Friday, June 12, 2009

A New Rating System for UBS GNT

United Bible Societies are ahead of their time! The July issue of Bible Translator (Technical Papers) is already out, just in time for the TTW –Triennial Translation Workshop, which is being held in Bangkok right now.

In this issue I propose a new rating system for Greek New Testament editions. It is a revised version of a proposal that was included in my monograph on Jude.

Tommy Wasserman. "Proposal for a New Rating System in Greek New Testament Editions." Bible Translator 2009 60/3:140-157.


Various rating systems have been employed in critical editions of the Greek New Testament ever since the eighteenth century. The current letter-rating systems in the UBS Greek New Testament editions have been subject to severe critique ever since they first appeared in the first edition of 1966. In my experience, however, there is still a need for this kind of basic guidance among Bible translators and students of the New Testament.

In this article, I present a new descriptive rating system based on the generally accepted principles of textual criticism that takes into account external and internal evidence. This, in itself, is a pedagogical advantage that may enhance the individual learning process and the increased engagement with the textual problem in question. The rating will show the state of the evidence from the perspective of the editor(s), and, at the same time, it will invite the users of the edition to struggle with this same evidence on their own.

Thus, the following symbols are used to describe the evidence:

{e + i}
External and internal evidence unequivocally support the adopted variant reading.

{e > i}
External evidence favors the adopted variant reading, whereas internal evidence is ambiguous.

{e < i}
External evidence is ambiguous, whereas internal evidence favors the adopted variant reading.

{e = i}
External and internal evidence are balanced or, alternatively, external evidence favors one variant reading, internal evidence another.

(The latter symbol is roughly equivalent to the bold dot in the ECM.)

In the article I provide several examples of these various ratings in variation-units in the Epistle of Jude. Scholars and editors can use this system when they make textual decisions in any given unit of variation, weighing external and internal evidence, however they are perceived. I propose it for the future editions of the United Bible Societies' Greek New Testament.


  1. My frustration in text critical argumentation is the following: Quite often 'intrinsic' and 'transcriptional' evidence point in opposite directions. Those working with a preference for part of the external evidence, tend to pick the 'internal evidence' that strengthens their choice, which was decided almost exclusively on external grounds.

  2. your system is sage. i hope the ubs editorial board is paying attention and will adopt it in their next edition.

  3. I don't think the GNT needs ratings at all. Ratings are cages: a person could deduce, by judging the size of an animal's cage, the size of the animal that resides in that cage. But what is the point of that when the animal itself is right there in the cage?

    Yours in Christ,

    James Snapp, Jr.

  4. Thanks for the comments.

    Gie, according to my system such decisions would then become more transparent, indicated by the symbol e>i which means internal evidence is ambiguous and that the decision was made mainly on the basis of external evidence. It is up to the editor to explain how they perceive this external evidence.

    In the article I also point out:

    "The perceived state of the evidence corresponds to the relative degree of certainty behind the various decisions. If, for example, a variation-unit has rating {e=i} the evidence is ambiguous and the decision highly subjective. Of course, the indivdual scholar or editor may in many cases come to wholly different conclusions depending on the adopted overall view of the textual transmission, and of the relative weight assigned to external and internal evidence, respectively."

    Rating systems has advantages and disadvantages. My system aims to be transparent and pedagogical, and invite the reader/user to struggle with the evidence for themselves.

    Yes, I hope they will adopt it. They are apparently interested. It was Roger Omanson who asked me to explain my system. At one point I presented it in Birmingham, and one UBS representative who was there at the presentation liked it very much. I also expect a review of my monograph on Jude by Harold Scanlin to appear in the Bible Translator, so it will be interesting to see what he thinks.

  5. Tommy, Congratulations on this excellent system. I certainly will employ it in my own work in textual criticism, and in my teaching of exegetical methods at Fuller Theological Seminary. I hope our paths cross at the Rome SBL conference. Peter Rodgers

  6. I agree, Tommy, that your system would be an enormous step forward. It would be even more transparent if 'intrinsic' and transcriptional would get different codes. I consider them more 'opposite' than 'external' and 'internal', as 'external' and internal could lead to the same conclusion,while 'transcriptional' and 'intrinsic' more often than not are in opposite camps.

  7. GV: "It would be even more transparent if 'intrinsic' and transcriptional would get different codes."

    Gie, I want to keep the symbols simple (and one may for example deduct from e>i that internal evidence is ambiguous), but such explanations should accompany the symbols.

    In Jude 14, for example, I use the symbol e>i and after the discussion I summarize thus:

    "In conclusion, external evidence definitely favors προεφήτευσεν in Jude 14, whereas internal evidence is ambiguous; although both forms were in use during the 1st century, the NT authors seem to have preferred the form ἐπροφήτευ(σ)-. The author of Jude, however, known for his competence in Greek, could well have used the other form, which may have caused some scribes to have harmonized it to the form most common in the LXX and in the NT."

  8. I agree with Tommy that some sort of rating system is necessary for the student and can be an excellent pedagogical device.

    I also agree that the current system is deficient in many ways, and that Tommy's proposal would be a huge step forward (plus, it would make the apparatus look even more esoteric and advanced, which would make me look even more like some good-will-hunting like genius when I expertly interpreted the dense sequences of symbols and letters, and that, in turn, would make me seem even cooler, and frankly I need every advantage I can get!)

    I really agree with Gie, however, that transcriptional and internal should really be separated. The two are so different in both logic and practice. The only thing they really have in common is that they are not "external" (though even that distinction can begin to blur a bit in regards to transcriptional if you begin to look at the scribe behind the tendency or the scribe behind a particular manuscript). It's been a pet peeve of mine for so long that transcriptional and intrinsic are lumped together under "internal" that way, and I think doing that is less helpful that more.

    Tommy, it looks like your current system works more easily with a binary comparison, but could there be some way to accommodate a tripartite distinction between external, transcriptional and intrinsic?

    Perhaps if we assigned 5 symbols, one for each of the following:
    -strongly supports
    -lightly supports
    -lightly opposes
    -strongly opposes

    then for each reading a 6 character code could be given, consisting of three letters (e for external, t for transcriptional, and i for intrinsic), each letter followed by the appropriate symbol?

    Thus, let's say (picking symbols randomly) that

    -strongly supports = #
    -lightly supports = \
    -ambiguous = ?
    -lightly opposes = /
    -strongly opposes = *

    then a given reading could be coded with something like:


    Ok, looking at that now, that looks fantastically terrible, but perhaps we're onto some sort of start of an idea maybe? Maybe if we used numbers instead of symbols?

    Just some thoughts. Looking forward to reading your article.

  9. -strongly supports = #
    -lightly supports = \
    -ambiguous = ?
    -lightly opposes = /
    -strongly opposes = *

    Make it intuitive:

    -strongly supports = ++
    -lightly supports = +
    -ambiguous = ?
    -lightly opposes = -
    -strongly opposes = --

    e#t/i? becomes e++t-i?

    Or simplified by just using «+» for supports and «-» for opposes and «?» for ambiguous, doubling the symbols only in extreme cases.

    I'll let y'all figure out the textual ramifications of using «+» vs «-».


  10. minusrusty,
    that's a great idea! In fact, taking it further, maybe we don't even need the qualifiers of "light" or "strong", after all, that's pretty subjective anyway. Perhaps then we could just use a simple "+" for "support" or "-" for "oppose" and the "?" for "ambiguous." Thus:


    Or, we could simplify it further and drop the "?" and just have it understood that an unqualified letter is automatically taken as ambiguous, thus:




    or if all three are ambiguous:


    That's looking better already I think.

    Maybe we could even simplify it even more by making a negative apparatus! Have it assumed that a given area (i.e. e, t, or i) supports the reading unless it is cited otherwise in the apparatus. In the apparatus, if the area is ambiguous then we can just give the letter, while if it opposes we can give the letter plus the symbol "-".

    Thus, a reading that is supported by external but opposed by transcriptional and intrinsic would look like this in the apparatus:


    a reading supported by external and transcriptional but ambiguous in intrinsic would look like this


    a reading supported by all three would look like this:

    while a opposed by external and transcriptional but supported by intrinsic would look like this:

    and so on.

    Just thinking out loud here.

  11. For your information, this is the frequency of the ratings I indicated in Jude:

    e + i 74 times
    e > i 14 times
    e < i 7 times
    e = i 6 times

    In the great majority of the cases evidence was unambiguous.

  12. The problem with the various {e} and {i} combinations is that in the end they still represent only the opinions regarding external and internal criteria that happen to be implemented and favored by a particular set of editors.

    I am certain that some of Tommy's evaluations in Jude would differ from those in UBS or ECM; similarly, my own application of the {e} and {i} ccmbinations would differ dramatically from those of the UBS editors.

    All this simply demonstrates the point Burgon made long ago:

    "Inasmuch as one expert's notions of what is 'transcriptionally probable' prove to be the diametrical reverse of another expert's notions, the supposed evidence to be derived from this source may, with advantage, be neglected altogether .... Notions of 'Probability' are the very pest of those departments of Science which admit of an appeal to Fact." (Revision Revised, 252).

  13. Maurice, that is why I say that "the individual scholar or editor may in many cases come to wholly different conclusions depending on the adopted overall view of the textual transmission, and of the relative weight assigned to external and internal evidence, respectively."

    However, I think that if the editor explains his or her overall view of transmission, the rating system itself can be useful, also in the case of a Byzantine priorist.

    On the other hand, a radical eclectic would not care much about the "e".

    In any case, this is a proposal specifically for the UBS edition, whose editors work with the generally accepted principles of external and internal evidence.

  14. Maurice, you touch on a great point that I was just discussing with a friend last night.

    Often times, I think, we over-estimate our ability to assess probability, and even then, I find that the truth often has an inconvenient habit of defying the odds.

    Completely unrelated example, but a great story all the same:

    A couple weeks ago I was sitting in the parking lot of our building getting ready to switch the snow tires for summer tires on our car. Working on the front first, I put the wheel chocks on the back tires, got the jack stands ready by the front, and slid the jack under the centre and was just starting to jack the car up with something interesting happened: A strange man, taking a short-cut through our parking lot, said to me as he walked by: "you'd better switch those chocks around to the other side!" and then he kept walking. I checked, and sure enough in a senior's moment I had put the chocks behind the tires rather than in front! Had I jacked the car up like that, given the incline of the ground, it liked would have rolled off of the jack right on top of me! Good thing that guy walked by when he did, right?

    But then I started to analyze the probability of what happened:

    I live in a small town of about 6000. Of that 6000, how many would have sufficient knowledge of auto mechanics that with one half-a-second glance they could identify misplaced wheel chocks? I mean, a larger number could probably figure out correct chock placement if given the time to think about it, but how many people would know it so well that they would know it instinctively, in the split second as they walked by? Surely only a small number! Let's say even a couple hundred in this town of 6000 had that level of expertise. But wait, of those few hundred, how many of them would be in the end of town that I live in on that saturday afternoon? Certainly only a fraction of that few hundred. And of that remaining fraction, how many would choose to trespass on our property, taking an illegal shortcut? An even smaller fraction, I am sure. And of that smaller fraction, how many would be taking that short cut at the exact moment I was starting to jack the car? Remember, there was a very small window or time here: 30 seconds too early and I wouldn't have placed the chocks yet; 30 seconds too late and the car would have already fallen. This guy had about a 20 second window, and he nailed it. And finally, assuming that some of that ever so small remaining fraction were cutting through our lot at exactly the right time, how many of them would have the gumption to say something to a stranger? We're canadians here, we're shy. I see strangers doing dumb or unwise things all the time, but I'm certainly not in the habit of correcting them! But this guy did.

    So given all those variables, from the necessary expertise, to the timing of walking through the lot at exactly that moment, to being willing to say something, how would we critically evaluate the probability of my story actually happening? Surely most scholars would assess it as having a very, very low probability! And technically, we would be right! It's fantastically improbable! And yet, it happened.

    And so that's the kicker of probability, and I've been thinking about that a lot lately in relation to our text critical decisions when we start waxing authoritatively about what a scribe would or would not have been likely to do.

    On the other hand, we do have an obligation to make our best scholarly assessment, even if it's not perhaps as certain as we would like it to be, and I think the subjectivity you are worried about can be largely mitigated by the types of consultations and committee decisions that the UBS text uses.

    So, bottom line, I agree with and commend your caution about our use of probability, but I still think we have the ability and the duty to make some assessment of it.

  15. I like it. I don't mind making up my own mind on the [e] and [i] possibilities (with the input of scholars who have spent some serious time and effort on the subject).

  16. [also posted at textualcriticism list]

    I have not yet read the article, but I'm sure it is very well reasoned and full of examples.

    My concern with ratings of any kind is the methodology underlying the ratings. Who decides? I would imagine that for most critics "number of mss" would not play any significant role in defining the weight of external evidence. Yet for others, number remains one of several important external criteria, among which are also provenance, age, scribal habits of particular mss (such as omission for Sinaiticus), etc.

    Pay particular attention to the final sentence of Griesbach's note in his Critical Commentary concerning the retention of the Majority text's WBHD instead of IWBHD in Matt 1:5: "For WBHD in this place, as in Luke 3:32, a few ancient manuscripts and some versions have IWBHD. Thus also in the books of Chronicles, uniformly in codex Alexandrinus and sometimes in others, this name is written. Many Hebrew names that appear in both Testaments begin with IW. Names of this kind, perhaps even the very celebrated IWB, appear to have been in mind for some scribes. Hence they imitated more easily also in this place the orthography that they remembered to have found in certain of the manuscripts. The form of the name consistent with Hebrew origin must be retained, and is confirmed by the majority number of manuscripts, even of ancient ones."

    Certainly number was not the deciding factor for Griesbach in retaining WBHD in Matt 1:5. Yet he probably would not have mentioned if it did not retain at least some critical value.

    I hope, if any (new) ratings are adopted, that number of mss might be allowed to play at least one of many supporting roles.

    Jonathan C. Borland

  17. To take another example, in Matt 1:6, the omission of O BASILEUS following DAUID DE would probably receive the rating of strong external and internal evidence in its favor according to prevailing critical methodologies. I.e., the "earliest and best" mss omit it (external), and scribes tended to add (internal).

    Yet the inclusion of O BASILEUS is not only witnessed by a large number of witnesses, but also has a good internal argument in its favor as well:

    Wettstein: "“In J. Mill's judgment (Prol. § 1245, 1471), [ὁ βασιλεύς | O BASILEUS] crept in from the preceding words. In our opinion, however, Matthew wrote both occurrences with a certain plan to honor Jesus by making a repeated mention of the dignity of his ancestors, among whom David in particular is numbered. Compare 1 Kings 1:1, 13, 28, 31, 32, 37, 38, 43, 47, where it is clear that this kind of repetition is not uncustomary for sacred writers.”

    See also Griesbach in his commentary: “ὁ βασιλευς | O BASILEUS is omitted in some, because in all the other members of this genealogy ἐγεννησε | EGENNHSE follows immediately after the man’s name and the particle δε | DE. Hence a scribe, accustomed as it were to the rhythm, after δαβιδ δε | DABID DE hurried at once to the ἐγεννησε | EGENNHSE. Mill thought it crept in from the preceding occurrence. I would concur, had the preceding δαβιδ ὁ βασιλευς | DABID O BASILEUS been in the nominative case.”

    So as Dr. Robinson (and Burgon) said, what is strong internal and external support for one comprises just the opposite for another. Perhaps it is better just to let the ratings rest in peace.

    Jonathan C. Borland

  18. Although ranging beyond the rating issue, let me address Ryan's wheel chock illustration (which seems to imply an act of Divine Providence, if I read him correctly).

    Our cat likes to play with a small rubber ball. We throw it to him and he bats at it, usually hitting it with the skill of a minor league baseball player. Today he not only hit the ball, but sent it flying across the room, directly to end up inside a very small bag with a very narrow opening (2 points!).

    The odds against anything of the sort happening again would be well-nigh impossible (and I suggest the event is virtually non-repeatable) -- but it did occur (Providential involvement is not suggested in this case).

    But, in reality, what difference then exists between such parallel instances of statistically improbable but actual situations?

    And is there a point to be made thereby?

  19. Ryan: "Or, we could simplify it further and drop the "?" and just have it understood that an unqualified letter is automatically taken as ambiguous, thus: e+ti-"

    Hi Ryan,

    No it would be best to stay explicit. The more stereotypical you can keep the code, the easier it will be to read and intuitively understand, much as standardized spelling makes reading easier, kwim? ;-)

    I agree that using only single signs is good, but I also think there are limited cases where being emphatic can be important, too. (I'm specifically thinking about the Johannine Comma and having e++, where there really is no controversy, but much rhetorical concern in some circles. Like I said before, those would be exceptional cases.)


  20. Ryan: "So given all those variables, from the necessary expertise, to the timing of walking through the lot at exactly that moment, to being willing to say something, how would we critically evaluate the probability of my story actually happening? Surely most scholars would assess it as having a very, very low probability! And technically, we would be right! It's fantastically improbable! And yet, it happened."

    -R: But you also have to balance that individual instance probability across the number of times and places that such a thing can happen. How many times do people place chocks underneath cars in a given day? How many times do strangers trespass in a given day? How many times do people speak up when someone is doing something dangerous in a day? How many days are there in a year, and in a decade, and in a century?

    An individual instance might be one-in-a-million, but with billions of people, and hundreds of billions of interactions among them in any given day across years and years of time, the probabilities can easily balance out often such that the nearly-impossible happens all the time.

    And with the human habit of selective memory... Well, that's just the way the world is.

    How this applies to textual criticism, however, is probably more of an art, but I'm sure there are trends that can be shown and applied to the textual data at hand to yield more-or-less scientific wags. ;-)