be
        indistinguishable from an editorial policy. This eliminates intermediate
        texts, as you've suggested.
        -------------
        BRUCE: I don't like the term "indirect" relationships, and prefer
        "intermediating" relationships. I agree that it is only under exquisite
        conditions that the case M>x>L can be even tentatively distinguished (where
        x is not extant) from M>L. To specify one such possibility, however, just
        for the sake of abstract theory: if the contribution of M to x is very
        strongly marked in x, and the tendency of L is to incorporate sources more
        or less passively, and if any other sources used by L are well
        differentiated stylistically from the immediate (intermediate) source x,
        then a skilled investigator of L might be able to detect (by its x
        stylistic common boundary) the presence of x, and also the presence of bits
        of M within that x boundary. From this the composite nature of x could be
        readily inferred, and a tentative M perhaps even reconstructed.

        I note with Stephen that, whatever the symbology, and even with collapsing
        of equivalent solutions, adding non-Synoptic texts to the definitions very
        rapidly expands the number of candidate solutions. I don't pause to quibble
        over the point that with the 3rd member of 3 there are 4, not 3, possible
        relationships to the preceding 2 texts. It is bad enough even if we only
        allow 3 relationships. He continues:
        -------------
        2. Devising tests to filter out candidate solutions

        You have argued for an initial "two-body" approach as a starting point,
        while
        I prefer to tackle the Griesbach Hypothesis straight on, examining all
        three
        gospels synoptically.
        -------------
        BRUCE: I think the difference here is at bottom temperamental: I prefer to
        investigate problems and not attack people. I note, in defense of a
        preference that might otherwise seem merely timid and therefore
        contemptible, that the result of a problem investigation is more
        psychologically viable for the adherents of a solution which that
        investigation as it were incidentally invalidates, than is the result of a
        direct attack on the named position which they uphold. Generals are
        criticized for knowing how to win but not what to do on the day after the
        victory. A problem-oriented investigation would avoid that censure in the
        Synoptic case. However, Stephen also has doubts about the effectiveness of
        my more impersonal investigation model:
        ------------
        STEPHEN: Here is an illustration why I think that the pairwise,
        two-body approach is fraught with difficulty. Kümmel, Neirynck and
        Fitzmyer
        have argued that, considering only Mark and Luke, an examination of the
        order
        of pericopes, i.e. the five or so transpositions of pericopes between Mark
        and Luke, leads one to the conclusion that Mark is prior to Luke. However,
        for each of the five examples, the order of Matthew and Mark is identical
        and
        therefore consistent with the Griesbach Hypothesis! [I believe Farmer
        noticed
        this, but I can't find the cite]. Therefore, this two-body test that
        apparently
        concluded Luke's dependence on Mark was in fact incapable of eliminating
        the
        Greisbach Hypothesis, even though the Griesbach holds to Mark's dependence
        on
        Luke (and Matthew). Maybe you can do better, but it seems difficult to
        avoid
        this trap without considering the third synoptic in the first place.
        -------------
        BRUCE: I can't speak with confidence without doing the work, but (1) I was
        envisioning a study of literary indebtedness in parallel material, not a
        study of the order of the material. (2) Some previous investigators report
        that there is only one variation in order of *narrative* segments (as
        against the more freely recombinable *wisdom* segments) between Mark and
        Luke. Evidently it makes a difference where you look for fixity of order.
        The word "pericope" blurs together all delimitable segments, and thus
        obscures this possibly important difference, which in turn affects
        experiment design. (3) Also evidently, questions of the arrangement of
        material bring in, and turn on, assumptions of the authorial intent,
        including the compositional predilections, of the respective Evangelists.
        If hard proof rather than plausible argument is what is called for, I think
        text priority is stronger than order retention (or rearrangement). As far
        as I have observed, it is the "authorial mind" arguments that are most
        frequently denounced as "reversible," and these seem therefore likely to be
        less effective in a contested situation. (4) Let me say again that I do not
        propose to stop with one two-body test, hence the word "only" in Stephen's
        quote is irrelevant to, and not a refutation of, my proposal. (5)
        Operationally, my point is that if the GH is wrong, and it is not
        eliminated on the first pass, it will infallibly be eliminated on the
        second or third pass. I don't have any special animus for GH (much as I
        deplore a certain stylistic tendency among some of its most prominent
        adherents), and I don't care whether it comes out in the first round, or
        second, or for that matter not at all. I just want to solve the problem.

        On the theoretical preference for initial three-body comparisons, take
        Stoldt p101 (not here reproduced, and in any case the dratted archive would
        only collapse the columns). He there notes that Wernle asserts that "quite
        frequently Luke's phrasing deviates from Mark's" and illustrates this with
        23 [actually 22] Mark/Luke pairs, where indeed the Lukan word differs from
        the Markan word. Assuming it previously established (at the segment level
        and not at the level of these particular words) that Luke and Mark are here
        in some sort of literary relation, Wernle must be conceded to have proved
        his case: Luke (if he is second) does not copy identically, but varies
        words within what (I presume) is otherwise known to be indebted material.
        This is not a test of dependence, nor without examination of the meaning
        and not the mere identity or difference of the wordforms is it a test of
        directionality of dependence (and if Wernle thinks it is, he is indeed
        totally wrong), but in that framework it would be a valid examination of
        Luke's way of handling borrowed material.

        Stoldt supplies the missing Matthean parallels (not available in two cases,
        #14 and #22, so we are down to 20 three-way examples). He points out that
        in most of the cases Matthew agrees with Mark. That is correct. Here is one
        example (in Stoldt's ordering of columns):

        1. Mk isk*uontes / Lk 5:31 ugiainontes / Mt 9:12 isk*uontes

        The other cases where this Mk = Mt pattern obtains, including cases where
        Mt differs slightly but is still closer to Mk than Lk is to either, are his
        #3-4, 6-7, 9-10, 12, 15-16, and 18-20. With the #1 case, these make
        altogether thirteen (13), or 65% of the total corpus of 20.

        There are also some other relational patterns. I should say at this point
        that I have sometimes been unable to verify Stoldt's readings in the Aland
        Synopsis, but for the moment conceding Stoldt his choice of text, whatever
        it is, then Luke and Matthew are closer (Lk = Mt) in:

        2. Mk epiraptei / Lk 5:36 epiballei / Mt 9:16 epiballei

        and in #5, 11, 13, and 21, a total of four (5) or 25%. The third
        theoretically possible two-way closeness, Mk = Lk, may (depending perhaps
        on segmentation; I don't know this language) be seen in:

        17. Mk promerimnan / Lk 21:14 promeletan / Mt 24:19 merimne#sete (total
        one, or 5%)

        and there is a three-way divergence in:

        8. Mk beble#tai / Lk 17:2 epiraptei / Mt katapontist*e# (total one, or 5%).

        There is the triplified Wernle material. Since Wernle's list was chosen to
        show differences beween Mark and Luke, it would be invalid to use it as a
        neutral test of three-way affinities. Despite this skew, there are some
        possible inferences. (1) In this sample, Mark and Matthew are often close,
        and frequently identical, this holding in two-thirds of the cases.
        Matthew's use of Mark (if such it be) thus tends to be respectful of Mark's
        wording. (2) Luke (assuming for the moment that he is third) is less close
        to either than those two are to each other. In 25% of the sample he is
        closer to Matthew; and in 5% closer to Mark. He thus (as far as this
        handful goes) is more respectful of Matthew than of Mark, but (3) It is
        still more likely than not that Luke *knew* Mark, in addition to following
        Matthew for places where Matthew and Mark are identical. The hypothesis of
        Matthew as an intermediary for Mark will not hold. It also happens that (4)
        Luke varies from all three, and in general his pattern with these 20
        phrases is more for rephrasing than for identical repetition. Luke thus
        comes across as a stylistic independent, preferring the wording of Matthew
        to of Mark in most but not all cases where both are available, whereas at
        an earlier stage Matthew was in a higher percentage of cases respectful of
        Mark's exact wording.

        On a Mark-last scenario, these conclusions would instead look like this:
        (1) Of the two earlier sources, Luke is not particularly close to Matthew
        in parallel passages, thus generating a number of variants for any third
        version to consider. (2) As that third source, Mark is very close to
        Matthew in a majority of cases, so that Matthew is his preferred source,
        but (3) he is nearer to Luke in a detectible minority of cases, and so must
        have known Luke as well as Matthew, (4) he sometimes diverges from both.
        This last shows that Mark too, if considered to be last, must be allowed to
        have exercised stylistic originality.

        Can we say which of these scenarios is the more likely? Not decisively, as
        far as I can see. The data are not well chosen for the purpose, and are
        very limited in any case. As with the Neirynck comparisons faulted by
        Farmer, I do not think these were meant as an investigation, but rather as
        a demonstration on territory presumed established. I could make a guess,
        based on these results, but not offer a serious conclusion. Can anyone get
        more out of it than this?

        STOLDT is bolder in his inferences. He says, "This *synoptic* compilation
        clearly shows that it is out of the question that Luke changed Mark's
        wording. The respective parallels in Matthew, insofar as they exist, show
        that in the majority of cases Mark corresponds in mode of expression to
        Matthew; in the remaining cases Matthew and Luke agree in their mode of
        expression against Mark."

        BRUCE: "Corresponds in mode of expression" is a symmetrical expression; it
        is a valid statement whichever of the two texts is using the other.
        "Changed Mark's wording" is an asymmetrical expression; it assumes that
        Luke comes second in the Mark/Luke relationship. The sentence "Mark
        corresponds in mode of expression to Matthew" neither confirms nor refutes
        the sentence "Luke used and varied from Mark." The latter sentence is thus
        not logically "out of the question." It is the correct statement if it is
        separately known that Luke came after Mark. The directionality between Luke
        and Mark is not established by this particular data set, so the question of
        whether the asymmetrical statement is valid is open. Not closed. Not "out
        of the question," but precisely in question. It is also not precisely true
        that the only results in the table are Mt = Mk and (minority) Mt = Lk.
        There is one case, despite a selection designed to prove the opposite,
        where Mk = Lk. There is also one case (as Stoldt points out in small type
        in the table himself, at #8: "Here Matthew diverges from both") where all
        three are different, and so whichever text is assumed to come last must
        also be assumed to vary at least sometimes from the two earlier ones. Thus,
        the requirement of authorial latitude is established (conclusion #4 in both
        versions of the inferences, above).

        STOLDT: Here we get exactly the same image as that left by the pericopes
        which, according to Wernle, Luke supposedly passed over of omitted from the
        Gospel of Mark. *To the extent that Mark diverges from Luke in these
        examples, he converges with Matthew.* And this confirms again the fact that
        a solution to the synoptic problem is possible only on a synoptic basis.

        BRUCE: Again, the statement that Luke converges with either Mark or Matthew
        in all these cases is not strictly true. The entire situation, as noted
        above, does not lead to a decision between Markan tertiority (GH) and
        Markan priority (FH). Still less does it lead to a solution of the synoptic
        problem. The final sentence is thus radically unjustified by the exposition
        that precedes it. One could better say, of this particular data set (and I
        venture to think that I could assemble a more suggestive one), that it is
        indeterminate with respect to the final synoptic problem (though, if
        uncontradicted by wider examination of similar material, it does contribute
        some limiting conclusions to any priority solution, once that is reached
        separately and on other grounds). If one wants a dramatic version of this
        sentence, and accepts Stoldt's three-way version of Wernle's two-way list
        as "a synoptic basis,"one might say that this little investigation shows
        that "a solution to the synoptic problem is impossible on a synoptic
        basis."

        ENVOI: This conversation is proceeding on the assumption that only one
        approach is going to be undertaken. But several are already underway.
        Stephen's four-color Gospel parallels look to me like the groundwork for a
        three-body approach in small segments (not whole Gospels, but also not
        single words). This is an alternative to my proposal and to his
        counterproposal, not to mention the absurdly constricted Wernle/Stoldt
        list. So far Stephen's only published summary from that exercise is that
        Mark and Matthew are closer than is Luke with either or both. That is
        visually apparent, almost instantaneously, from the color text, in cases
        where the three texts run in parallel (we also got it out of Wernle/Stoldt,
        but on much narrower data and in a way that certainly requires this broader
        confirmation). On the other hand, where only Matthew and Luke are in
        parallel, a similar mutual intercoloration sometimes appears between those
        two. A sufficient number of such cases might tend to establish as a
        philological fact the alternate reliance of a third Gospel on the preceding
        two. Is it fair to ask Stephen how he sees the experiment design aspect of
        the color project?

        Bruce

        E Bruce Brooks / University of Massachusetts
      Your message has been successfully submitted and would be delivered to recipients shortly.