Monday, April 22, 2019

Robinson and Bordalejo on the CBGM and 1 Peter 4.16

49
A recently posted article on Academia by Barbara Bordalejo and Peter Robinson spends some time on the interesting change in the NA28 at 1 Peter 4.16. There, the ECM and NA28 read ἐν τῷ μέρει τούτῳ instead of ἐν τῷ ὀνόματι τούτῳ. The change is noteworthy because μέρει is somewhat awkward but much more so because it’s only attested in the 9th century and later. All the early evidence—Greek and versional—is on the side of ὀνόματι. Hence the many editions and commentators that prefer ὀνόματι.

Bordalejo and Robinson are interested in this change because they see it as a case where
the effect of the CBGM is to act as a kind of sleight of hand, with the textual flow diagram suggesting that all the older manuscripts and versions amount to a single line of descent, and so (indeed) having no more stemmatic weight than the single line of descent represented by the Byzantine text (p. 22).
The problem they have is in the textual flow diagram which shows 01, 1739, and P72 deriving ὀνόματι  from 03.


As they explain,
the change appears first in Vaticanus (03) and thence descends to Sinaiticus Alexandrinus (01 02) [sic; they seem to think the “/2” after 01 means 02 but it doesn’t]
 P72 and 1739. This is in accord with the way in which the CBGM shows textual flow working. Because Sinaiticus has more variants from the Ausgangstext (350) than has Vaticanus (280), the textual flow shows the text of Sinaiticus as descended from Vaticanus. But this is simply not true. Sinaiticus is comprehensively not a copy of Vaticanus or descended from it. It is here that the exclusion of sub-ancestors from the textual flow diagrams becomes a problem. [Andrew C.] Edmondson’s representation of the textual flow appears to show all of 01 02 [sic] P72 and 1739 descending from 03. But this is not the case. Not one of these four [sic] manuscripts is a descendant of 03.
The problem here is that they read the textual flow diagram exactly the way it should not be read, namely, as if it were a traditional stemma. A textual flow diagram is not and should not be read as if it is telling us what the actual, historical ancestors were. Instead, it shows us the most closely related witness that has more ancestral readings (i.e., the closest potential ancestor in CBGM terms).

Because of this, they are exactly right to note that “It is possible that there might have been an exemplar below the archetype from which all of the uncials, P72 1739, and all the versions, might have descended.” This is indeed a possibility, especially given the loss of NT manuscripts. But what the coherence shows is that no matter which reading we set as initial (ὀνόματι or μέρει), the coherence suggests that the text probably did changed from μέρει to ὀνόματι in the course of transmission. As Tommy and I have written
This suggests that, if reading b was the original source of reading a, reading a must have nevertheless developed from b a number of times as the text was subsequently copied. The simpler explanation, in light of the transcriptional evidence already discussed, is that it also developed in this way in the first place. (New Approach, p. 73)
To be sure, anyone is welcome to set aside the evidence of coherence at this point in favor of what they might consider weightier, alternate evidence. The CBGM never forces our decision, it only provides additional evidence, evidence that must be interpreted. But it is to misread the evidence from the textual flow diagram to think that they suggest that 03 was the historical exemplar of P72, 01, and 1739. This is precisely what Tommy and I warn against in our book (see p. 92).

Bordalejo and Robinson also seem to think there is some difficulty with the reading ὀνόματι but I don’t see it and their appeal to the CEB and Good News Bible is a bit odd—as is their citation of the ETC blog as an example of “groups of fundamentalist Baptists with the motto ‘King James Only’, and a group of well-qualified scholars who assert the value of the Byzantine text under various labels: as ‘textus receptus’ or the ‘majority text’” (p. 18)!

Robinson and Bordalejo have long been at the forefront of so much in the field of digital stemmatics and I have profited from reading so much of their work. In this one case, however, I think a misunderstanding has led to wrong conclusion about the CBGM.

49 comments

  1. I spoke with Peter (Robinson) regarding this issue in our Studia Stemmatologica meeting a couple of weeks back...the problem seems to be that we now have two promising computer-assisted stemmatological approaches, CBGM and New Stemmatics, with very little interaction between them. This has led to unnecessary confrontations between these viewpoints. I think that there is too much of jealous attitudes toward the methods we are using, when in fact, we should learn from one another. I believe that collaboration between CBGM and New Stemmatics would benefit us all. How great it would be to bring all the prominent stemmatological minds (Mink, Wachtel, Robinson, Howe etc.) around the same table and try to solve the problems we face together...

    ReplyDelete
    Replies
    1. Thanks for sharing. I agree that collaboration and comparison of methods would be good. Wachtel did that with Howe and Spenser some time ago in fact.

      Delete
    2. I don't think that the article you referred is particularly good example of the reasonable collaboration between these approaches I was talking about, quite the contrary in fact. Wachtel, for example, seems to think in the interview conducted by Lin, couple of years after that article, that evolution and progress is one and the same thing. This is simply a misunderstanding of the nature of evolution. In other words, the lack of interaction between CBGM and New Stemmatics leads to successive misunderstandings and unnecessary confrontations between these approaches.

      Delete
  2. PG,
    Of course Robinson and Bordalego misread the Textual flow diagram, how could they not? The mystics in Muenster designed the Textual Flow Diagram to look exactly like a traditional stemma. All argument to the contrary, even after reading broadly and deeply on the CBGM, the TFD is in fact treated, even by proponents, as if it was a real stemma. Of course, as you point out, there are always statements in the pro-CBGM literature that make it seem like true believers realize that CBGM is just another ‘tool’, yet as you did, these same believers always point out the obvious reasons why any variation from the CBGM is mistaken.
    Surely, ETC is not a KJVO or fundamentalist Baptist site, but it is clear that some authors on ETC, despite protestations otherwise, are in fact CBGM-Onlyist.

    Tim

    ReplyDelete
    Replies
    1. Are you a TR, Byz, or KJV proponent yourself, Tim?

      Delete
    2. PG,
      I am not. I am a proponent off documentary evidence, particularly the earliest manuscripts.
      Tim

      Delete
    3. I have to say, Tim, I can't agree with your assertion "how could they not?"

      Back when I was first learning textual criticism (from a short little chapter written by Mike Holmes, no less) I was taught, like most students I'm sure, to prefer the reading that could best explain the others. Thus, if reading A was shorter while reading B was a longer version of A, then - assuming we still ascribe to the lectio brevior heresy - we would prefer A and conclude that B was simply an expansion of A. In that sense, B "flowed" from A.

      Lectio brevior aside, I think most critics would agree that's a normal way to handle internal evidence.

      I also think that most normal critics would not then immediately assume that whatever manuscript A was found in was therefore an historical exemplar of whatever manuscript was attesting B! I mean, we might draw internal conclusions from such internal evidence, but I don't know anyone who would then run off and draw such manuscript conclusions. Do you?

      In every introduction to CBGM I've ever read - and I was in Muenster at that time and actually helped a little bit in editing the first introductions to it - they were extremely careful to explain that they were showing the flow of readings, of texts, not the historical manuscripts that bore those texts. I think they've been more than clear on that point.

      Delete
    4. Juniper,
      Of course you don’t agree, as I articulated and you confirmed, anyone who has been associated with Muenster can’t see any problems with CBGM or its explanation, no matter how many scholars or lay people tell them that it appears to be voodoo. Repeating the Muenster Mantra doesn’t clarify anything.
      This probably sounds harsher than I mean it. I want to believe that the CBGM can provide help with contamination. The concerns I expressed above I believe are legitimate. Muenster could help themselves if they tried answering the concerns raised even here on the ETC blog with something other than the same party line.


      Tim

      Delete
    5. Tim, (Juniper is my wife's account), I'm not sure you're helped much assuming that anyone who disagrees with you must "of course" be doing so simply because they've been bought off or brainwashed by the other side. This is textual criticism, not american politics.

      Isn't there a chance that I'm disagreeing with you not because I'm blindly repeating the muenster party line but because I authentically think that you are incorrect? Isn't there just a small chance that I think that way because the evidence leads me there, and indeed should lead anyone there?

      If you go to the muenster website and click on the introduction to CGBM, there the very first thing you read is:

      "Elements of a genealogical hypothesis are not the manuscripts but the states of the text that they convey and that may be far older than the respective manuscript. The text with its respective state will be referred to here as witness, not the manuscript."

      That makes it clear - undeniably clear - that they are referring to texts, not physical manuscripts.

      If you read just a little further down the page, they also state:

      " the relation between the states of text of the existing witnesses can only be the subject of a genealogical hypothesis. It cannot be the objective of such a hypothesis to reconstruct in detail historical processes. That is impossible. Rather it is the objective to find the structure which interprets in the most straightforward way the genealogical relations between the available states of text and which is not falsified at any place of variation."

      This may just be my opinion, but I think their explanation is clear enough that if anyone reads it and walks away still thinking that a CGBM stemma is showing the literal genealogy of physical manuscripts, well, I think that person is going to have to take responsibility for their own reading.

      If Peter's summary is accurate (and I have no reason to think it is not) then the authors have simply made a blunder.

      I'm certainly not inclined to hold that against them. I mean, it happens, it just does. I've made tonnes of blunders. Huge blunders! And in print, too. They happen. Something catches your attention, you just happen to miss something you normally notice, you just get thinking about something the wrong way. It happens all the time. But if I, for example, make a blunder and, I don't know, misread a critical apparatus - I don't get to blame the apparatus for that.

      Delete
    6. Ryan,
      So you haven’t read the article yourself? Hmmm!

      Delete
    7. Nope! I leave that to the people who get to do this full time.

      Are you saying Peter's summary is not to be trusted?

      Delete
    8. I don't think that the text vs. manuscript distinction makes much of a difference here, since the text of Sinaiticus is not a descendent of the text of Vaticanus.

      Delete
    9. SC,
      This is, for me, a fundamental issue with the CBGM. The idea that a text, isolated from the manuscripts that contain them can somehow establish which one is genealogically earlier. Then to use that determination to decide Coherence, seems particularly problematic.
      Tim

      Delete
    10. The text vs. manuscript distinction serves primarily (in my opinion) to set aside the paleographical evidence for the attested dates of the texts. It is true that a later manuscript can embody an earlier text, but since no text is a perfect copy of another there is always some difference. The CBGM's use of an extant text as a proxy for the text of a potential ancestor is probably OK for the well-sampled Byzantine text, but it degrades for texts close to the initial text. That is why there are so many texts whose first potential ancestor is the initial text, and that state of affairs is what I read Bordalejo and Robinson critiquing.

      Delete
    11. Still in a bit of a rush with class, but let me quickly say, Stephen, that I think you’re spot on to ask about the possibility of things breaking down closer to the initial text where we have less extant witnesses to fill in the gaps. I should have more thoughts on that in a forthcoming piece, I hope.

      Delete
    12. As you well knows, I'm just repeating myself from my earlier remarks in http://jbtc.org/v20/TC-2015-CBGM-Carlson.pdf . I look forward to your thoughts on the question, Peter.

      Delete
    13. Just to be clear. We did NOT misread this diagram. We are fully aware of the distinction between texts and manuscripts. The point of our article was .. well, you might want to read it again. To explain again: finally we have to deal with historical probability, and this means hypotheses about lost ancestors within the tradition. The central conundrum of the article is that in cases where manuscripts do not show evidence of significant shared variation, hypotheses about lost ancestors are problematic. To put it another way, you need a "global stemma". Or, a way of working towards historical understanding without a global stemma. Or something...

      Delete
    14. Peter Robinson, apologies for my lack of interaction. I’ve been too busy to follow all the comments. But let me add that the global stemma accurately incorporates all the local stemmata in the CBGM. So it should be quite able to accurately trace this development in 1 Peter 4.16. And, in fact, the top portion of the global stemma for the Catholic Letters IS available. Tommy and I show it on page 97 of our book.

      Delete
    15. By “accurately,” of course, I mean “in accord with the editors’ theory at each point of variation” not necessarily “in accord with what actually happened.” We are working with historical reconstructions and all the difficulties that entails. As to the relation of history and the CBGM, see my chapter in my dissertation where I attempt to deal with that issue at length.

      Delete
  3. Some of the comments from P. Robinson and Bordalego are problematic; yet the point seems valid that if the least complex textual-flow diagram implied by the CBGM is historically impossible, what good is it?

    ReplyDelete
    Replies
    1. It’s good for showing us coherence among our witnesses. And that’s good for detecting how the text might have changed. Textual flow diagrams are like maps, but not detailed topographic ones. They’re more like subway maps.

      Delete
    2. Those who have worked more closely with the method are welcome to correct me where I'm wrong here, but this is the impression I have of the textual-flow diagram within the larger CBGM process.

      The textual flow diagram for a given variation unit is part of an iterative circle of steps in the CBGM workflow. Each iteration starts with two inputs: (1) pre-genealogical similarity calculations, which are fixed and do not depend on user input; (2) local stemmata of readings (including the initial or "A" reading) at each variation unit, which are oriented by users on the basis of existing methodology or conclusions about coherence made in previous iterations.

      If the user is certain about the initial text and the local stemma at a given variation unit, then nothing else needs to be done for that unit. But for variation units where further evidence is needed, the textual-flow diagram can be used to evaluate coherence under different hypotheses about that unit's local stemma, and the results of the evaluation can help users revise this local stemma for the next iteration. The textual flow diagram, despite looking like a stemma, is basically just an abstract representation of a calculation being done on the textual witness: for each witness, we're asking, "Which other witness is closest to this witness (based on pre-genealogical similarity across all variation units) and has the same reading or a parent reading (based on the local stemma of the given variation unit)?"

      Historical impossibility is a bigger issue for the global stemma, which is an (intended) output of the CBGM that combines and accounts for the textual flow relationships established at all variation units. Robinson and Bordalejo's criticism about historical inaccuracy would have been better directed at a global stemma, although it's worth pointing out that even the global stemma is not intended or able to capture the history of manuscripts very well, because, as they point out in their paper, the CBGM does not use hypothetical ancestors in the construction of its global stemma. It's also worth pointing out that they wouldn't have had a global stemma to apply their criticism to, because no one has been able to produce one for reasons of computational complexity.

      Delete
    3. What global stemma? As far as I am aware, none has been published yet...

      Delete
    4. Yeah, that's one of the last points I made in my previous comment; as far as I know, no one's been able to generate a global stemma for all of the witnesses collated in the catholic epistles (although, if I recall correctly, Peter was able to produce a global stemma for the Harklean group and some Byzantine witnesses in his paper "The Harklean Syriac and the Development of the Byzantine Text: A Historical Test for the Coherence-Based Genealogical Method (CBGM)").

      I think that for me, the difficulty of producing global stemmata and their resulting absence in published literature is an even bigger obstacle to the CBGM's credibility than the question of what those stemmata would represent. The advantage of more traditional stemmatic approaches is that they have more performance-optimized tools for their tasks, which gives practitioners of stemmatics more to show for their work in the end.

      Delete
    5. Right, it is possible to produce a global stemma for 1 Peter according to known phylogenetic techniques, but not, as far as I am aware, according to the published methods of the CBGM. (There's a simple proof using complexity theory for this.) Nevertheless, it is difficult to resist the urge to think stemmatically about the textual tradition, even with repeated disclaimers to that effect for the textual flow diagrams of the CBGM.

      Delete
  4. Thanks for this Peter Gurry. I appreciate you and Tommy Wasserman's continuing efforts at blowing away the fog and helpfuly clarifying the CBGM at points of confusion.

    ReplyDelete
  5. I for one am not a baptist.

    ReplyDelete
    Replies
    1. Am I the only one who heard that in Richard Nixon's voice??

      Delete
    2. Some of us are Baptist, but hardly of the KJVO variety.

      Delete
    3. Of course I am a fundamentalist.

      Delete
  6. I am still wondering ... Absent the CBGM and without worries over the resultant "flow" involving specific individual MSS, it seems much easier to see μέρει as the contextually more difficult reading, and ονόματι as due more to attraction from 4:14.

    In other words, a more traditional approach that apparently works just as well without the computerized mysteries (but then I am biased, and just not as open-minded to current technological claims, right?) .

    ReplyDelete
    Replies
    1. I haven't seen any argument, however, prior to Mink that μέρει is the harder reading in 1 Peter 4:16. Personally, I think ἐν τῷ μέρει τούτῳ ("in this respect") makes great sense on the surface.

      Delete
    2. Interesting. I’ve read a number of commentators who reject μερει as original but but admit that it’s the more difficult reading internally. Schreiner is one that comes immediately to mind. I think Tommy and I cite him in the book.

      Delete
    3. Schreiner is not prior to Mink.

      Delete
    4. His 2003 commentary? I need to check at home but I’m pretty sure he wasn’t using the ECM for that and I know he doesn’t refer to Mink in his discussion of that variant.

      Delete
    5. Good point that he could be independent of Mink, which counts for something.

      Delete
    6. Ummm... I am quite independent of Mink. Does that count?

      Delete
    7. You always count in my book, Maurice.

      Delete
  7. I have just come across this, and I have been reading the comments with great interest. Firstly: the version of the article you have been reading is a draft only. It is being changed even as you read this, and the revised version will be on-line later today. The changes are mostly refinements along the way of being fairer to the CGGM -- or rather, to reflect our acceptance of the soundness of the fundamental approach, and to refine away some comments which appear to result from our misunderstanding. So some of your favourite quotables may be missing.

    Secondly: I think we do understand very precisely the difference between texts and manuscripts, and one of us (Bordalejo) has actually published an article on just this subject in Digital Studies in the Humanities (see her Academia site). The comments we make of the CBGM are not a result of misunderstanding that crucial difference. As the title of the article implies: the fundamental problem is to do with manuscripts which cannot be grouped into families. To put it another way, this shows itself as the impossibility of creating a convincing "global stemma". That is a problem with all four traditions surveyed in the article. It is particularly acute in that one cannot establish a global stemma for the witnesses closest to the original. I think our observations centre on this. For the later reading to be historically possible at this point in 1 Peter, one would need to hypothesize the existence of an archetype which had this reading, but which somehow failed to be copied by any of the extant early lines of descent up to around 800 AD. This is exactly the same question Dante scholars face with the existence (or not) of "alpha", a hypothetical subancestor distinct from the original from which (if it ever existed) almost all the extant mss descend.. or not.
    I second Pasi Hyytiäinen's comment: a meeting (NOT a confrontation!!) between people from the different approaches would be a really useful idea. There is a lot of goodwill here (many of us have been collaborating for decades, in different ways). I don't believe there is any fundamental incompatibility between the different approaches. And there is a very definite need for better tools and better understandings. This story is just beginning.

    ReplyDelete
    Replies
    1. Peter, I'm in meetings and teaching all day today and hope I can respond more later. But for now let me just say thanks for responding here and I am glad to hear there is much goodwill (as I assumed there was given the close association over the years between you and Münster). In my experience, those working in computer stemmatics have been very open to learning from others.

      Delete
    2. Peter (Robinson), I appreciate your critique of one aspect of the textual state of affairs implied by the CBGM, and I look forward to seeing the revised paper in publication.

      Delete
  8. Ramsey Michaels4/24/2019 12:44 pm

    Mostly I just lurk on this list, and greatly enjoy it. But I can't resist calling your attention to my 1988 Word Commentary on 4:16, in support of merei, with reference to 2:12 and 3:16.

    ReplyDelete
  9. New Testament Textual Criticism is a most humbling field,-even "a perilous field" as E.Miller calls it. In my estimation it can take 10-20 years of constant study just to get your feet wet. Accolades and expertise aside, P.Robinson and Bordalejo have great courage (and should be applauded for such!) in trying to tackle CBGM. Especially, coming from slightly differing fields of study. With that said, I believe the "Science" of NTTC is in need of more emphasis on the "Holy Spirt" and less emphasis on computer analysis (as very odd as that might sound). The reason being is quite simple, the Sacred Deposit is the birth right and inheritance of the body of Christ, not secular Academia (or "Christian" Academia for that matter). -Grace and peace

    ReplyDelete
    Replies
    1. Indeed, I have been involved in TC since 1997, before there was much of anything available online (the very first website I ever viewed contained digital images of the DSS), and I feel that I am just now beginning to get good at it. I should say something about the phobia of 9th century readings: most of the readings in the Hebrew Scriptures don't even go back that far, and no one seems to be ready to toss out the first 2/3rds of our Bible over it. The fact is that there are very few Byz mss found in the sands of Egypt, and those few are very small or confined to the Gospels only. Extrapolate their text out to the entire NT, and you would have no lack of 4th and 5th century readings that otherwise "don't appear until the 9th century" (as if stepping out of the 9th century is something currently being done). This phobia is both unnecessary and unreasonable, and I'm thrilled to see that someone has finally been able to get past it in the pursuit of truth.

      Delete
  10. A huge advantage of CBGM, if I understand at least that part of it correctly, is that it is blind to age bias in manuscripts, something no human critic seems to be capable of.

    ReplyDelete
  11. Daniel,
    Did a MT/TR guy just find something good in the CBGM because it doesn’t favor the earliest readings?
    The last I understood, the CBGM actually is dependent on the decisions by the committee in the beginning. Therefore, it is the committee who has found new respect for the Byzantine Text which causes their CBGM results to be the way they are. I see the editors ‘circular reasoning regarding the Byzantine Text as a downfall of their output.

    Tim

    ReplyDelete
    Replies
    1. I don't have much respect for the TR other than that, for example in Galatians, it differs from the MT in only six places, compared to hundreds for Aleph and B (which are as different from each other). So if the TR were a mss, it would be, if I read Stephen Carlson right, the second most accurately copied book of Galatians in existence (in my opinion). But yes, I'm glad to see the CBGM output a reading "despised" by Westcott & Hort, which human critics of their ilk have been unable to accept for a century and a half, despite mounting evidence of the antiquity of more and more "distinctive Syrian readings" since 1899.

      Delete
  12. Peter, a broad CBGM question if I may. I haven't seen (as far as I can tell) some historical observations or data in favor of the four principles of parsimony in your monograph. How do Mink et al. go about justifying these? For example, principle 1 seems valid as far as it goes, but I could see more value in weighting it against some kind of metric. When you say (regarding principle 2) that a scribe varying from his main source is more likely to have done so with reference to an additional source, do you mean by "vary" the kind of variation one might encounter in a passage of continuous text, or simply any kind of deviation (however minor)? Again, is there any historical basis for such a tendency? In the case of principle 3, although I'm not familiar with the historical situation in monasteries, scriptoria, etc, I could easily imagine there being a broad range of variation in terms of the number of manuscripts consulted (or available on hand) as sources. Is this meant to only be the most general of tendencies?

    My apologies in advance for the jumble of questions!

    Kaspars

    ReplyDelete