An entry by Kamran Karimullah for the forthcoming Translating Cultures Glossary
Image. The ‘critical apparatus’ in Maximilian Wallies’ edition of Alexander of Aphrodisias’ commentary on Aristotle’s Topics (Berlin: G. Reimeri, 1883).
The critical edition is one of the most important document types represented in the Latin, Greek and Arabic subcorpora that have been built by the Genealogies of Knowledge research team. Every text in these ‘premodern’ subcorpora is based on a critical edition of a lost original. Yet, the process of editing premodern texts in a critical edition is not a straightforward, objective, and empirically verifiable process. Interpretation is involved in the editing process as well. Critical editions are, thus, important to Genealogies not only as sources of text, but also as forms of translated text, broadly conceived.
Though text-critical techniques involving emending texts and reporting variants have existed since antiquity, the critical edition in its contemporary garb springs from philological interests among renaissance Humanists to reconstruct the Greek and Latin secular classics and the Greek New Testament. Textual critics distinguish between the author’s original text (the ‘autograph’) — a somewhat fictive object in not a few cases — and the manuscript to which all subsequent manuscript witnesses of the text can be genealogically traced (the ‘archetype’). Assuming that the archetype is, in the words of Paul Maas, ‘closer to the original than the text of any witness’, textual critics attempt to recover the autograph as far as possible by reconstructing the archetype. Like the autograph, the archetype no longer exists. Its existence, however, is hypothetical rather than fictive. Textual critics insist that the recovery of the archetype should proceed from understanding the genealogical relations that link the extant manuscripts to the archetype, and by setting down criteria that allowed editors to mechanically determine which readings go back to the archetype, nineteenth-century editors claimed they could scientifically reconstruct the archetype — and by extension, the autograph — of any text.
Roughly speaking, then, we think of the critical edition as a published version of a text — often an ancient text — that is reconstructed on the basis of surviving witnesses of the author’s original, which is normally lost. Frequently, however, there is, in fact, no author to ascribe the text to — there are critical editions of Hippocrates’ works, which represents the work of many different hands. It is not necessary that the original be a written text — many argue that Homer’s Odyssey was probably not written down until centuries after Homer’s death. Normally, the manuscripts are not direct witnesses of a lost autograph or archetype — what we know today as the collected works of Aristotle is in large part thanks to the herculean efforts of Andronicus of Rhodes, who died some 300 years after Aristotle.
In other words, our working definition oversimplifies. But, in a way, it could not be otherwise. The positivist’s faith that there could be an objective scientific method for mechanically reconstructing texts, a method that bypasses interpretation, foundered on postmodern shoals. Conjuring up Aristotle’s Poetics out of a messy textual reality, raising its spectre before the reader in a volume bearing Aristotle’s name, summoning it from manuscript folios, the text editor’s and the magician’s craft are akin. In the words of E.B. Tylor’s Researches into the History of Mankind (1865): ‘any association of ideas in a man’s mind, the vaguest similarity of form or position, even a mere coincidence in time, is sufficient to enable the magician to work from association in his own mind, to association in the material world’. By nineteenth-century definitions, then, the critical edition works magic on the humanist’s mind. Yet, this charm is what makes critical editions indispensable to humanist research; they are certainly essential to the digital humanist.
Computational techniques used by corpus linguists further reify the critically edited texts included in the Genealogies corpus. The ‘physical’ attributes of the digitised text (its ‘metadata’) are encoded in a computer-readable syntax. These metadata constitute the texts’ tenuous link to the concrete reality from which they sprang. The extensible syntax of XML may allow conscientious researchers to capture the uncertainty, messiness and some of the arbitrariness in producing a critical edition. We might use attributes of the AUTHOR or TRANSLATOR elements to indicate levels of certainty, ranging from ‘certain’ to ‘doubtful’, about authorial responsibility. In the TITLE element, we might note that the original title, if it had one at all, is ‘unknown’ or use one by which the text eventually came to be known much later. In this way, we try to dispel somewhat the critical edition’s charm on the corpus user.
Yet, the move from physical codex to XML-annotated text-file, from book to dataset, involves a number of transformations that affect how the text is experienced. These do not relate to the text itself, but involve omissions that qualify how the reader reads and experiences the text. For example, for texts to conform to the requirements of the concordance software and to limit “token-noise” in the corpus indices, we have systematically omitted the critical apparatus from premodern critical editions. With the record of manuscript variants, additions, omissions, lacunas and editors’ conjectures removed, the premodern texts that make up the Genealogies corpus are endowed with an aura of monolithic objectivity, of textual stability, and with an appearance of permanence and ‘givenness’ that they do not in fact possess.
Revisions of the positivist-inspired text-critical project allow us to view critical editions as forms of interpretation and translation. Yet, as forms of translation, critical editions pose real problems for the Genealogies project. Our Sinclair-inspired theory of language enables us to trace conceptual shifts in the movement from source to target language, using lexical and syntactic patterning as evidence. We, and our software tools, take for granted that translation is interesting, above all, as a semantic activity. This leaves us ill-equipped both conceptually and computationally to examine the forms of ‘translation’ involved in the preparation of critical editions, namely forms of ‘translation’ where the source and target language are identical; where the interpretative element essential to translation activity involves the selection of the so-called correct reading from a number of textual variants; where the editor-cum-translator treats the source text not as a semantically replete object, but as a syntactic-morphological object whose physical structure must be transmitted.
In sum, critical editions are an essential part of the Genealogies corpus. Yet, this document type imposes limits on corpus users, while also suggesting new lines of research. Corpus users must be conscious of the ‘magic’ that critical editions practice on them. Observing differences in semantic prosody between source and target text may tell us a great deal about the author’s and translator’s world-views and how each approaches his craft. But observing differences in collocation frequency in two editions of Aristotle tells us less about Aristotle’s philosophical discourse and more about the editorial habits and interpretive choices of August Bekker and William Ross, Aristotle’s nineteenth- and twentieth-century editors.
 P. Maas, Textual Criticism, trans. B. Flower, Oxford, Oxford University Press, 1958, pp. 2–3.
 E.B. Taylor, Researches into the Early History of Mankind and the Development of Civilization, Boston, Estes & Lauriat, 1878, p. 130.
Grafton, A., G. Most and S. Settis (eds.), The Classical Tradition, Cambridge, Mass., Harvard University Press, 2010.
Timpanaro, S., The Genesis of Lachmann’s Method, trans. G. Most, Chicago, University of Chicago Press, 2005.
West, M., Textual Criticism and Editorial Technique, Berlin, Walter De Gruyter, 1973.