Sunday, May 15, 2011

Textual Critics' Report Card (Part 1)

  Critical Editions of the NT can be rated on a variety of scales, and many of these measures are reasonably objective.  For instance, some obvious and basic categories are:

(1) Completeness of Apparatus:  In many cases, important variants can be left out of an apparatus.  A recent trend has been to 'dumb down' the apparatus for students (and apparently translators!), including only those Variation Units deemed of use or relevance to translations.

(2) Accuracy of Apparatus:  Historically, many a 'good' apparatus has turned out to be based on incomplete or inaccurate collations, which lowers reliability and confidence in support claims for readings.

(3) Accuracy of Reconstructed Text:  The philosophy, theories and methods of, and the data available to various editors significantly affects their results, and this can mislead researchers hoping to use their findings.

Many such scales and ratings are straightforward,  but rating the quality of textual reconstruction can be complex, and involve subjective components.

Reliable Subsets of Variation Units:

One can however turn to the more reliable and solid general data, such as studies of scribal habits and errors, to select Variation Units (VUs) that can be classed by identifiable physical features, such as probable homoeoteleuton errors (h.t.).

No solution to a Variation Unit can be absolutely certain, and all such evaluation must be based on probability.   But scientific decision making in such cases can and will be based on reasonably objective probability estimates, independently of philosophies or personal preferences.

The great majority of special VUs with unique homoeoteleuton features will indeed be homoeoteleuton errors (h.t.).   So, although we cannot know in any individual case its exact cause and transmission history, or even be absolutely certain of its correct identification as h.t., we can rely upon probability to make the reasonable assumption that the majority of VUs with homoeoteleuton features are in fact homoeoteleuton errors (h.t.).

For instance, although editorial glosses and marginal insertions sometimes happen, it is extremely implausible that the majority of such cases would have h.t. features.  Glosses and insertions arise independently in many times, places, and circumstances, and there is no plausible mechanism that would justify any claim that any significant numbers of these would have such features.
Even marginal insertions would be accidental or naive in nature, and would not be deliberately given h.t. features in the process of incorporating them into the text.
Only a very sophisticated interpolator could deliberately incorporate h.t. features into an interpolation.  But now the motive would be lacking.  The majority of  h.t. Variation Units have no theological or historical importance.  They don't support orthodox or heretical doctrines, and they don't impart significant information to the story.   Deliberate edits to the text invariably have doctrinal impact and political motive, but they are rarely disguised in any manner to appear to be something else.  This level of sophistication is simply absent from cases currently known.

In a word, if it looks like a duck and quacks like a duck, its a duck. This makes  Variation Units with h.t. features ideal for testing reconstruction methodology.

Evaluating Critical Greek Texts:

Since it is extremely unlikely that the majority of h.t. errors would be anything else but h.t. errors, any method of reconstruction that fails to handle the majority of  errors correctly must be considered a failure.   This observation can form a basis for checking and evaluating various methods and attempts at NT textual reconstruction.

We have taken as a base the 15 probable h.t. errors found in Matthew, which have been taken seriously as possibly something else (i.e., they are included in most apparatus*).

13 Critical Greek Texts Evaluated:  Click to Enlarge
Alongside each editor is his 'score', that is, his success-rate at correctly identifying h.t. errors and avoiding the mistake of incorporating the omissions into his text.

A word or two on each textual critic is appropriate here, by way of explanation for the scores.

The Success Stories

Hodges/Farstad, Robinson/Pierpont:  These two editor teams have managed to avoid mis-diagnosing 90% of h.t. errors, simply by following a rule which has confined their work to the Byzantine text-type (i.e., follow majority readings).  Since most of the h.t. errors on the list were sourced from Alexandrian manuscripts with little numerical support, they were lost as background noise to the Byzantine stream of transmission.  It may be, in the case of both of these teams, that they came to prefer the Byzantine text as a result of examining the features of other text-types, including h.t. errors. But their method in fact enables them to avoid these mistakes automatically.  But, Matt. 27:35 was rejected by them for the same reasons: The verse isn't in the Byzantine text.  We are not concerned with the correctness of any particular reading.  Our evaluation is only based on probabilities for the VU group as a whole, and the group of readings chosen by each editor as a group.
For whatever reason, these two teams have scored high honours on our report-card.

Bloomfield / Duncan:  Bloomfield, bringing a vast wealth of knowledge to the task, and applying a conservative approach, has also scored equally high honours.  He was not constrained to prefer the 'Majority Text' (note: Matt. 27:35!), and he was quite willing to make reasonable amendments to the TR, in the same manner as Burgon.  His knowledge and caution served him well here, helping him to avoid most errors.  Duncan probably follows Bloomfield's lead here, although he is well aware of others; he duplicates most of Griesbach's apparatus.

Griesbach / Schott:  These two score remarkably low, but their methods were not yet on a sure scientific footing.  What has saved Griesbach here from many errors has been his own knowledge of scribal errors, and his caution in emending the text (something later editors were notably lacking).  Schott's method is similar.  Both also have respect for the Latin textual stream, which was later regarded with suspicion and abandoned by other Protestant editors.

Scholz:  A popular editor who produced a conservative text, he was satisfied in most cases to note what he considered more significant variants.  Although lacking the most accurate collations, he had a wealth of textual evidence at his disposal, not significantly altered by subsequent discoveries and publications.  His high regard for the Byzantine text-type has assisted in keeping him out of trouble with Alexandrian h.t. errors.  Ironically, this Roman Catholic editor scores the highest, showing that the popularity of his text in England may have been well founded.

The Failures

Tischendorf / Tregelles:  These two largely followed the theories and methods proposed by Lachmann,  especially choosing reliance upon the "oldest evidence".   Unfortunately, they have failed to properly assess both the significance of common scribal habits, and the witness of the majority of MSS, and so have failed to identify Alexandrian h.t. errors. Tischendorf scores higher, in part because of his reliance upon the Latin tradition, which has largely escaped the h.t. errors of the Alexandrian stream.   Tregelles however, with his dogged insistence on only using the oldest MSS, gets bogged down.  Tregelles was aware of the potential for scribal errors, correctly identifying some, but let age of MSS override his caution and judgement here.

Merk:  This Roman Catholic editor depends largely on the work of von Soden.  He does better than Tregelles, but ultimately fails with this aspect of the text, as he also was under the heavy influence of Lachmann and the state of contemporary textual theories.  His reverence for the Latin also helps him to avoid a few of the worst errors here:  Had he trusted the Latin text more, he would have probably got a passing score, even with a poor method and his lack of understanding of scribal habits.

 Hort / Nestle-Aland:  Its no real surprise that Hort, although knowing quite well the problem of homoeoteleuton errors, scores the lowest here.  He elevated Griesbach's  'Canon', "Prefer the shorter reading" (originally heavily limited), into a universal overriding principle.  His (unstated) purpose appears to have been to create the shortest possible text.   The Nestle/Aland text was largely taken over by Aland, and adopted by the German-based UBS group.  The agenda here appears to be to maintain a text distinct from the King James Version at all costs, and the various NA/UBS editions have largely ignored both new evidence and theoretical advances.  The UBS text has been adopted almost universally by translators of modern versions, mostly it seems under the funding and influence of the Roman Catholic church.


* (Many other virtually certain h.t. errors are never noted or included in any apparatus, being unanimously recognized by all textual critics as h.t. errors, if for no other reason than that they are singular readings found only in one manuscript.)

1 comment:

  1. Griesbach / Schott: These two score remarkably low,

    Should be "remarkably high"

    "singular readings found only in one manuscript"

    may as well read "only one manuscript (other than, of course, Vaticanus, Sinaiticus, or Alexandrinus)"