Paolo Monella, Methods in Digital Philology

What is the added value of the computational technologies in a critical digital edition in classics? Why should a textual scholar choose digital over print in this specific field? The drawbacks of the digital choice include additional training and costs, slower workflow, less interoperability and durability. Do its 'pluses' balance them out, or even surpass them? \#\#\# 1. Funding The **additional funding** that research funding agencies seem eager to pour on projects involving digital aspects is producing an (etymologically) preposterous situation: instead of inquiring *whether* digital methods can help their research, many applicants first decide that some digital must be involved, they try to find out how. But I think that we can all agree that, if we want to be intellectually honest, we should not take the economic metaphor of the 'added value' so literally as to consider increased funding opportunities *per se* a 'plus' for classical research. Otherwise, the digital contribution to our research will be mostly cosmetic today, and the relative funding will eventually dry out -- as it happens with shallow trends. The same, but from the opposite perspective, goes for the individual careers of digital classicists (full disclaimer: like me) and for the destiny of digital humanities (DH) centres and institutions: we should not ask ourselves what Classics can do for the DH (how to find a spot for them in our projects and universities), but what the DH can actually do for Classics. And this question should be asked by digital classicists and 'classicists at large' together, "tearing the wall" between them -- as Samuel Huskey said in a previous seminar of this series. \#\#\# 2. Availability The **availability** of large textual *corpora* is compensated, on the negative side, by the interoperability and durability issues I mentioned above. Also, the goal of creating and making comprehensively classical *corpora* available has long been achieved, and hardly complex digital models and formats such as plain text, Beta code or PDF are sufficient for the purpose of browsing and reading them, while plain text is sufficient for simple, Boolean or lemmatised searches. \#\#\# 3. Annotation and analyis The advantages of digital (also semantic) **text annotation and analysis software** are apparent. Such features arguably make an edition more 'scholarly', since commentary and analysis are part of the editorial work at large, but do they make it more 'critical'? In Classics, the very definition of a 'critical edition' relies on an original work of *consititutio textus* and on its documentation through an *apparatus criticus*: what is the added value of digital methods in this? \#\#\# 4. Phylogenetics **Phylogenetic** algorithms make a strong claim at providing 'specifically philologic' added value, since they help in the phase of the *recensio*, an important step towards the *constitutio textus*. Their experimentation is ongoing, amid some diffidence from 'traditional' scholars and a methodological tension between 'counting' and 'weighing' variants. Also disputed is the cost-benefit analysis of the application of these algorithms, compared to traditional variant evaluation. \#\#\# 5. Witness facsimiles and enhanced recording of variants/versions An important 'plus' of digital scholarly editions, upon which its success in fields outside the critical edition of 'canonical' classical texts, is an increased potential of recording philological data closer to the 'document' -- papyrus, inscription, manuscript, *incunabulum* etc: - A quite common feature is the linking of the text to the **facsimiles** of witnesses. - Another such feature is the possibility of recording an extensive amount of **variants or parallel versions** from a textual tradition, potentially much more than that which would find place in a print edition, - either by comparing/collating full **transcriptions** of witnesses, possibly encoded at different layers, (e.g. a paleographic/orthographic one and a normalised one) - or by adding **annotations** to a master file with the *textus constitutus*. If those variants are tagged and distinguished between 'orthographic', 'substantive', 'stemmatically meaningful' etc., visualisation sofware can sort, show or hide them. But what is the added value of this enhanced availability of philological data, especially for a classicist? Let us start by saying that facsimiles and enhanced recording of variants/versions can serve two goals: - Improving the *visualisation* (the human reading) of additional philological data; - Allowing the *processing* (computation by algorithms) of such data. The *visualisation* of facsimiles, parallel versions and non-Lachmannina variants can produce three scientific advantages: **5.1 Text/document** They are contributing to the success of digital philology in those sub-fields of textual editing in which information about the **document** itself is key -- papyrology, epigraphy, codicology, editions of texts with meaningful text-image interaction such as Leonardo's notebooks or of unique textual historical sources such as the *Old Bailey* or the *Codex Sinaiticus*; As far as Classics are concerned, only texts with unitestimonial traditions such as inscriptions and papyri have taken advantage of this aspect of digital editing. **5.2 'Plural' texts** The added value of digital philology is also apparent in research fields or in which textual variants or even parallel versions of a text are considered culturally significant *per se* -- medieval textual *mouvance*, genetic editions and *filologia d'autore* of contemporary authors, New Philology. However, as I argued in a 2018 article (), most of the variance in witnesses of classical textual traditions has been introduced much later the author's time, so it is not considered meaningful enough by classicists to give them reason to record the enormous number of orthographic variants or *lectiones singulares* found in the witnesses of Greek and Latin texts. This is why in Classics, among texts with multi-testimonial traditions, only a few 'non-canonical' ones have digital editions -- including Homeric papyiri, non-literary works with multiple ancient versions -- because for these traditions different versions are considered culturally significative. **5.3 Popperian falsifiability** While the value of 5.1 is perceived only by part of Classical studies (papyrology, epigraphy) and 5.2 does not appeal classicists in general because is is not directly useful for the *constitutio textus*, a third advantage does apply to all editions, including those of 'canonical' classical author: an increased availability of philological data would greatly improve the *falsifiability* of the edition in a Popperian sense, thus its scholarly value. A very influential definition by Patrick Sahle () says that a digital (not digitised) scholarly edition "cannot be given in print without a significant loss of information or functionality". The advantages described in 5.1, 5.2 and 5.3, focussed on the mere *visualisation* of additional philological information, can be confuted by an interesting objection raised by Daniel Kiss during his talk in this seminar series: Dr. Kiss argued that with the availability of enough pages, every quantity of information, including facsimiles and myriads of variants, can theoretically be "given in print" too, not only in digital form. However, digital editions live up to Sahle's definition if we move from *visualisation* to **processing** of formalised philological information, i.e. if we create algorithms (software) that *process* that wealth of data and produce further information for the consideration of scholars. Processing produces further potential advantages: **5.4 Analysis of paleography and orthography** I shall mention two examples, taken from my own digital editions: (a) my edition of Ursus Beneventanus is based on a 'graphematic' transcription (), enabling computational analysis of spelling and abbreviation practices; (b) my edition of the *Chronicon* by Romualdus Salernitanus () attributes a 'type' to each variant ('substantive', 'transposition', 'orthographic' etc.), enabling analysis of Latin medieval orthography. Again, the scientific advantage is apparent for sciences such as paleography or linguistic history (Latin, Greek and Neo-latin languages alike), but not for the traditional goals of classical studies. **5.5 Analysis of variants** Human comparison and study of select variants can be aided by spreadsheet-like technologies, ranging from simple Office suite spreadsheets (see S. Rozzi, ) to full-fledged databases. This approach undisputably makes a specific contribution to the *constitutio textus* (only apparently similar to phylogenetics, since variants here are selected, weighed and evaluated by the human, not by an algorithm). However, one might wonder whether the added value it provides compared to 'traditional' comparison of variants counterbalance the additional work of populating variant tables. In any case, points 5.4 and 5.5 fall within the realm of Sahle's "functionality": the latter, in fact, "cannot be given in print", not even with unlimited abundance of pages. \#\#\# Conclusions In an article that I published in 2018 () I focussed on point 5 above, and concluded that the potential added value of digital philology would not appeal classicists unless they would open themselves to a new type of edition, in which medieval (secondary) textual variance would be considered meaningful *per se* (see point 5.2 above), for instance to provide materials for the study of the history of Latin and Greek language, orthography and paleography in the Middle Ages. In another article, published the next year (), I argued, somewhat more optimistically, that digital editions could host *both* the *textus constitutus* *and* additional data on the different stages of the tradition. After further years of experience, I tend to return to the somewhat gloomy conclusions of 2018 for the application of digital philology in Classics, though with some hope *in cauda*: - The building of large plain-text *corpora* (point 2 above), as well as annotation and textual analysis (point 3), will likely continue to thrive, but they do not belong to the core of textual criticism, at least as it is conceived in Classics; - If phylogenetic methods (point 4) will reach such a level of maturity to convince classicists, they might become part of scholarly practices; - Yet, a wider recording of textual variance, either to increase falsifiability of philological hypotheses (point 5.3) or for further computation and analysis (point 5.4) seems to be very far from penetrating editorial practices in Classics (except, maybe, for point 5.5). - In my 2018 article I suggested that Classics might join forces with scholars of medieval culture and with language historians to create editions that record larger amounts of philological information in medieval manuscripts (points 5.1 and 5.2). This dream has been perceived as utterly unfeasible in the reception of that article of mine. But some hope may come from projects such as the one I am working on at this time, PAGES (Priscian's *Ars Grammatica* in European Scriptoria, ERC Grant 882588, ), which aims at taking advantage of the potential of digital philology to produce not only a Lachmannian critical edition (in digital and in print) of Priscian's *Ars*, but also to record and publish an extensive database of transcriptions of Greek passages, Greek hands, glosses, Renaissance interpolations and other philological data which, in fact, broaden the scope of the study of Priscian beyond the classical age.

Paolo Monella Curriculum
DH bibliography
Paolo Monella home page