This page is about what is usually called "text criticism" and some of the subtleties that apply to the rise and acquaintance and corruption of texts in general.,
In philology in any language, it is useful to distinguish between a text's authorial growth process (the different stages, if any, being called T1, T2, etc) and its later scribal corruption process (represented by manuscripts M1, M2, etc). The two are separated by a point P at which the text comes to be copied by scribes for wider distribution.
Authorial Phase (text formation, T1 > T2 > T3) . . . . P . . . . . Scribal phase (text corruption, manuscripts M1, M2, M3)
The Authorial Process
When an author (or a succession of authors or proprietors) successively modify a text, each version is replaced by the one following, and ceases to exist as a separate entity. This is what the arrows in the above model are meant to represent: T1 > T2. Once Point P is passed, there come to exist several physical manuscripts, each containing the final authorial text. Their parallel existence is represented by the commas in the above model: M1, M2. Once multiple manuscripts exist, there is the danger that one or more of them may diverge from P, the author's text as it was first made available for copying. The manuscripts must thus be compared, and any differences noticed. Where differences exist, one of them is presumably anterior to the other(s). Which one? This involves a directionality determination, to distinguish the original reading from the errors or improvements that were made by the other scribes. How does one detect the original reading? In this question lies the whole art of text criticism. The basic guideline was articulated by Tischendorf in 1849. Here is a summary of that statement by Epp (Perspectives 147):
More probable than others is the reading that appears to have occasioned the other readings, or that still contains within itself elements of the other readings. Taken broadly, says Tischendorf, this is the foundation of all rules.
This principle had been articulated before, but Tischendorf was the first to identify it as the key to the whole business. Without that guiding principle, the adjudication of manuscript variants quickly reduces to mere subjective preference, or to rules of thumb, such as a preference for the shorter reading, or the reading of a particular manuscript, or whatever. That unthinking reduction was brilliantly satirized by Housman, and all who venture into this sort of work need to be warned by reading Housman. The sad fact is that variants cannot be adjudicated without thinking. They require the same kind of thought that must be given to understanding the formation of the text in its authorial phase.
We are not always lucky, and it can happen that none of the extant manuscripts contains the original reading, but the original reading can sometimes be deduced, on the Tischendorf principle and some other guidelines (certain scribal errors and confusions are very likely to happen) from the nature of the various readings.
After working through the variants, always with an eye on that part of the text that is not put in doubt by the existence of variants, or the whole that seems to be emerging from the adjudication process itself, we arrive (in favorable cases) at the archetype (A): the master copy from which all later copies can be thought to be descended. This is sometimes called the "author's original," but this is a delusion; it is merely the best text that study of the available manuscripts allows us to reconstruct. If we rewrite the above model to include A, we would have
Authorial Phase (text formation, T1, T2, T3) . . . . P . . . . . A . . . Scribal phase (text corruption, manuscripts M1, M2, M3)
A, the thing to which we get back by adjudicating the M1, M2 variants, may be far removed from the author's original. Thus Lachmann, reconstructing the NT from the Greek manuscripts available to him, estimated that his archetype A was the text of the 4th century, not that of the 1st century. Later scholarship has confirmed that Lachmann was about right in that estimate. In the case of a late forged text, there is not, properly speaking, any author's original at all: the supposed author may be altogether fictitious. In such cases, we have an A, but no P, and no authorial process; no T1, T2 etc.
Some traditions are entirely fictional; there is nothing behind them. Those who find this shocking need to familiarize themselves with the kind of invention and legendary exaggeration that actually happens in the evolution of traditions. The older manuals of historical method included long sections on just these things. We may particularly recommend Langlois et Seignobos (1898, tr 1908) and John Martin Vincent (1911); there are also good things in Marc Bloch (1949; tr 1953). The ground rules for all historical inquiry were articulated, in suitably gnomic form, by von Ranke, from 1814 on.
Lower and Higher Criticism
Suppose two manuscripts have the readings ABC and AC, where one has, and the other lacks, a portion B, We consider the nature of the material. If B seems inconsistent with its surroundings, A and C, and if when B is removed from that series, A and C join together to form a satisfactory and consecutive text, then we will judge that the manuscript with the reading AC is original, and that a scribal interpolation has taken place in the other manuscript. That is, our analysis looks like this:
A B C
Or, if the manuscript with AC is choppy or discontinuous, and if the one with ABC reads satisfactorily, then we will judge that one scribe has mistakenly omitted the B segment, and that the one with the ABC reading is the original. In that case, our conclusion looks like this:
A B C
That is how the lower criticism works.
Now suppose we are through with manuscript adjudications, and we have the archetype, the earliest state of the text that can be reached with the lower criticism. The archetype may be (and with some Latin authors, it is) centuries later than the author's work. Plus as we know Horace (for one) and Thucydides (for another) did, the author himself may have modified his work, leading sometimes to inconsistencies or formal interruptions. This is where the higher criticism comes in.
In working with the archetype, we encounter a passage ABC in which there seems to be a problem of continuity, one which is solved by removing B. In that case, as in the above one, we will judge that B is an interpolation (whether an ill-advised one by the author at a later period, or by a scribe whose copy we no longer possess). That is, we reach this decision:
A B C
The process of adetermination is exactly the same in both cases. The only difference is that, in the latter case, we had no manuscript differences to call the passage to our attention. We had to notice it ourselves, just in the course of careful reading.
That is, the lower criticism and the higher criticism are the same thing, except that the lower critic has variants which bang on his window, getting his attention. The higher critic works in silence.
Subtleties of Acquaintance
There is a difference between a text that has been handed over to copyists, and one that has not. But even when it has not, there may be shades of publicness. An authorial or other closely held text, even before it reaches Point P and becomes generally available, may not be entirely private. Without being handed over to copyists, it may have a penumbra of outside acquaintance. Consider these simultaneous examples:
- Vergil read parts of his unfinished Aeneid to friends; such literary evenings were common in Roman patronage circles.
- The classical Chinese texts, though closely held by the respective schools, came to be mutually known through personal contact.
- Luke's Gospel was probably known not only in his own church, but among neighboring churches, one of which may have been Matthew's. Both were in Antioch, or that is the consensus of modern opinion, so the suggestion is far from fantastic; it is instead the likeliest way in which the two could have become known to each other. Antioch is a big city, and the slummier parts of town, where Luke's people had their storefront church, probably did not adjoin Matthew's ritzier community. Nevertheless, Matthew, his curiosity aroused by rumor, could probably have walked the whole distance without working up a sweat. We do not have to hypothesize a taxi service to account for this journey. We do not have to account for anything. That kind of contact is intrinsically likely.
There are also some variations in the ways in which successive texts in a series can be known to each other - how far up the line he awareness of a late text may reach. Consider these shorthand representations:
- Mk > Mt > Lk implies that each text in the sequence is aware only of the immediately preceding one
- Mk >> Mt >> Lk implies that each text in the sequence is aware of all the preceding ones
Finally, "public" has a different meaning in an open culture and in an oppressive culture. The "silence" of the evidence may have different meanings in those different situations. Here is a short summary of some of the possibilities, in both circumstances:
These situations vary with local conditions, including cultural specifics, but should be borne in mind, and the reader should be alert for signs of coded or reticent statement in a text.
We will now look at the corruption process a little more closely.
The Corruption Process
There are two types of scribal changes: the inadvertent and the intentional. Inadvertent errors involve such things as mistaking a word for one written similarly, incorporating into the text what was meant as a marginal gloss, returning to the text which lay before the copyist (the German term for that text is Vorlage) a line above or below the correct place, producing either a repetition or an omission, misconstruing scribal abbreviations, and so on. Intentional changes are made to improve the text for its next reader, by harmonizing it with other texts (the text of Mark is sometimes in this way assimilated to the more highly regarded Matthew), or making it more respectful (adjectives registering the divinity or majesty of Jesus tend to pile up as time goes on), or removing embarrassments (the lowly pallet on which the paralyzed man is brought to Jesus in Mark, or Luke's report that Jesus was crucified between two other criminals), or simply making things easier for the reader. In principle, text criticism should be carried out before the text is analyzed on its own terms, or used as evidence for history. In practice, a certain amount of circularity is necessary. For instance, one valid criterion of genuineness is whether a given variant in a passage of Mark is typical of the style of Mark. But what do we actually know of "the style of Mark" before we have arrived at the text of Mark? Handling such circularities is one of the things that text critics strive to be good at. As Housman has said, text criticism is not a mechanical process. It requires experience, judgement, and a sense of reality. It requires thought.
With this distinction firmly in mind, the standard handbooks may fruitfully be consulted. Probably the best for the beginning NT student is that of Metzger. We have already had one statement of the Tischendorf principle. Here is Metzger's:
"Perhaps the most basic criterion for the evaluation of variant readings is the simple maxim 'choose the reading which best explains the origin of the others.' We all follow this common-sense criterion when confronted with errors and "variant readings" in modern books . . . " (p207).
Just so. Viewers may note that their automatic mental correction of a misprint in the morning newspaper is not based on the comparison of different manuscript readings. In this case, there can be no different manuscript readings: it is the nature of the printing process that all copies of that newspaper will have the same error. The correction of such an error is based on recovering an error process for which we have only the single erroneous text as evidence. In such cases we see that the Tischendorf Principle still applies: the probable original can be deduced by reflection and experience from the single, unassisted erroneous morning newspaper. That principle is not limited to the adjudication of multiple variants. This is an important theoretical extension of the usual "multiple manuscript" situation. A single text, properly handled, can also be evidence for the process which lies behind it. This is what we have above called the"higher criticism."
Corrections made from a single text, rather than from the comparison of two texts, are sometimes called "conjectural emendations." Some regard them as unsound in principle, but this is not correct; they are no more or less conjectural than a choice between two manuscript variants (a choice which can also be made inexpertly). It must in the end be said, of the correction of errors whether from one or multiple versions of a passage, that some people are better at it that others. Bentley and Madvig were among those who have been proved to be uncommonly good at it.
It helps to have a wide experience of the language in question, and a close acquaintance with texts similar in type to the one under study. These things are recommended herewith, along with a close rereading of Housman.
The now standard critical text of the NT is that of Westcott and Hort. For their reasoning, see their companion work, the Introduction (1882), and particularly the included Notes on Selected Readings. The WH text has been further revised in later years, though not always wisely. The successive United Bible Societies (UBS) editions, in particular, besides taking account of new and valuable manuscript evidence, also reflect the loss of critical acumen which affected all branches of textual science over the course of the 20th century. As one result of that loss, some beloved but ill-attested passages rightly excluded by WH were restored to the UBS text. For the best recent opinion, see again a work of Metzger, his Textual Commentary to UBS 3 (1971) and UBS 4 (1994). In our considered view, Metzger's dissents from the UBS committee's majority opinions are nearly always right, and should be followed (against that text) by the discerning student.
Some Specific Passages
In the rest of this section, we notice certain high-profile passages in various NT texts, chiefly in the Gospels. Whether they are original in their respective texts, or were later added to them, has been much debated. We give our own best judgement.
- Some Important Text Critical Results
- "Nine Western Non-Interpolations" in Matthew and Luke
- Eighteen Further Western Non-Interpolations
- Eight Additional Suggestions by Wieland Willker
- Other cases where Bezae has the better reading
- The Fitzmyer Exclusions in Luke
- The Pericope Adulterae in Luke and John
- The Doxology in Romans
- The "Lord of the Sabbath" in Mark
Differences between manuscript readings define a problem, one or more parts of which are later changes. Getting rid of the nonoriginal variants is how we get rid of the later history (the corruption history) of the text, and get back, as near as we can, to the original text; that "nearest point" is properly called the archetype. The literature on methods and specific solutions in this area is vast, and we have no intention of reviewing it, beyond the above list of cases which are of special importance for this overview of early Christian texts. Otherwise, we will generally accept the verdict of decades of scholarly effort, beginning with Benson and continuing through Lachmann, and will regard the Westcott/Hort critical text, itself chiefly based on Codex Vaticanus (4th century) as our base text; the thing we have to talk about.. Otherwise, we end our glimpse of the "lower criticism" at this point.
The remaining pages in this section are chiefly concerned with what is properly called the "higher criticism" - the study of prepublication growth processes in the texts.
25 Aug 2010 / Contact The Project / Exit to Alpha Index Page