Wednesday, January 23, 2019

The All-or-Nothing Problem with Byzantine Priority

16
The most noteworthy feature of the Byzantine priority position to me is not that it seriously values the Byzantine text (I think that’s generally good); it’s that it always prefers a Byzantine text. It’s that “always” that I’ve never been able to fully stomach. I find it incredible that the text found in any one manuscript or subset of manuscripts is always right.

This is for reasons both logical (could one scribe or series of them really get it right every time?) and empirical (I haven’t actually found a manuscript or group that always convinces me it’s right).

In the case of the empirical reasons, I primarily have in mind internal evidence, evidence that I don’t see how a Byzantine prioritist can use consistently. I’ve written about this before and today I want to look at another example of the same problem. Here is the text of 1 John 4.19 in the RP2005 edition of the Byzantine text and in NA28:
  • RP2005: Ἡμεῖς ἀγαπῶμεν αὐτόν, ὅτι αὐτὸς πρῶτος ἠγάπησεν ἡμᾶς. 
  • NA28: ἡμεῖς ἀγαπῶμεν, ὅτι αὐτὸς πρῶτος ἠγάπησεν ἡμᾶς
The only difference is the minor one of the direct object of the verb ἀγαπῶμεν (RP2005 note no differences within the Byz tradition itself). Now, how would someone who thinks the original reading is always found in the Byzantine manuscripts explain the omission of αὐτόν here? The simple answer is homoioteleuton. The -ν ending led some scribes to omit αὐτόν by accident and the scribes of the Byzantine manuscripts preserved the longer form faithfully for us.

The problem with this explanation arises when we consider the other readings at this point. Here is the data from the ECM:
  1. omit 02. 03. 5. 6. 323. 424C. 623. 945. 1241. 1243. 1739. 1852. 1881. 2464. L:V. K:S. G:A1. Sl:E 
  2. αυτον 044. 88. 94. 104. 181. 254. 307. 321. 398. 453. 459. 467. 720. 915. 918. 1523. 1524.1678. 1836. 1838. 1844. 1846. 1875. 2186. 2298. 2492. 2544. 2818. Byz [424*]. G:G-D. Sl:ChMSiS 
  3. τον θεον 01. 048. 33. 61. 81. 206. 218. 326. 378. 429. 436. 442. 522. 614. 621. 630. 642. 808. 876. 1067. 1127. 1292. 1359. 1409. 1448. 1490. 1505. 1563. 1611. 1718. 1735. 1799. 1831. 1832. 1837. 1890. 2138. 2147. 2200. 2243. 2344. 2374. 2412. 2541. 2652. 2805. L:T. K:BVV. S:PH. A. Ä
  4. τον κυριον 629
  5. lacunose P9. P74. 04. 025. 0245. 0296. L60. L590. L596. L1126. Pr
Since all three of the longer readings end in ν, how does a Byzantine prioritist decide which of them is original? He might reason, easily enough, that readings three and four are natural expansions of the simple pronoun αυτον. And so I would agree. The problem, however, is in giving a convincing reason why that same argument doesn’t also apply to reading two. If scribes wanted to make the implicit explicit here as all must agree they did (why else would we have these particular variants here?), then why shouldn’t reading two be rejected for that very reason? I can’t see a way around this except to accept reading one as the best explanation for the origin of the other three readings. (Side note: this is why using a small apparatus can be misleading.)

This, it seems, exposes a flaw in the Byzantine priority scheme. If it cannot use the same internal argument consistently within the same variation, what grounds does it have for claiming that “transmissional considerations coupled with internal principles point to the Byzantine Textform as a leading force in the history of transmission” (RP2005, p. 564)? Here it seems that internal considerations point away from it. It’s true that John’s writings usually have a direct object with ἀγαπάω (though cf. 1 Jn 3.14b; 4.8, 20), but that could also be a further argument for preferring the shorter reading as the harder one.

I confess I would find the Byzantine Priority position more consistent if it said that “transmission considerations” (i.e., external evidence) is decisive and that internal evidence is only needed secondarily for large splits in the Byzantine witnesses. At that point, we would still have to discuss the logical objection I mentioned, but that would be a post for another day. As it is, the internal evidence is a key reason why I find any all-or-nothing approach to the Byzantine text (or any other text) unconvincing. As Tommy et al. have put it: “no single manuscript, manuscript family, or larger group of manuscripts (a text type) can be accepted uncritically as representing the original version of a biblical passage or book...” (Dictionary of the Bible and Ancient Media, p. 414).

Yes, Byzantine manuscripts are very good overall. But that doesn’t mean they’re always right.

16 comments

  1. A canonical text doesn't have to be "always right" it has to be "more than reasonably reliable." And it has to be closed, fixed, final and received by the church as such. A never finished scolars' text seeking what is "always right" can never be received by the church, because no one knows what it will read in the next edition.

    ReplyDelete
    Replies
    1. But it is in fact received by the church around the world.

      Delete
  2. Wouldn't the argument be that some scribes dropped the pronoun by accident and other scribes supplied a referent for the pronoun on purpose? I don't see why the behavior of one scribe must, in the name of consistency, dictate what a completely different must do, particularly when it is alleged that one variant is accidental. An I missing something about the consistency argument here?

    ReplyDelete
    Replies
    1. Good point, Stephen. Maybe my problem is that readings 2 and 3 both explain reading 1 equally well, so how do we decide which one led to reading 1? I guess by asking which of those two best explains the other. But it still seems more economical to say that reading 1 led to both reading 2 and 3.

      Delete
    2. OK, economy is a different argument than consistency, and one that seems more appealing (at least to me) as an explanation.

      Delete
  3. Peter, your opening statement is not fully accurate: "The most noteworthy feature of the Byzantine priority position to me is not that it seriously values the Byzantine text (I think that’s generally good); it’s that it always prefers a Byzantine text." Hodges and Farstad did not do this: 15 minority readings in the PA and 150 in the Apocalypse were published as the text in their edition of the MT. Of course, Pickering said that his mentor Zane had given the theory a black eye, and the RP text has, so far as I can tell, only majority readings (is that right, Maurice?).

    When Hodges taught the course on NT textual criticism in the late 70s at Dallas Seminary, I recall him saying that his genealogical work would prove (he thought) that the Byzantine text was always right. To his credit, he allowed his method to overturn his presupposition 165 times.

    ReplyDelete
    Replies
    1. Dan: "the RP text has, so far as I can tell, only majority readings (is that right, Maurice?)."

      Only where Byz/Gothic "M" is undivided (but this also holds for the Hodges-Farstad edition).

      Where Byz itself is heavily divided, the RP edition, like H-F in many instances (e.g. 1Cor 13.3), follows a reading with somewhat lesser but still substantial numerical support (not always the case with H-F due to their stemmatic approach). Of course, most such occurrences in both H-F and RP occur in the Apocalypse, but in RP the H-F type of stemmatics is absent.

      I should also note that in the PA most significant variant readings necessarily are "minority", given that the primary transmissional lines there (von Soden's μ5, μ6, μ7) often divide so as to produce only "non-majority" variants within that pericope.

      Delete
  4. Peter Gurry: "I find it incredible that the text found in any one manuscript or subset of manuscripts is always right."

    An important part of BP theory is the idea that large elements of the minuscule tradition, the K^X group in particular, cannot easily be collapsed into a "subset" such that they all are basically one voice. Of course no one MS or set of very closely related MSS contain the true text in every particular.

    ReplyDelete
  5. I would suggest that in most units of variation external evidence alone is sufficient to establish a particular reading as likely autographic, yet also suppose that that reading, being authentic, can reasonably be expected to bear a certain degree of intrinsic evidence in its favor. By the same token, many, if not all, errant readings will, upon sufficient study, be seen to bear the marks of error. Hence, I would see Byzantine-priority theory advancing claims of an internal nature in favor of its preferred reading not as superfluous or as self-contradiction, but as “confirmatory” (MAR's word). (As noted, in some units of variation, the external evidence is, at least given the present state of MS collation and classification, equivocal enough to make internal evidence our only recourse identifying the reading most likely to be authorial.)

    ReplyDelete
    Replies
    1. Stephen,
      Exactly, this true for any form of Textual Criticism that relies on external evidence as the primary method for determining the original text, not Just Byzantine Priority.
      Tim

      Delete
  6. I agree with Stephen C. regarding the issue of consistency; rejecting one reading for a particular reason (haplography, harmonization, you name it) doesn't obligate one to reject all other readings that can be explained the same way. Such decisions depend on (1) the probabilities you assign to different types of transcriptional changes (e.g., based on scribal habits) and (2) how external evidence plays into your evaluation of the variant (if it does at all). The only approach I can think of that could possibly demand such strict internal consistency would be thoroughgoing eclecticism, and I don't even think that it does so in practice.

    Byzantine priority, to the contrary, places a premium on external consistency; MAR's emphasis on "transmissional considerations" and his article on "Rule 9" illustrate this point. As Stephen B. has already described, its general strategy is to decide intrinsic probabilities based on manuscript support and to use intrinsic and transcriptional evidence only to break ties in the external evidence or confirm the chosen reading. The effect of this, as you point out, is that the Byzantine prioritist's decision process sometimes appears to ignore transcriptional probabilities whenever they don’t favor the Byzantine text.

    The severity of this perceived problem really depends on (1) the disparity between transcriptional probability in each direction and (2) coherence in transmission. Given a Byzantine reading and a non-Byzantine reading, let’s say there’s a probability p that the Byzantine reading gave rise to the non-Byzantine reading, and a probability q of change in the opposite direction. Generally, the non-Byzantine reading will have coherent support in a small group of witnesses, so we conclude that if it was a scribal innovation, it only arose on one occasion, giving a total probability of p. As Stephen B. has pointed out, Byzantine priority rejects the notion that the Byzantine tradition can be treated as a single later witness, so Byzantine prioritists would argue that the common reading, if it was secondary, could not have coherent support. In other words, for some number k > 1, the common reading must have arisen k times independently. If indeed these k transcriptional events were independent, then the total probability of the Byzantine reading arising as it did would be q^k. So even if, at face value, the non-Byzantine reading better explains the Byzantine reading (i.e., if q > p), it will often be the case that once coherence is taken into consideration, the Byzantine reading will be transcriptionally more likely (i.e., p > q^k). Thus, in practice, Byzantine prioritists only need to demonstrate beyond a reasonable doubt that the Byzantine reading could give rise to the other reading; as long as the probability p isn’t too small, it will do. The most important questions we might ask in this scenario are (1) what the actual values of k really are, and (2) whether or not all k transcriptional events were indeed independent. The CBGM seems capable of answering both questions; classical stemmatics could do even better, provided contamination is accounted for.

    On a related note, I’d be interested to know the coherence for each of readings 1–3 in 1 John 4:19. From a Byzantine-priority standpoint, it wouldn’t be hard to imagine reading 2 giving rise to the relatively popular reading 3 and the singular reading 4 (both by way of expansion), and reading 3 subsequently giving rise to reading 1 (by homoiotelueton).

    Also, thanks for the visual shout-out to the SRGNT! :)

    ReplyDelete
  7. It's encouraging to me that we can so easily discuss the possibility of a Byz reading being the right one. That has certainly changed in the quarter-century that I have been involved in TC. And as Stephen pointed out, it's not a certain family of mss that we look to for the correct reading, but the whole picture. HOW did it turn out that most mss have auton, unless it was in the text to begin with? But Peter, can you decode that cryptic last section of the ECM apparatus for those of us not on the cutting edge?

    ReplyDelete
    Replies
    1. The last section? Do you mean after the fourth reading?

      Delete
  8. It's at the end of every reading, e.g. L:V. K:S. G:A1. Sl:E

    ReplyDelete
    Replies
    1. That's versional evidence. L = Latin, K = Coptic, G = Georgian I think, Sl = Slavonic. The sigla after the letters represent specific translations or manuscripts of those versions.

      Delete
    2. Okay, thanks. So I get Latin: Vulgate and Koptik: Sahidik. Syriak: Phyloxenian and Koptik: Boharik (what is V_v_?). And A angstorm is Armenian?

      Delete