![]() | | ![]() |
A great number of works written in Middle English are in fact translations, in one sense or another, of originals written in French, Latin, Italian, and so on. This has long been recognised by mediaevalists in the voluminous investigation—and frequent publication—of "sources and analogues", where the critic might be able to observe the structural, narrative, linguistic, and aesthetic changes wrought by the translating author. On the textual side, there has been much editorial concern for establishing specific readings (and justifying them by appeal to these sources) in works by major authors. But there has been comparatively little study of the theory of such textual relations; the attention of critics, literary and textual, has been on the particular manifestation of translation in a given author, a given work, or a given reading, and the discussion of translation has therefore tended to be embedded either in editions of Middle English works or in critical examinations of mediaeval authors. For example, although Vinaver attempted to draw general theoretical maps for charting the possible range of scribal errors that could occur in the act of copying,[1] when it came to the editing of his Works of Sir Thomas Malory he was content to produce stemmata that reflected the specific textual conditions which obtained between the Winchester manuscript, the Caxton edition, and the French sources.[2] There was no theoretical study of the possible routes of transmission across the translation gap, but only of the actual (and therefore inevitably limited) translation process as used by his author—and this is perhaps only right when an editor is concerned to defend the individual emendations introduced into his text. Where arguments over translation have occurred, as for example in David C. Fowler's attack on George Kane and E. Talbot Donaldson's smoothing of the B-Text of Piers Plowman to conform with the A, they have
![Click to Enlarge Page 132](https://iiif.lib.virginia.edu/iiif/uva-lib:911568/full/!200,200/0/default.jpg)
In fact, there are theoretical positions underlying the
work
of Vinaver on Malory, Kane-Donaldson and Fowler on Langland. For
Vinaver, it might be appropriate to suggest that his models for translation
all rest upon the notion of fidelity, so that where alternatives in the
Winchester and Caxton versions occur, the test is an appeal to the French,
on the assumption that this will yield Malory's likely reading.[4] For Kane-Donaldson, the argument
What is currently missing from, or undeveloped in, the theory of editing in Middle English is a model or models which can be employed in evaluating the conditions to be found in the translated and translating languages and texts.[6] It may also be that similar models could be useful in other periods
![Click to Enlarge Page 133](https://iiif.lib.virginia.edu/iiif/uva-lib:911569/full/!200,200/0/default.jpg)
I propose to begin this process of creating models for stemmatic analysis of translation by citing textual examples from the work of one of the most prolific and influential of Middle English translators, John Trevisa, a translator who moreover left considerable discussion of the nature and mechanics of translation.[8] I therefore select Trevisa in part because his known (and avowed) position and practice on procedures for translation make him an ideal author for such a "control" necessary for the preliminary charting I suggest, and in part because his actual work appears to be governed by what we may call the "clere and playne" doctrine, involving a self-effacement of the translator which reduces the potential creativity (or idiosyncracy) of the derived text to a manageable or discoverable level, despite the lack of holographs
![Click to Enlarge Page 134](https://iiif.lib.virginia.edu/iiif/uva-lib:911570/full/!200,200/0/default.jpg)
I posit three theoretical classes, generating four ideal models between the translated and the translating languages:
—the first class, what we may call the Perfect Linear Class (MODEL A), with a consistent source and a consistent derived text.
—the second class, with two types of the Imperfect Linear Class (MODELS B and C), the first with a translating language variance (i.e. a consistent source with an inconsistent derived text), and the second with a translated language variance (i.e. an inconsistent source with a consistent derived text).
—the third class, or Parallel Variance Class (MODEL D), with an inconsistent source and an inconsistent derived text, clearly the most problematical model of the series. Finally, I consider the question of determined or auctorially conscious variation (as opposed to the postulated scribal variants charted in MODELS A-D), discussed under the two possible directions such determination can take: the gloss-type (or divergence from the norm of the source, usually in the form of an addition), and the neologism-type (or divergence from the norm of the current medium, usually in the form of a substitution).
![]() | | ![]() |