 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
 |
 | |  |
Notes
A version of the computer sections of this article was presented at a conference on "Computer Technology and the Contemporary Scholar—Applications for Research", held at the Graduate Center of The City University of New York in October, 1983. A version of the editorial theory sections was presented at the 1983 meeting of the Medieval Institute of Western Michigan University in Kalamazoo. Much of the information in the article is based on material supplied by Peter P. Farley of Adelphi University, who has been responsible for the computer part of the Hoccleve edition. The article is also dependent on a good deal of very practical advice on both the theory and practice of normalisation offered by scholars in medieval and other fields. These scholars include Fredson Bowers, John Burrow, E. Talbot Donaldson, A. S. G. Edwards, John H. Fisher, W. Speed Hill, David J. Nordloh, M. C. Seymour, and G. Thomas Tanselle. In a series of colloquia, the Hoccleve editors (Charles R. Blyth, Peter P. Farley, Marcia Smith Marzec, Jerome Mitchell, Gale Sigal, and David Yerkes) have made valuable contributions to both the evidence and the theoretical arguments of the article.
G. Thomas Tanselle, "Classical, Biblical, and Medieval Textual Criticism and Modern Editing," Studies in Bibliography, 36 (1983), 21-68. See esp. p. 24, "What has traditionally been called 'textual criticism'—or, more recently, 'textual analysis'—is this attempt to fix the relationship of the surviving documentary witnesses; and though many of the theories of textual criticism have entailed certain assumptions about how the editor's critical text should be constructed, the focus of attention has normally been not on the 'editorial' phase (the actual selection or emendation of readings) but on the prior analysis of the texts that results in the assignment of relationships among them." See also Tanselle's criticism of Dearing's theories of 'textual analysis,' pp. 31-35.
Paul Maas, Textual Criticism, trans. Barbara Flower (1958). See esp. p. 5 for Maas' "ideal" stemma, demonstrating graphically that the evidence of "separative" or "conjunctive" errors would not enable the editor to move further up the genealogy (towards the author's putative fair copy "x") than the stage represented by "α", the archetype created by the comparative data embedded in the hyparchetypes, themselves dependent on the identification of error of one type or another in the extant witnesses.
Piers Plowman: The B Version . . . . An Edition in the Form of Trinity College MS B. 15. 17 . . . ed. George Kane and E. Talbot Donaldson (1975). See esp. Section III (pp. 70-97) for a discussion of archetypal corruption in the B text, a corruption which is, of course, quite predictable according to the orthodox interpretation of Maas' genealogical system.
As has been widely observed (by Kane, Donaldson and their reviewers), the full apparatus of historical collation in the Athlone editions of both the A and B texts is intended in part to allow precisely this re-editing of the text by the critical reader. That is, the degree of license assumed by the editors requires that all of the evidence for reconstruction of the text be made available so that the editorial rationale can be continually tested against the documentary witnesses.
The only possible "authority" in the Regement manuscripts is in the so-called "patronic" versions, the copying of which is very likely to have been overseen by Hoccleve himself. In a forthcoming article on the genealogy of the Regement manuscripts, Marcia Smith Marzec argues that, both on stemmatic evidence and from such extra-textual considerations as the survival of dedicatory verses and the documented patronage of Hoccleve's other works by members of the royal circle, there were probably as many as eight such authoritative copies of the Regement, of which two survive (British Library MS. Arundel 38 and British Library MS. Harley 4866). The second of these two was selected by F. J. Furnivall as copy-text for his 1897 edition of the Regement for the Early English Text Society (Extra Series 72), on no better grounds than its containing "the best portrait of Chaucer" (p. xvii). Furnivall even admits that Harley 4866 displays certain Northern orthographic features which could not represent Hoccleve's London usage, thereby severely qualifying Harley's potential status as "best-text" on the basis of dialect or accidentals—the usual major rationale cited by mediaeval editors (along with the evidence for genealogical authority) in support of a single witness embodying the greatest degree of auctorial intention. The present edition's selection of Arundel 38 as copy-text is supported in part by its genealogical position (it is arguably only one, or at most, two removes from the putative fair copy, and might even be the exemplar, or at least collateral to the exemplar, of Harley 4866), and in part by its orthographic and dialectal features. As is noted in more detail below, the degree of editorial intervention required in the restoration of auctorial "accidence" (see footnote 12) is comparatively slight in the case of Arundel (and certainly much less so than with Harley), and while it might be theoretically possible to restore the substantives of hyparchetypes in other branches of the stemma, these inferred witnesses would not only be genealogically "inferior" to Arundel (i.e., at two or three removes from the fair copy), but would also lack any consistent evidence for the reconstruction of auctorial accidentals, since they would be at best conflated states of the text representing the variants in as many as nineteen or twenty different manuscripts. On both Lachmannian and Gregian terms, therefore, Arundel 38 would stand in a greater position of authority as copy-text than any other extant witness, even if we were not attempting the reconstruction of auctorial accidentals which is the subject of this article. Once normalisation is to be undertaken, the usefulness of Arundel as a vehicle for this procedure becomes unassailable. See, in addition to the Marzec article mentioned above, M. C. Seymour, "The Manuscripts of Hoccleve's Regiment of Princes," Transactions of the Edinburgh Bibliographical Society, iv, part 7 (1974), 255-297 for a description of Arundel, Harley, and the other manuscripts, together with a preliminary account of their filiation. See also A. S. G. Edwards, "Hoccleve's Regiment of Princes: A Further Manuscript," Transactions of the Edinburgh Bibliographical Society, v, part 1 (1978), 32, and R. F. Green, "Notes on Some Manuscripts of Hoccleve's Regiment of Princes," The British Library Journal, 4, number 1 (Spring 1978), 37-41. Seymour gives a brief account (differing in some minor points from the evidence cited here) in the Commentary (esp. pp. 113-114) to his Selections from Hoccleve (1981).
The fullest account of the status and possible influence of Chancery English occurs in John H. Fisher's article, "Chancery and the Emergence of Standard Written English in the Fifteenth Century," Speculum, 52 (1977), 870-99. The degree to which Hoccleve's idiolect corresponds to the most characteristic features of Chancery English as identified by Fisher is now being investigated by the Hoccleve editors. An interim analysis suggests that, while Hoccleve, as a Londoner, inevitably shares several of the paradigmatic forms which Fisher recognises as belonging to the "proto-Chancery" that would have been in use in the early decades of the fifteenth century, the degree and type of consistency in the concordance generated from the Chancery documents is often of a different nature from that discovered in our analysis of Hoccleve's practice. A fuller report will be published shortly.
For surveys of recent work on the autobiographical element in Hoccleve's verse, see Jerome Mitchell's chapter (pp. 49-63) on Hoccleve in Fifteenth-Century Studies, ed. R. F. Yeager (1984), and D. C. Greetham, "Hoccleve's Development of the Chaucerian Persona," in Chaucer at Albany II: Essays in Honor of R. H. Robbins (forthcoming).
Huntington Library MS. HM 111, Huntington Library MS. HM 744, Durham MS. Cosin V. III. 9, and the "non-literary" (Privy Seal) British Library MS. Add. 24062, which (unfortunately) is almost all in Latin or French. See H. C. Schulz, "Thomas Hoccleve, Scribe," Speculum, 12 (1937), pp. 71-81.
As Jerome Mitchell (Thomas Hoccleve: A Study in Early Fifteenth-Century English Poetic [1968], p. 14n.) remarks, "Hoccleve is the only significant Middle English writer whose works have in large part been preserved in autograph manuscripts." The only other substantial piece of Middle English verse available in an orthography which seems to represent auctorial practice is the notoriously idiosyncratic Ormulum; in Hoccleve's general period, only the possibly holograph text of Chaucer's Equatorie of the Planetis offers a potential parallel to the Hoccleve documents, and the Equatorie is uncharacteristic of Chaucer's oeuvre, being a scientific work in prose. In other mediaeval literatures the situation is a little better, for there are holographs of, e.g., Petrarch and Boccaccio, and William Wallace is to edit the autograph version of Thomas Aquinas' scriptum on the third book of the Sentences of Peter Lombard for the Leonine series. In general, however, it must be accepted that the survival of the Hoccleve holographs is an aberration in textual transmission, and does not in itself suggest that there are great documentary riches yet to be discovered. No doubt if Hoccleve had not been a Privy Seal scribe, his orthography would be as lost to us as that of any of his fellow-poets.
Jerome J. McGann, in his recent A Critique of Modern Textual Criticism (1983), attempts at several points (see esp. pp. 15-36) to draw the Greg-Bowers and Lachmannian dispensations together, particularly with regard to auctorial intention. However, his case depends very largely on proving that post-Greg textual critics have (deliberately?) misread or enlarged upon Greg and does not, I believe, modify the general assumptions about base-text/copy-text which motivate this present article: namely, that the Lachmannian genealogical system has been traditionally seen as a device for identifying the position of a witness in a hierarchy developed from a study of the substantive errors in texts (with comparatively little concern for the accidentals as they might be represented genealogically), whereas the Greg doctrine on copy-text has separated the authority of substantives and accidentals, while regarding both features as a necessary element in the reconstruction of auctorial intention, be it final (for substantives) or original (for accidentals). This is obviously something of a simplification of the distinction between Greg and Lachmann, but that some such distinction is eminent and valid is surely indisputable—at least from a charting of editorial theory in both the "ancient" and "modern" fields.
That is, where a normalisation theory occurs in mediaeval studies, it is associated primarily with the resuscitation of, for example, dialect, not idiolect. Such a normalisation theory might surface in the smoothing of Old English to tenth-century Wessex forms or in the selection of a South-East Midlands manuscript as copy-text on its dialectal features, but since there is virtually no opportunity for the reconstruction of an auctorial idiolect, there is little development of the sort of distinctions (i.e., between substantives and accidentals) which have led to the evolution and dissemination of Greg's theory elsewhere.
I use the term "accidence" in a special sense here and elsewhere in this article— i.e. to refer not to the classification of grammatical forms embodied in an author's "style", but to the entire system of surface features of which grammatical alternatives are but one of the choices presented to an author within the parameters of a particular language at a particular time. Thus, a choice between long and short "s" might be regarded as a purely orthographic component in the system of accidence (although there is some argument as to its possible phonological significance), whereas a choice between, say, "-e" and "-eth" as 3ps present tense inflection, or a predilection for a past participle with the "y-" prefix might have grammatical import (though not "meaning") as well as orthographic. Both types can still, however, be regarded as genuine "surface features" and both are therefore a part of the "accidentals" from which an identifiable system of "accidence" can be derived.
Tanselle, op. cit. passim, e.g., p. 42, "The editor is finally responsible for establishing both substantives and accidentals, and to assume that scribal accidentals are too far removed from the author's practice to be worth preserving is to ignore the connections between accidentals and meaning. One may not in the end accept these accidentals, but the question of what accidentals to include in a critical text must be faced. Authorial accidentals in ancient texts may be more conjectural than in modern texts, but the attempt to approximate them is not necessarily to be rejected in favor of standardized spelling and modernized punctuation." Tanselle's basic argument is stated in his last two sentences (p. 68), "Editing ancient texts and editing modern ones are not simply related fields; they are essentially the same field. The differences between them are in details; the similarities are in fundamentals." It is, of course, this belief which has motivated the Hoccleve edition to make use of both ancient and modern textual practices in the editing of the Regement.
See Tanselle op. cit., p. 48 and fn. 33 for an account of the argument between Greg and Maas over copy-text. See A. E. Housman, Selected Prose, ed. John Carter (1961), p. 35-36 on the concept of "best" MSS, p. 59 on "the authority of a MS. against the exercise of the judgment," and p. 141 on "sincere" MSS.
Bowers' views on the Hoccleve problem are, of course, similar to those expressed in his published writings. See, for example, the "doctrine" that a "definitive edition . . . in its texture of accidentals, as well as in its words, [should] conform to the closest approximation to the author's own linguistic and orthographic characteristics that can be recovered." (Textual and Literary Criticism [1959], p. 141).
The well-attested forms in HOCCLEX parallel exactly the same forms in the copy-text. In practice, this level accounts for, perhaps, 50% of all cases in the normalisation procedures, thereby confirming Arundel's status as copy-text on Gregian criteria as well as its function as base text on "classical" principles.
While copy-text might occasionally use a variant form, it clearly prefers the same form that is consistently used in HOCCLEX.
Copy-text does show the form used in HOCCLEX, but as an "indifferent" variant, in comparison to the consistent usage in HOCCLEX.
While copy-text clearly prefers a form other than that in HOCCLEX, the holograph usage is still absolutely invariable.
Even where the form does not exist in copy-text, the holograph preference is unambiguous. Usually, of course, the HOCCLEX form would therefore not occur in the edited text at all, except perhaps in circumstances where the scribal inflection seemed an error (e.g., in number, agreement etc.), when the form derived from HOCCLEX would take precedence.
Note that the required "level" of evidentiary support in HOCCLEX is higher than in the corresponding entry for copy-text (line 2 of model). This reflects the editorial desire to represent auctorial (as opposed to scribal) choice with greater surety, and, in general, not to admit HOCCLEX (or NORMLEX) forms as paradigms unless there is either a clear "blank entry" in the copy-text data or a statistically and morphologically well-attested norm in the great majority of auctorial or auctorially-derived forms.
Given the general assumptions (regarding auctorial and scribal usage) under which the normalisation model is constructed, this level should be very rare (and in fact turns out to be so), but could contain occasional anomalies. If one were using orthodox base-text theory (or copy-text theory in its usual restriction to the features of a single text), then the 100% usage form in copy-text would clearly support an unambiguous preference, albeit scribal, or at least not demonstrably auctorial. If the copy-text form is therefore highly atypical as measured against the HOCCLEX extant patterns (say, less than a 25% concurrence with HOCCLEX), then great care would obviously be needed in establishing preferred auctorial usage. Thus, although this level, particularly towards the lower statistical end just cited, occurs infrequently in the normalisation (perhaps once every 500 lines), it is by far the most problematical. Note, furthermore, that we had originally employed the same series of discriminations for HOCCLEX (high, indifferent, low) in this category as in levels 2-4 and 9-11 for copy-text. However. given our general principle that a demonstrable consistency of usage in HOCCLEX is necessary for normalisation of copy-text to proceed, anything under a 90% usage (particularly as measured in this category against a 100% usage in copy-text) seemed possibly suspect and might therefore lead to a selection of copy-text over HOCCLEX, as the model indicates. Therefore, a single category representing this level of indeterminacy, rare though it be, is a more honest response to the statistical conditions than a three-tier break-down would be.
While the required form does not exist in the holographs, it would be easy to create it by morphological extension. This required form corresponds exactly with the form in copy-text, and would merely confirm a conservative reading of copy-text by reference to NORMLEX.
Copy-text preference would again confirm the form created by morphological extension in NORMLEX. Since this form is at least derivable from auctorial usage in HOCCLEX and supported by the majority readings in copy-text, its consistency would be employed throughout the normalisation of the copy-text variants.
An indifferent variant in copy-text, but with the NORMLEX form established as a consistent and unambiguous derivative from HOCCLEX.
A logical extension of the previous two levels, where despite the statistical evidence of the copy-text, a minority form in the copy-text would be preferred since it would be derivable unambiguously from NORMLEX (and therefore from HOCCLEX). Note that the "presumptive authority" of copy-text would be employed, not in cases like levels 9-11 (where the authority of the consistent external evidence— i.e., NORMLEX—is paramount), but in cases where the "indifference" of the copy-text scribe would be matched by a similar (in incidence, if not always in exact type) indifference in HOCCLEX.
The normalised form is not extant in the copy-text (the scribe presumably preferring other variants of the same basic word) or in HOCCLEX (presumably because of the lack of concurrence of the two lexicons). The form could, however, be fully supported by NORMLEX in evidence drawn from related forms in HOCCLEX. This is therefore a level which logically proceeds quite straightforwardly from the previous three, where the normalised form was gradually having its statistical base in copy-text reduced. Note that, while the statistical base might appear to have vanished, in fact this could conceal a very high level of authority for morphological extension. As suggested above, this level could describe cases where a particular word happens not to occur in the holographs (and therefore does not turn up in the HOCCLEX column), but does occur in the copy-text, in, however, a form recognisably different from the clear auctorial preference that could be established in NORMLEX if the word had been used in the holographs. Say, for example, Hoccleve had happened not to use the word (ModE) though in the holographs, but it did show in the copy-text. We can establish from HOCCLEX that a) he never used a 'yogh' and that b) he never used a 'þ' except in the contracted form þt=þat, (ModE that), and that therefore a copy-text form of þouyogh would be an impossibility in HOCCLEX. The normalised form, based on 100% usage patterns in HOCCLEX, would be perhaps though or thogh. Thus the form (if it represented a choice between an initial 'þ' and 'th', or a final 'yogh' and 'gh') could merit the 100% level in NORMLEX, while not occurring in HOCCLEX or copy-text in the form created by NORMLEX and used in the edited text.
This would cover very few cases, and would be most prominent in nonce-words, often proper nouns, which for obvious reasons would often not be replicated in HOCCLEX. The 100% in the copy-text column, therefore, would often represent in actual incidence a very small number, usually a single occurrence—for the scribe of Arundel, while closer to Hoccleve's usage than the other copyists, does not share his author's passion for orthodoxy in orthography to the same degree.
Again, a very rare level, where NORMLEX would not be capable of creating a consistent reliable form by morphological extension, but, on the contrary, several forms which would mirror the apparent indifference in HOCCLEX. In such cases, we would invoke the principle of copy-text as residual authority, as long as the forms showing in copy-text reflected putative auctorial choices. If the type, but not ratio (or incidence) of indifferent forms in copy-text were dissimilar to those in HOCCLEX and NORMLEX, then presumably the use of these latter forms would have to be considered. However, since this does not seem to appear in the editing of the Regement, the case is included here only as a logical completion of the statistical types cited, and an additional testament to the conservative principles upon which the normalisation proceeds.
See Robert L. Oakman, Computer Methods of Literary Research (rev. 1984), pp. 70-75 for an account of the KWIC and KWOC formats.
It should be emphasised that this reverse lexicon, even though it is used for charting suffix forms, is different in format and purpose from the sort of suffix-removal or lemmatisation programs described, for example, in Susan Hockey's A Guide to Computer Applications in the Humanities (1980), pp. 101-106. See also Oakman, pp. 80-81, 92-96 for references to other types of automatic lemmatisation and normalisation, including BIBCON, which can alphabetise variant spellings in Middle English under the appropriate modern form. See Oakman pp. 76-77 for a brief account of genuine reverse lexicons (or unfortunate lack of them) in other fields.
The final "-e" in wole could, of course, have metrical, as opposed to purely orthographic value. Judith Jefferson of the University of Bristol is making a metrical analysis of the Hoccleve holographs (dealing with such questions as the use of pronounced final "-e" and of variant spellings for metrical purposes). The results of this study will obviously be of great help in establishing the degree to which copy-text use of final "-e" represents a likely auctorial preference and could therefore have significant impact upon the metrical features of the Hoccleve edition.
The 7 in the text here is one of the numerical codes used to represent various scribal features not available on the conventional keyboard. The codes occurring here include 1 =semiinv. 2 = long "s" ("longs"); 3 = "re"; 4 = the supralinear flourish indicating an omitted "r"; 5 = "þ"; 6 = the macron signalling a following nasal; and 7 = the final flourish transcribed as "-e". (* = a supralinear "t"). Thus, "5 *" = þt .
| | |