4. A Protocol for Scholarly Digital Editions? The Italian Point of View

Marina Buzzoni

Preliminary remarks

This chapter discusses whether it is desirable to establish a protocol that would provide, if not a standard, at least some guidance on how to structure the core elements that one should expect to find in a scholarly electronic edition.

A preliminary examination is thus needed to determine which features should be defined as fundamental. Though the debate on the issue is still intense, many scholars in the field of digital philology1 now agree that there are at least five domains in which scholarly digital editions may offer important advantages over paper editions, namely:2

1. the possibility to present and manage quantities of data that are not normally publishable in a paper book;
2. the relationability of the data provided, i.e. the possibility of making connections between data and processing them at a speed, precision and complexity otherwise unattainable;
3. their interoperability, i.e. broadly speaking, the ability to share information in computing environments and—in principle—between different computer systems, thus enhancing the possibility of interaction within the scientific community in time and extension which a traditional book does not allow for;3
4. their multimediality and multimodality, which allow for the organisation of data into hierarchically structured hypertexts, as well as the inclusion of non-textual data in the edition (e.g. audio and video files);
5. and, last but not least, user interaction.

More specifically, these domains represent what can be considered the added value of scholarly digital editions, not in terms of mere application but rather in terms of theoretical and/or methodological improvement. For example, the availability of space offered by the digital edition, together with the relationability of the data provided, are prerequisites which allow users to account for the choices made by the editor more easily and economically than in printed form;4 and accountability is a necessary component of scientific reliability. Paradoxically, the ‘new’ digital medium goes in the direction indicated by thoroughly traditional philologists like the Italianist Domenico De Robertis, according to whom an edition can be called critical in a strict sense only if it furnishes the reader with all the documentation necessary to evaluate it and to produce another, maybe different edition that is nevertheless based on the same material.5 More recently, Alfredo Stussi—among many others—has called attention to a closely related issue:

Once the textual-critical problems regarding both content and form have been resolved, however many witnesses there may be, the manner in which this information is presented to the potential user of the critical edition is highly important. A critical edition is, in fact, a working hypothesis, and hence the reader must be able to verify it point by point, and, indeed, to disagree [with the editorial choices made].6

Digital editions as ‘editions-in-time’

In principle, a Scholarly Digital Edition should present the five features listed under section 1 regardless of the subtype to which it belongs.7 Furthermore, these same features should be fulfilled in every single module that makes up an edition, otherwise the potential of the digital environment would not be fully exploited. As recently claimed by Dino Buzzetti: ‘One might be tempted to say that present-day digital editions, for all their merits, are not yet fully digital, since they do not fully exploit the distinctive features of the digital form of textual representation to obtain better critical and analytical results’.8 And further on: ‘[…] the means of rendering a text—spoken, written, printed, digital—affords a different and distinctive approach to seizing it. In this respect, an “image”, or representation of the text in digital form, can considerably enhance our opportunities of penetrating deeply into its discourse’.9 A major advantage of a digital edition is its potential to provide a model capable of embodying the edition-in-time (‘edizione-nel-tempo’) as a result of text-in-time (‘testo-nel-tempo’) postulated by Gianfranco Contini.10 A text can be considered as a dynamic entity that originates from the tension between the initial, multi-faceted creative process and the subsequent re-elaborations and modifications which it inevitably undergoes. According to Contini, the edition too should be ‘in time, opening up in the “pragma” and making the editorial choices subject to a variable teleology’.11 In fact, in order to convey more precisely the mobility of the text, an edition cannot carry the connotations of a rigidly defined structure; rather, it should aim at injecting history into the critically reconstructed text by taking into account the different synchronic stages that make up its diachronic dimension, namely the evolutionary line of the textual tradition.12 In this perspective, Contini’s view of edition-in-time represents an antidote to Bédier’s radical scepticism towards any kind of reconstruction. The inherent risk of Bédier’s well-known argumentations against the so-called ‘Lachmannian method’ is to open the path to editions which, in the name of the dogma of the witness as vehicle for an alleged historical text, result instead in an inactive and almost frozen object, suffocated within the borders of its own materiality, fixed once for all. Broadly speaking, this is what happens with his critical edition of the Chanson de Roland based solely on the Oxford manuscript.13 In order to give new life to a paralysed text, the Italian philologist Cesare Segre, on the pattern laid down by Contini, opens it up again to the diachronic dimension —both in the direction of a rehabilitation of the reconstructive process based on a thorough recensio, and simultaneously by representing the after-life of the text, as well as its dynamicity, which reaches the highest peaks in two families of witnesses (labelled as γ and δ, respectively).14

In order to grasp the dynamic nature of a text (its inherent mouvance, as defined by Paul Zumthor15), not so much the witness itself but rather the ‘critical apparatus’ acquires a crucial role, as underlined on several occasions by the Italian philological school.16 It is no coincidence that back in 1974 Cesare Segre argued that the apparatus should be the location where the tension between respect for the antigraph and the innovative thrust of the copyist is brought to the fore:

There needs to be a turnaround [...] in the hierarchical relationships between the text and the apparatus, give greater emphasis to the apparatus and consider the text as a neutral surface [...] on which the philologist has grafted the readings which he deemed certain among the many considered. However, the edition deserves the attribute of being ‘critical’ through the apparatus, if discursively problematic: because it summarizes the diasystem of the tradition, and because it carries out a full assessment, even if not always conclusive, of the readings.17

It is therefore in the apparatus that the diasystem18 of the tradition is best highlighted, and its historicity fully appreciated.19 Since an edition is always a working hypothesis (see above, footnote 6),20 the critical apparatus is the key that allows the reader to understand the choices made by the editor to present the text in that particular shape. It is in the apparatus that the reader finds information about the editorial process that resulted in the text he or she is reading—thus enabling her/him to evaluate the editor’s decisions—as well as the different shapes assumed by the text itself in the period in which it was composed and committed to posterity.21 A crucial problem which arises when dealing with the apparatus in the digital medium is how to make it respond to modern scholarly needs,22 without either drastically reducing or completely concealing its critical nature.23

The ‘digital’ apparatus and quantity

In most of the digital edition projects developed so far, a specific form of apparatus—when provided—seems to have gained particular success, namely the so-called ‘horizontale Kollationspartitur’24 exemplified by the famous edition of Chrétien de Troyes’ Lancelot (= Le Chevalier de la Charrette, ca. 1180), which was produced at Princeton by a team of romance scholars between 1997 and 2010.25

Yet, from a theoretical point of view this ‘horizontale Kollationspartitur’ is too dependent on categories that still stick to a linear representation of the object text, not completely fulfilling the digital features mentioned above. For example, in order to meet the requirement of quantity (see above, section 1, point 1), a critical apparatus should drop its traditional focus on single words in favour of a sentence-oriented or even text-oriented approach. In other words, it should be centred on dimensions that would allow one to go beyond the ‘chopped’ variants with which readers are usually presented in paper editions (see Fig. 4.1) and which make it extremely difficult to identify the potential relationships between them:

Fig. 4.1 A page taken from Taeger’s 1996 edition of the Heliand, showing a linear critical apparatus focused on single words.

To overcome this word-oriented perspective would be to increase the number of potential users, in particular, though not exclusively, among scholars. A linguist, for example, might be interested in features attesting intra- and inter-linguistic variation for the representation of sentence-level syntax, semantics and discourse. This is the kind of information that is generally not present—or at the very least rare—in a traditional linear apparatus.

A brief example taken from the text currently being edited by the present writer—the ninth-century Old Saxon alliterative reworking of the Gospel titled Heliand—illustrates this point. The poem, about 6000 lines long, has come down to us in a nearly complete form in two manuscripts: a continental MS M (München, cgm. 25, preserved at the Bavarian Staatsbibliothek) and an English MS C (Cotton Caligula A.vii, preserved in the British Library). A further four fragments transmit short passages of the text: i.e. V (Codex Palatinus Lat. 1447, discovered by Karl Zangemeister in 1894 and now housed at the Vatican Library in Rome; ll. 1279–1358); P (formerly preserved at the University Library of Prague, now in Berlin, Bibliothek des Deutschen Historischen Museums, R 56/2537; ll. 958b–1006a), S (the Straubing fragment, currently held in München, Bavarian Staatsbibliothek, cgm. 8840; ll. 351–722), and—last but not least—L, the newly discovered Leipzig fragment, found in 2006 (Leipzig, Universitätsbibliothek, MS Thomas 4073; ll. 5823–5870a).26 Since the major editions produced so far27 are based on M—which has always been considered the guide-manuscript in editorial practice — many sentence-level linguistic phenomena that only C displays have been completely neglected. A word-oriented linear apparatus has contributed considerably to these phenomena being left out, and they have become so difficult to detect that even trained experts have serious trouble discerning them. A paradigmatic example is represented by the so-called attractio relativi (or ‘case attraction’),28 frequently attested in early Germanic languages such as Gothic, Old English, Old High German, as shown by the following sentence taken from the Old High German Liber evangeliorum (Otfrid, I.17, 38):29

Here the relative pronoun (then) features the dative case in agreement with its antecedent (uuórtun), even though the subordinate clause would have required the accusative.

Not surprisingly, Old Saxon relative clauses may also display case attraction. However, the phenomenon is not recorded in any of the most common and widely used Old Saxon grammars. The reason for this omission can be traced to the very nature of the available editions. Attractio relativi occurs almost only in the C-text, therefore the variant readings that should confirm the phenomenon are relegated to the apparatus. Since the latter is built as to focus on single words, it only accounts for the variation of the pronominal form rather than of the whole sentence, making it very difficult to pinpoint the structure. In a nutshell: a traditional linear apparatus risks concealing most of the complex linguistic and textual features that a thorough scrutiny of the manuscripts has brought to the fore. An interactive hypertextual apparatus would instead make them (more) visible.

The ‘digital’ apparatus and relationability

As regards relationability (see above, section 1, point 2): a scholarly electronic edition permits the presentation in the hypertext of all the evidence which the reader requires to grasp both intertextual and intratextual connections.30 In principle, a scholarly electronic edition should allow the editor to present the critically reconstructed text, as well as the different versions and the many forms the text assumes when it becomes part of a historical transmission chain.31 The interesting point is that, contrary to what is too often assumed, these two perspectives (the genetic/reconstructive and the historical/material) are frequently complementary, rather than mutually exclusive. In most cases, in fact, they do not preclude one another. The digital environment proves totally adequate for representing both the reconstructed interpretative text (when considered necessary by the editor) and its chronological dimension. The conventional paper edition, on the other hand, tends to privilege either a single stage or a few stages of the tradition.32 Furthermore, conventional paper editions hardly ever allow for cross-checking of the data since, if it is true that most of the times the apparatus accounts for the choices made by the editor, it is equally true that the elaboration of an alternative proposal by the reader remains a chimera, due to the paucity of information, and to the fact that the data provided are usually unstructured. Thus, the historically transmitted texts become almost unrecognisable in the apparatus of variant readings.33 In some textual traditions, especially those belonging to the Middle Ages, the aforementioned diachronic issue comes on top of other crucial editorial problems, such as how to represent the stratification of intertextual relationships within the same work.34 In a traditional paper apparatus, these features find their place in additional registers (e.g. the apparatus fontium) that accompany the lectio variorum. Similarly, Francesco Stella’s digital edition of the Latin Corpus rhythmorum provides a multi-layered apparatus which accommodates Loci vetustiores, Loci coaevi and Loci seriores, along with the lectio variorum given in the right upper part of the box, as shown in Fig. 4.2:35

Fig. 4.2 Corpus Rhythmorum: multi-layered apparatus.

This modality of representing intertextual features is undoubtedly an improvement over printed editions, especially if one considers that the apparatus can be accessed in different ways, and can point both to a synoptic view of different redactions (Fig. 4.3), as well as to one specific redaction (Fig. 4.4).36

Fig. 4.3 Corpus Rhythmorum: synopsis.

Fig. 4.4 Corpus Rhythmorum: single redaction.

Yet, one cannot fail to notice that the apparatus is linearly structured—which means that it mimics what may also be found in book form. In fact, the boxes that accommodate the various kinds of information are not mutually interconnected, which opens the path to the discussion of the next feature, i.e. multimediality/multimodality (see above, section 1, point 4).

The ‘digital’ apparatus and multimediality/multimodality

Multimediality and—when relevant—multimodality are the most important requirements the critical apparatus module should meet, since these two features subsume the previous ones and make them fully available to the readers. The improvement offered by these options should therefore be a major theoretical concern and perhaps also a priority in the critical debate. The possibility of connecting the apparatus with other windows showing, for example, the transcriptions of a single witness or the manuscript images is not simply a question of providing more information; rather, it allows the contextualisation of each variant reading, which can thus be studied in vivo, rather than in vitro. The apparatus gains new life and is hopefully used by more readers in a variety of different ways in the hyper-textual environment. The possibility of giving more context, as well as more ‘paratextual elements’—both epitextual and peritextual37—is a major improvement from the scholarly point of view, since it provides the tools to better interpret the text, and ‘enhance[s] our opportunities of penetrating deeply into its discourse’, just as envisaged by Dino Buzzetti.38

An ongoing project that meets the requirements of multimediality is the Parzival-Projekt, in which three research teams (based at Bern, Berlin and Erlangen), under the guidance of Michael Stolz of the University of Bern, are preparing the ground for a new electronic edition of Wolfram’s Parzival. Their theoretical assumption is that

A new critical edition of Parzival will have to come to terms with the abundance of variant readings and the not inconsiderable problems of establishing a text against the methodological background of the polarity of New Philology and New Phylogeny.39 A challenge that was voiced in the Parzival scholarship of the 1960s now seems more relevant than ever before. It was then argued that it was necessary ‘to publish all the material that was collected for critical assessment before the question of manuscript interrelation could be clarified’ (E. Nellmann). Perhaps the idea, when it was voiced in 1968, had a Utopian ring. Today, however, it can be put into practice, step by step, with the aid of computer technology, and at reasonable expense. A critical electronic edition will constitute a work-base that would be an indispensable prerequisite for any new edition of Parzival.40

Fig. 4.5 The Parzival-Projekt template.

Indeed, what the research teams claim to be their programme is embodied in the fourfold interface chosen to present the reader with their ‘edition’ (see Fig. 4.5).

In the upper left window a normalised critical text is given—based on manuscript D, which scholars have always considered as the guide-witness. The lower left window accommodates the apparatus of variant readings, while the windows on the right contain the transcriptions and facsimiles of each single witness. All the windows are interconnected by hypertext-links, and permit users an interactive interchange between base-text, apparatus of variants, transcriptions and facsimiles. On the screen, every variant is fully contextualised.

Similar theoretical concerns have inspired the ‘Electronic Heliand Project‘, started at the University of Venice in 2006 under the guidance of the present writer. The template used in this project is based on a series of click-and-drag resizable windows,41 which can be activated or deactivated by the user, so that he or she can freely choose the material to view and in which order, according to his or her own interests.42 The windows are not isolated items; they are connected by hyperlinks. Thus, by clicking on a word in the main window (at the top-left of the screen) the user can activate other windows, such as, for example, one containing the image of a manuscript, or one providing its transcription (see Fig. 4.6 and Fig. 4.7).

Fig. 4.6 The ‘Electronic Heliand Project’ template: flexibility of text representations.

Fig. 4.7 The ‘Electronic Heliand Project’ template: text and manuscript images.

What is relevant for the present analysis is that all the hyperlinks are centred on the apparatus (in that they proceed from it and return to it); therefore, the latter is only apparently shaped in a linear fashion, since each variant reading included in it by the editor and selected by the user can be contextualised at different levels: e.g. word, phrase, sentence, text and paratext,43 material document.44 The centrality of the apparatus module is perfectly in line with the view of the edition as a working hypothesis discussed above. Its hypertext structure greatly enhances the possibility of deriving information from the data provided, while conforming to a standard for the representation of the apparatus in digital form, in particular the standard developed and maintained by the Text Encoding Initiative (TEI) consortium that favours—among other things—interoperability and hopefully improves survival over time.

TEI encoding scheme: Main issues and future perspectives

Two of the most crucial issues related to the application of the TEI encoding scheme to the critical apparatus are, on the one hand, the ‘method’ followed to link the apparatus to the text and, on the other hand, certain specific encoding procedures. As for the former, the preferential adoption of the ‘parallel segmentation method’ to the detriment of the other two (‘location-referenced method’ and ‘double end-point attachment method’45) seems to be common to many born-digital projects that are TEI-conformant.46 This might depend on a series of variables, including the lack of easy-to-use tools supporting the encoding process, which can be difficult to carry out manually, especially in the case of double end-point attachment.47 Yet, parallel segmentation also has an evident theoretical advantage over the other two since the much-disputed concept of ‘base-text’ remains in the background: ‘the texts compared are divided into matching segments all synchronised with one another. This permits direct comparison of any span of text in any witness with that in any other witness’.48 Further on in the same section one reads that the parallel segmentation method ‘will also be useful where editors do not wish to privilege a text as the “base” or when editors wish to present parallel texts’.49 Note, however, that even when no base-text is postulated, no philologist would consider the critical apparatus simply as a ‘repository of variants’. The apparatus is indeed different from the descriptive lectio variorum one can get by, say, applying any collation software to the transcription of the witnesses. The apparatus is critical—i.e. interpretative—in that it accommodates certain variant readings, and excludes some others, according to the editorial principles to which the philologist conforms.50 A (Neo-)Lachmannian editor might want to include solely the readings which bear a stemmatic value, thus eliminating those of the codices descripti and the lectiones singulares. If the editor is, instead, more interested in highlighting the linguistic features displayed by a text, or the history of its transmission, he or she might want to include formal variants too, thus accepting for example the lectiones singulares in the apparatus. In both cases, the editor’s task is to discriminate between variants and/or groups of variants, and this can be done electronically only by structuring them through encoding.51

Coming now to specific issues related to TEI recommendations, a positive feature is undoubtedly represented by encoding flexibility. The fact, for example, that readings may be encoded either individually, or grouped for perspicuity using the <rdgGrp> element helps philologists to discriminate between variants and, if this is the case, to create a hierarchical order. The most problematic areas, instead, are constituted by specific phenomena like transpositions (which cannot be marked up explicitly), the handling of punctuation (of which no encoding examples are provided),52 as well as of omissions/additions or lacunae. Generally speaking, even the representation of the palaeographic features of the witnesses in the apparatus module can be quite problematic.53 Encoding a long (‘verbose’) apparatus entry is most of the times extremely time consuming, very tricky and often frustrating, especially when information needs to be suppressed because of the lack of elements and/or attributes, or even because it is not always possible to break up the information into consistent base-units in order to keep all witnesses synchronised. In those cases, adding a note (in the <rdg>?) would help—though not always; yet, it seems like avoiding the problem rather than facing it.

Apart from these difficulties, a major issue until recently has been represented by the ‘status’ of the <app> element itself, which seems to have been considered as a phrase- or word-level only element. Editors of texts whose length varies considerably between the witnesses (for example: poems showing a different number of stanzas, or both poetic and prose texts with many omissions) have found the <app> element, for example, nested in the <l> element, as prescribed by the Guidelines, somewhat awkward.

The editor was forced to repeat the encoding of the <app> string for every varying line (in this case, for example, from line 1 to line 84, which are attested only in the C-redaction of the Heliand). Clearly the time was ripe for a revision that would lead not only to a more economic encoding practice, but also to a practice more respectful of the actual hierarchy of textual content. Fortunately, the TEI has now (October 2015) accepted a proposal to make <app> a block-level element so that <rdg> can now contain <p> and <l> elements, which means that its content is no longer restricted by any line or sentence border.

An experiment in the wake of Lavagnino (2009)

The issue of the drawbacks of apparatuses that fail to serve their purpose regularly comes up in the scholarly debate. In an essay devoted to a critical rethinking of the idea of access to data, John Lavagnino argues that editors should represent their work as providing not simply data but rather critical points of view on the texts they are offering.54 Though the argument that by focusing on the ‘activity of the editor’, rather than that of the ‘user’, too many editions ‘offered access to the wrong thing’55 is not entirely convincing, I do share the opinion that an edition—and all the more its critical apparatus—should suit different audiences, as well as present ‘ways of filtering’ the data provided by the editor that could enable readers to interrogate also the text’s transmission history.56 The integration of textual scholarship and textual criticism that Lavagnino seems to call for in his article may offer a useful model for a more discursive critical apparatus, which would record the relevant information traditionally included in separate sections, for example in the introduction. An experiment in this approach is represented by the Digital Ramusio project, sustained by Ca’ Foscari University of Venice and coordinated by Eugenio Burgio, Antonella Ghersetti and the present writer. The project, whose main aim was to provide a hypertext edition of Giovanni Battista Ramusio’s Dei Viaggi di Messer Marco Polo gentiluomo veneziano (1559) capable of representing Ramusio’s own ‘desk’ in a virtual environment, was launched in February 2015 and is now available Open Access at the following website: http://virgo.unive.it/ecf-workflow/books/Ramusio/main/index.html. In the Digital Ramusio, the modal windows allow the user to visualise a chapter of the main text (R) in parallel with its major sources (Z, V, VB, L, P and VA, F), three of which are given in new born-digital editions. Each chapter and chapter section of R is accompanied by a philological commentary made accessible through pop-up windows which present the relevant interface to the user. The philological commentary—containing the identified sources and their variant readings against R, the analysis of their manipulation by Ramusio, as well as some informative notes—serves the purpose envisaged by Lavagnino, namely that of filtering the data provided by the editor through the provision of a narrative able to explain and make sense of them, as shown in Fig. 4.8.57

Fig. 4.8 Ramusio’s Dei Viaggi di Messer Marco Polo: the philological commentary by chapters and sections.

It goes perhaps without saying that this editorially radical choice was made possible by the peculiarity of both Ramusio’s text itself and the Polian tradition to which that text belongs, whose intrinsic mouvance is well recognised by scholars in the field.58

The ‘digital’ apparatus and user interaction

After this brief digression on encoding procedures and experimental apparatuses, let us return to our main path and discuss the last core feature that the apparatus module should display, i.e. user interaction (see above, section 1, point 5). The idea of setting up a shared virtual environment, which would facilitate a dialogue between the editorial team and the (critical) reader, is indeed tempting. This, in principle, would enhance scholarly debate, promote new research and allow quicker updates of the edition. In fact, many questions are still open, among which: how can a ‘social/collaborative edition’59 be reconciled with the idea of the editor’s scientific responsibility? Should we plan on filtering the information, and how? In a ‘social edition’, what are the valid indicators of real user interaction? What can we use to create an accurate model for evaluating socially enhanced editions? With the exception of some essays that focus on the first of these issues (i.e. the editor(s)’ scientific responsibility), to my knowledge there exists no detailed study of user interactions in the field of ecdotics.60 A study of this kind would be most welcome, since it would prepare the ground for the philologist to make operative choices on the basis of objective data, rather than of subjective points of view.

Final remarks

In lieu of a conclusion, I would like to provide a provisional answer to the question posed in the title and at the beginning of this chapter: is a ‘protocol’ for scholarly digital editions desirable? As shown by the discussion on the apparatus module, the answer seems to be yes, provided that the protocol encompasses the domains in which scholarly digital editions may offer important advantages over paper editions, without being too strict as to orient the editor to follow a specific ecdotic praxis. The philologist should take full responsibility for his or her choices, which depend both on the theoretical framework to which he or she conforms, and on the peculiarities of the manuscript tradition he or she is dealing with. Nonetheless, the adoption of a (shared) protocol would have positive repercussions both on the products (i.e. the editions) and on the core features themselves, the former acting as material to test the reliability of the latter when deemed necessary. This study has also shown that the critical apparatus is the site where the dynamic nature of a text can be conveyed better by the editor(s) and grasped more easily by the readers:

Anyone who uses a critical edition should be able to grasp with ease the criteria followed and the decisions made at each level. A critical apparatus serves this purpose no less than a detailed introduction. [...] the apparatus allows convenient comparison of the readings accepted in the text with those discarded.61

Particular attention to the representation of that module—often dismissed as too technical—in digital form should perhaps be the major concern of anyone who intends to produce an edition that truly deserves to be called ‘scholarly’. In this respect, a revision of the section of the TEI Guidelines dealing with the Critical Apparatus to make it meet the theoretical needs of the philologist would not only help enhance the value of the apparatus itself, freeing it from being considered simply as a ‘repository of variants’, but would also enhance the digital edition as a whole.