Digital work in premodern studies is a treasure house of riches. Projects on medieval manuscripts, artifacts and art, intricate maps, and cosmopolitan cities, thrive and proliferate today, presenting extraordinary results and, also, special challenges.

Some of these projects drive cutting-edge technological innovation: Peter Bajcsy's group at iCHASS, working on the manuscripts of Jean Froissart (chronicler of the Hundred Years war between England and France), is developing widely-applicable tools for image analysis and pattern recognition. To keep track, Chris Baswell and Matthew Fisher are compiling a central database of all European manuscript digital projects. Beyond manuscripts, there is the visualization of cities, artifacts, monuments, and sites: among the highlights of the 2008 HASTAC conference were UCLA's virtual Ancient Rome and the collaboratively-produced virtual Temple Mount of Jerusalem http://www.ust.ucla.edu/ustweb/Projects/israel.htm & http://www.archpark.org.il/

Excitingly, this work is no longer confined to premodern Europe. The Timbuktu manuscripts of medieval Africa are being digitized: http://www.sum.uio.no/timbuktu/index.html http://www.aluka.org/page/about/news/nytimes20080521.jsp International Dunhuang, mentioned by Cathy Davidson on HASTAC forums, is putting online the fabulous resources of a cave-complex at a famed cosmopolitan city, Dun Huang, on the old eastern Silk Road.

For the Global Middle Ages initiatives that aim at studying an interconnected world in deep time, the biggest challenge is how to weave together disparate existing projects and our own, new, efforts, in narrating a multi-layered, multi-dimensional past that privileges no single region in the world, but offers as many points of entry, and analysis, as there are subjects to study them.

We have many digital projects brewing, all clustered under the name of Mappamundi ('map of the world') and focusing on different locations in time and space.

'Discoveries' of America is a visualization platform that depicts geographic imagination around the world--in Islamic civilization, China, and Scandinavian Europe--as societies imagined, or traveled to America before Columbus. It will show how maps flow and circulate in networks of cultural exchange, raising questions of timelines and knowledge. Along with navigational techniques, instruments, and cartography, Ayhan Aytes (the projected designer of 'Discoveries') anticipates grafting 3 layers onto the platform: a semantic layer with taxonomies that allow for varied tagging functionalities, a chronologic layer offering multiple timelines (episodic, linear, cyclical, discontinuous), and a spatial layer that intersects terrestrial with celestial navigation. Envisaged as a model for how Mappamundi could function, people will be able to contribute wiki-style modules to 'Discoveries' so that there's ongoing collaborative knowledge-making.

Ana Boa-Ventura, who hosted the very successful HASTAC Scholars Forum on Metaverses, is proposing her project on farming social media like YouTube, FlickR, and Wikipedia for resources to support study of a global Middle Ages, to the National Center for Supercomputing Applications this month as part of its 1 million CPU hours initiative, and hopes to work with colleagues at the NCSA to mine social media for attitudes to and materials on the Middle Ages.

Susan Noakes is collaborating with researchers at St Andrews, Scotland, in a Byzantium-Constantinople-Istanbul project associated with the Byzantinist Paul Magdalino. The St Andrews team has digitized the Topkapi palace, and are digitizing the Hagia Sophia, and sunken ships and their contents excavated from the Byzantine harbor at Yenikapi, Istanbul. Thanks to the efforts of Kevin Franklin of iCHASS, our Africanist colleagues will soon launch a collaboration with the Center for High Performance Computing in South Africa, to begin digitizing medieval Africa. We are also planning collaborations with supercomputing centers in Cyprus and in Beijing.

All these projects bring many challenges: from questions of funding, of how to find time to do all this extra work, how to amass data, and devise tools and interfaces, to--most fundamentally--the human challenges of working across cultures, languages, races, genders, politics, disciplinary training, institutions, and a wild and woolly mix of personalities, mindsets, and habits. Our colleagues in high-performance computing are often flummoxed by the "fuzzy logic" of humanities scholars (fuzziness that we, of course, know as creativity and inspiration). Humanities scholars can find it hard to translate complex research issues and ideas into "data" that can be queried by applications devised for computers. If the Global Middle Ages Project, as Kevin Franklin puts it, has the potential to become as large as the human genome project (!), it must build many kinds of bridges, through many kinds of conversations (including discussions like this HASTAC forum), and across many divides. Collaboration requires patience, and a good deal of time.

Some of these divides are, at base, political. When considering whether Byzantium-Constantinople-Istanbul could be digitized by the center at Cyprus, we were advised by an Ottomanist that locating the project at Cyprus would alienate the government of Turkey and colleagues in Istanbul. In the politics of the neighbor, the wars of the Middle Ages, it seems, are not yet over. And an Africanist warned that we should delicately and carefully approach the subject of digitizing Africa through the location and technology of Capetown, since South Africa is often seen by the rest of the continent to harbor hegemonist cultural tendencies. But thinking about how the politics of the present shape the production of the past--instead of how the past determines the present--can also be invaluable. Since my blog post in the HASTAC Scholars forum on the Future of the Digital Humanities, a group of us at the University of Texas has decided to take our project on Holy War into the slew of how contemporary politics has used the past. The reasoning goes like this. For a thousand years now, the countries of the West and the countries of Islam have been plagued by wars described--often without irony--as "holy." To some in Islamic countries, the holy wars known as the Crusades, beginning in 1096, are the foremost paradigm of an expansionist West, and the West's militarized past. To some in the West, the religion of Islam was birthed in the context of Holy War in the 7th century, with the Qu'ran authorizing the conquest of the non-Muslim world for the spread of religious domination.

I teach a graduate seminar called Holy War Redux as a series of questions in the classroom and in scholarly research. What is a holy war, and what conditions are necessary before a holy war can be declared? Once declared, do holy wars ever end? How are holy wars distinguished from wars waged for political, economic, territorial, and other purposes? Must holy wars be justified ethically and morally as just or defensive wars? How many societies have instituted holy wars in the history of the world, and what is the role of nationalism, ethnicity and race, tribalism, regionalism, caste and class interests in the waging of holy wars? Who are understood as combatants in holy wars, and what is the place of innocent victims? How is society re-organized as a result of holy war? What social, economic, political, psychological, or demographic changes result? How is the rhetoric of holy war refreshed over time, and how do cultures of the neighbor shape such wars? What logistics and statistics underlie holy wars?

These questions are not only of historical interest. Two renowned scholars--one a medievalist, the other an international relations theorist--have argued that the West and the countries of Islam are locked in a primordial struggle that will continue to produce holy wars in our current millennium. Bernard Lewis, an advisor to the Bush administration who urged the invasion of Iraq, called this struggle "the clash of civilizations." Samuel Huntington, in a famous book whose title borrowed Lewis' phrase, gave the phrase contemporary currency by suggesting that the end of the Cold War left a vacuum that is now filled by the "clash of civilizations" between the West and the East (the East being constituted by Islamic countries, and "Sinic" societies that form a natural partner for the countries of Islam).

Humanities scholars from several disciplines have argued vociferously against the simplistic and dangerous reductionisms of the Lewis-Huntington theses, but their arguments have thus far been driven only by conceptual reasoning, not by data. We are thinking to see if the Digging-into-Data Challenge can offer us an opportunity to discover if large-scale data mining can furnish evidence of a different kind--persuasive in different ways--from all our earlier arguments against the dangerous, simplistic, and mischievous reductionisms of Lewis-Huntington.

My example, in this, is Robert Pape's use of large-scale data in his Dying to Win: The Strategic Logic of Suicide Terrorism, which amassed a database of suicide terrorist attacks around the world from 1980 to 2004, using sources in Arabic, Hebrew, Russian, Tamil, English and other languages. Mining the data, Pape found that "Islamic fundamentalism is not as closely associated with suicide terrorism as many people think." His data showed that "the world leader in suicide terrorism" are the Tamil Tigers of Sri Lanka, a Marxist secular group that invented the "suicide vest." The data also showed that the incidence of suicide terrorism rapidly diminishes when "modern democracies" withdraw military forces from the territory that the terrorists view as their homeland. The data drove Pape to conclude that "suicide terrorism is mainly a response to foreign occupation and not Islamic fundamentalism."

We are thinking that our project for the Digging-into-Data Challenge should be to mine several different databases to test the dual Lewis-Huntington hypotheses promoting the idea of a "clash of civilizations."

If the data show that the incidence of holy war is linked to specific historical events, like the discovery of fossil fuels in Saudi Arabia in the 1930s (leading to "oil crusades") or the establishment of the state of Israel in 1948 (note Osama bin Laden's repeating references to a "Crusader-Zionist" alliance), it will be possible to argue that specific events, not primordial relations of enmity, shape the relationship of the West to the Near East. Rather than an essential struggle to which no end can ever be envisaged, and for which all solutions are necessarily inadequate, specific solutions can be envisaged to answer to specific events.

If the data show that the rhetoric of holy war was invoked at historical junctures before or during the Cold War, the force of Huntington's assertion of a "clash of civilizations" in the wake of the Cold War's closure is also diminished by data. Our group in Texas is collaborating with colleagues at the University of Sheffield's Humanities Research Institute on the subject of the Cold War.

Our working hypothesis is that no "clash of civilizations," either as formulated by Lewis, or as elaborated by Huntington, exists. Data could thus supply powerful alternative ways to prove incorrect these twinned pernicious formulations that have widespread influence today. So widespread is that influence that Fareed Zakaria, in the latest issue of Newsweek, just reflexively reinvoked this non-existent "clash of civilizations" again, as if the term witnessed conditions of reality, rather than political ideology.

We are thinking that, in the first instance, Holy War Redux could thus be a hypothesis-driven project of some value to contemporary political culture and to work in the humanities. Beyond that, we could apply data mining to address the larger research and teaching questions related to the history, incidence, and conceptualization of holy wars around the world. If we do, indeed, work on this project, there will be many challenges for the team at the Texas Advanced Computer Center, a supercomputing center that is attempting humanities projects for the very first time, to solve, along with our Sheffield counterparts at the HRI. But their solutions could have valuable and widespread applicability for humanities work on this campus and elsewhere.

This is not so far a stretch from digital work on manuscripts and images as one might think. This is simply another way for medievalists to work, as we remind people of the uses of the past--remember, for the Bush administration, Bernard Lewis' authority was as a medievalist par excellence--it's just an extension of the possibilities made available by data, digital textualities, and tools today.

• Login or register to post comments

As co-facilitator, I echo the welcome given by Michael and encourage all to contribute their throughts to the discussion. One reason I suggested this forum topic was the great potential for interaction between scholars of all disciplines as well as between scholars & administrators. Additionally, we all have a shared intellectual history in texts: this can be a very personal matter if one realizes what is at stake. I also invite futuristic speculation on how the fate of premodern texts in a postmodern age might influence the fate of postmodern texts in the distant future, as well as thoughts on the ultimate limits of preservation & digital representation.

For anyone desiring to read Danuta Shanzer's study on the status of palaeography and editing in US classics departments, here's the full citation (I couldn't find a nice link for it):

?Editions and Editing in the Classroom: A Report from the Mines in America,? in Vom Nutzen des Edierens: Akten des internationalen Kongresses zum 150-jhrigen Bestehen des Instituts fr sterreichische Geschichtsforschung, ed. B. Merta, A. Sommerlechner, & H. Weigl = Mitteilungen des Instituts fr sterreichische Geschichtsforschung, Ergnzungsband 47 (2005): 355-368.

I wanted to comment briefly on a bit that Michael wrote:

"There are great masses of manuscripts that people haven't even examined yet, much less considered for digitization. One immediate concern that occurs to me is that of selection biases. Since the process is expensive and laborious, those manuscripts deemed most important are first in line. That group would likely not include, however, works like those Yager examines or that many, many other scholars focus on. Does this problem mean we should just go through the archives in order of shelf marks?"

Michael has hit upon a significant problem in digitization efforts, a problem that has pervasive effects throughout all disciplines. (I will use a classical/medieval example simply because it is familiar.) With regard to digitizing ancient & medieval texts, too often digitization projects focus on two main categories of texts:

1. Texts transmitted by the "prettiest" manuscripts, i.e. the greatest number of illuminations and elaborate initial capitals.

2. Texts written by those esteemed as great authors (or otherwise widely known texts).

While these two categories of texts are vital and absolutely need to be digitized, overprivileging them does little to benefit the scholarly community. The "prettiest" (in modern eyes) manuscripts often are late medieval Bibles or Books of Hours - made for display or for the wealthiest of patrons. The texts themselves are well established & most lack substantial marginalia (because the purpose of the books was not necessarily practical). These pretty books may be of use to art historians and students of the development of scripts - but beyond there is nothing new or strange about them. When a library has a limited amount of funding available to digitize manuscripts, I humbly suggest that perhaps a manuscript sans illuminations or a text more obscure than the Bible might be a more substantial contribution than the most aesthetically pleasing codex.

At this point it is also somewhat unhelpful to overprivilege texts of authors already renowned. I am a bit more hesitant about this category - we are still missing texts from great thinkers such as Aristotle and Cicero (among many others); and the marginalia in school texts can also be a unique record of a medieval student's thought process. However, the academic community would benefit immensely from the digitization of more obscure works. Texts often go unstudied because no one has broken the pathway. Few people want to be the first to edit or translate a text - especially if there's no guaranteed payoff. (It's worth noting that editing is rather a thankless task in the US - numerous scholars have told me that one cannot count on a edition to bring tenure, a job, or even the Ph.D.) Texts go untranslated or unedited for numerous reasons, many of which relate to value judgments: the author is unknown or the famed Anonymous, the script is difficult, the grammar is seen as corrupted or "tainted" by vulgarisms, the subject isn't currently en vogue, the genre is generally seen as derivative or "mere compilation" (e.g. encyclopedias & chronicles), the location happens to be in an out-of-the-way monastery or library, etc. The point is that quick judgments made centuries ago by a frustrated cataloguer or bored scholar can banish a text to willful obscurity. These underprivileged texts are least likely to be selected by libraries beginning digitization projects, but they are the texts most in need of accessibility and study.

• Login or register to post comments

This has been an amazing forum. Thank you for sharing and inspiring so much.

Congratulations! And thank you.

• Login or register to post comments

Jorie, the costs you relate stunned me. They also make me wonder what other unintended consequences result from the impulse to digitization that's so prevalent now. I honestly have no idea what many of them may be, but I hope others will enlighten me.

I particularly like your suggestion that scholars coordinate their work to share costs and resources. It fits well with the urge to collaboration and more openness in scholarship. I wonder if there's anything like a clearinghouse website where scholars can self-organize in this way. If not, I think such a site could be a concrete and useful result of this forum. I know the new HASTAC site will offer a number of Wikis. A similar model could work for manuscript work, as well.

• Login or register to post comments

The following links don't solve the problem of undigitized material & the cost of digitizing it. See my reply to Mike's post about the idea of an academic clearinghouse site - which is a brilliant idea. Thanks to both of you for coming up with that collaborative thought! I, too, am pained by the cost a scholar must shoulder (often without reimbursement from his/her university) for microfilming or digitizing a manuscript. I recently inquired about digitizing a couple rolls of microfilm - the cost is just as high as the priced you cited!

At any rate, I wanted to list a couple links that help with finding material that has been digitized. As many of us know - once something is digitized and online, there is the entirely separate task of locating it.

Catalogue of Digital Manuscripts - The goal of this initiative (led by UCLA professor of English Matthew Fisher) is to index the many digitized manuscripts that are scattered around the Internet. There are links to the material; it can be searched via a variety of fields (language, location, etc.). Not everything is there, and I do wonder how they will handle links that change or cease to work, but this is a commendable project.

Digital Scriptorium - A collaborative archive of digital manuscript images

I'm sure there are others. I, frankly, got tired of looking through all the individual library collections, which would be silly to list here. This, I think, brings us to another issue: how do we find what we need? How can the process of locating items be made more efficient?

The issue with locating materials is that there is so little collaboration between universities and institutions at this point. Individual libraries digitize their own material and put it up in dribs & drabs using their own taxonomy (often mss. and papyri are not listed in general library catalogs, and when they are - let's just say the quality and completeness of entries varies). Libraries publish their own manuscript catalogues, but other than Worldcat, there exists no main database (to my knowledge - am I wrong?) for finding manuscripts of a particular work, digital or otherwise. Doing a project involving manuscripts (and incunables) involves a lot of scouring bibliographies and catalogs of individual libraries. After all of this time-consuming work, scholars still find out about hidden/unknown/obscure mss. after-the-fact. It's beyond frustrating.

In searching for an article on the UCLA digital mss. cataloguing project, I came across this article:

Time to Change our Thinking: Dismantling the Silo Model of Digital Scholarship(by Stephen Nichols)

To summarize Nichols' views would make this long post even longer - suffice it to say that he argues strongly for changing the way we think about digital humanities projects. Collaboration, he says, is key to digital projects on numerous levels. I think his point is sound & the world needs to hear it - he says a lot more as well - go check it out & see what you think.

I've included an excerpt below (in italics since blockquote isn't working):

It is not news in 2009 that digital humanities require a wholly new mind-set. That is what is meant by ?the cogito? of digital humanities. We cannot continue to focus simply on digital projects while ignoring the intellectual and social context in which they take place. We must begin by accepting the very different social, intellectual, and institutional context fostered by data-driven research. Digital scholarship creates a potentially productive network at many levels, and it entails significant change at the level of the individual scholar, in terms of operational methods, and in the kind of intra- and extra-institutional partnerships required.

The typical digital project cannot be pursued, much less completed by the proverbial ?solitary scholar? familiar to us from the analogue research model. Because of the way data is acquired and then scaled, digital research rests on a basis of collaboration at many levels:

• First, as a partnership between scholars and IT professionals;

• Second, as a dynamic interaction between scholars of the same and different disciplines, since the data is too large to be handled by a single scholar, and too varied to be encompassed by a single discipline;
• Third, in concert with a team of IT professionals responsible for designing the site, developing functionality as requested by the scholars, posting the data, and, not least of all, assuring access to end-users around the world.

• Login or register to post comments

You can compare digitisation with Open Source. "Everything that is already there is free. You have to pay for what is not there". This means that it is reasonable to have you pay for the work that needs doing when a digital copy has to be created.

There are several issues with this.

• You cannot predict in advance what is available and therefore the costs
• People will restrict their work to what has already been digitised
• Important material may not get digitised in this way
• When an archives burns or collapses, everything that has not been digitised is lost

There are problems in the execution as well. When scans are stored in the JPG format, they are compressed consequently these images are not suitable for digital restoration. They are at best useful to show on websites.

A good modern archivist is aware of these issues and he can also tell you about big investments that had to be redone because of changes in technology.

Thanks, GerardM

• Login or register to post comments

You also raise the difficulty of how to effectively (I won't say "accurately") represent texts in digital format if we cannot agree on exactly what constitutes a text. Current manuscript digitization projects, for the most part, do little more than provide high-resolution scans of each page. In some ways, this approach has much to recommend it since it leaves all decisions about the structure, nature of the content, genre, etc. to the scholars examining the texts. At the same time, however, it leaves little room for advanced cross-referencing, aggregation, or other techniques databases make possible. The manuscript becomes, rather than a text, pure image. Obviously, we must confront the theoretical questions about the structures of a text when designing new tools.

I also like your desire for a "broader tent". The problem with premodern manuscripts, though, is that most of them have yet even to be transcribed (itself eye-destroying labor), much less translated. Maybe the high-resolution scans could lead to more open transcription efforts, which would be great training in paleography, but I wonder how much broader than that the tent could be in practice. Once texts are transcribed, though, translation efforts could certainly come from a much larger coalition of interested people. I'm sure there would be some scholarly skepticism of the results, but so long as the entire process remains open and well-documented, anybody could go back and check the work when necessary.

One final point, before this post gets too long. The opportunity for variant readings that refuses to fix the text would reduce the need for editorial decisions about which works in the stemma to privilege. It would also be nice, as a reader, to have the variants easily at hand rather than hidden away in an appendix. Often, the variants make little to no difference to one's interpretation, but it would be nice to find that out quickly and (almost) immediately rather than flipping back and forth in a bulky tome.

• Login or register to post comments

Social projects are relatively new to the humanities or old if you think of the symposium as a humanities tradition of social research with food. I doubt they map on directly to "traditional" projects as easily as SUDA Online did. The two I'm involved in answer questions differently:

The Dictionary of Words in the wild lets us gather examples of public textuality. It is a hypothesis that is still trying to figure out its theory.

The Day of Digital Humanities is an experiment that will happen next week where about 100 digital humanists will blog a day of activities to answer the question, "Just what do digital humanists do?" We will see how it works.

There is the Distributed Proofreading part of Project Gutenberg and a bunch of science and art ones.

My question would be how can we imagine social research solving the problem Michael points out of scanning documents so that there is something for scholars to work on. If we take into account the political sensitivities that Heng points out we have an interesting problem akin to the issues we face here in Canada when university linguists try to "save" First Nation's languages. How can we help others to digitize cultural materials in their custody? Does it have to be to our standards? The linguistics department here has developed a successful summer school that might be a model, Canadian Indigenous Languages and Literacy Development Institute.

Geoffrey Rockwell, Univesity of Alberta, http://theoreti.ca

• Login or register to post comments

Have a look at Wikiprofessional. This project brings three things together. Literature in Pubmed, databases like UniProt and a Wiki with structured data. Combine this with a clever data miner that brings these things together and you will find information that was not available without one of those three parts.

Currently Wikiprofessional is bio-medical and English only. We are working on adding support for Portuguese and Spanish. Have a look at what it does on a the Wikipedia article on malaria... you can try it with any other Wikipedia article as well. :)

Thanks, GerardM

• Login or register to post comments

One simple tool along the lines of what you describe is probably not readily available, however you can likely piece together what you need from some of the following resources ...

Check out the TAPoR software page:

http://tapor.ualberta.ca/Resources/TASoftware/

See also TextGrid:

http://www.textgrid.de/en.html

See also Philoline and MONK for text comparison type tools.

http://code.google.com/p/text-pair/

http://monkproject.org/

Apart from tools like these I know of several toolkit frameworks like T2K, UIMA, GATE, MorphAdorner, WordHoard etc ... that perform various types of text annotation and analytics including POS Tagging, tokenization, clustering, information extraction, classification, and advanced search and sort.

http://alg.ncsa.uiuc.edu/do/tools/t2k

http://gate.ac.uk/

http://incubator.apache.org/uima/

http://morphadorner.northwestern.edu/

http://wordhoard.northwestern.edu/userman/index.html

HTH,

-- Duane

• Login or register to post comments

Another tool linguists are fond of is R. http://www.r-project.org/

Geoffrey Rockwell, Univesity of Alberta, http://theoreti.ca

• Login or register to post comments

Ishan, thanks for raising these points. I agree completely and have been having similar discussions about tagging digitized texts with a coder friend of mine. It looks like the tool Angela found, Inventoriana, is pretty close to what we need already, too. There are, in any case, numerous implementations of this technology, it just needs application to manuscripts and journal articles.

I hadn't considered that ancient languages would be easier for the machines to handle, but it makes perfect sense.

By the way, I love last.fm. Here's my library.

• Login or register to post comments

First of all - I think tagged texts would absolutely be useful for scholars and non-scholars alike. I would encourage tagging in the original language, though, if only because a word like the Latin vis has so many English equivalents. Sticking to the original tongue would reduce searching problems introduced by synonyms. Or perhaps two "layers" of tags - one in the original language for those who know it and one in a modern language for those who don't. (In this system, the option of hiding one "layer" of tags would be necessary, I think, to avoid clutter.)

I want to share a few thoughts regarding the usefulness of machine translation with regard to ancient languages. However, before I post, I want to be clear that I'm not against technologies like machine translation - just dubious about their ultimate yield for scholars as well as non-academics at this point in time. I have read over my thoughts & they sound so pessimistic! In actuality, that's not how I would describe my attitude toward the technology. My concerns are based on the fact that there is only so much money put toward textual research tools - and I'm not convinced that machine translation for ancient tongues will produce benefits equal to the costs of development.

My comments below are based on my own use of a variety of modern machine translation programs (free & shareware). Also, I can only speak about the specifics of Latin, Greek, & Old English, so perhaps things are different in Sanskrit other ancient tongues. Preemptive thanks to Amy Oh & Daniel Abosso (also Ph.D. students at the University of Illinois) for exchanging emails with me on this subject. Many of their thoughts are also represented here.

1. Ancient "prescriptive" grammars: If you limit Latin/Greek grammar to the forms & style learned in the first year of study, then perhaps I would agree that the grammar is - at this point - prescriptive enough to be translated easily by machine.

However Latin/Greek each survived for centuries, undergoing myriad diachronic changes along the way. The corpus of Latin/Greek literature is far from complete, and much of the untranslated material includes text written by people who themselves did not know or heed the rules of "classical" Latin grammar to perfection. They used idiosyncratic spelling (significant in an inflected language), anomolous tenses, creative structure. What would a translation program do with "bad" Latin such as that written by the marginally literate lower classes on curse tablets or Latin corrupted by centuries of manuscript transmission or the quasi-Latin intoned by certain Carolingian priests?

Compare the "creative" case usage in the formula rattled off by an unlearned priest: Baptizo te in nomine patria, et filia, et Spiritus Sancta with the correct version: Baptizo te in nomine patris et filii et spiritus sancti.

In my experience, small errors can completely break even a modern language translator (e.g. wrong gender of German adjective).

Anyway, the untranslated stuff out there tends to be imperfect Latin - even Latin mixed with early Romance borrowings/grammar. The language was constantly changing. I don't know as much about how Greek changed after the Koine period (though I imagine Byzantine Greek introduced many changes). Perhaps Greek after Koine enjoys a stability that Latin does not, since it (relatively) seamlessly segues into modern Greek, while colloquial Latin goes to seed, eventually producing the Romance languages. But I am just speculating here - a linguist would know more.

It's probably important to remember that untranslated texts are untranslated for a reason. There haven't been loads of new texts discovered recently. For whatever reason - unpopular genre, bad grammar, corrupted transmission, incomplete, difficult palaeography, etc. - they were neglected or spurned by some of the most productive scholars known in the field. This reason alone tends to exclude texts adhering to perfect Latin/Greek grammar from the list of literature needing translation. There are of course exceptions to what I've just said - but I think remembering the sociology of academic work does help when considering what texts one might wish to feed into a machine translation program.

2. Genre: Genre may make all the difference in whether a machine translation program can handle a text. Most poetic texts would be quite difficult to do well, I think, especially later poetry. The big epics have been translated, so I'm not sure what use machine translation would be there. Nuances of poetic texts (political allusions, underlying sexual connotations, etc.) would be completely lost. History/hagiography - possible, depending on the level of innovation and corruption. Same for some types of dialogues and sermons. Rhetorical prefaces? These are difficult even for scholars to translate into English - those of certain medieval & Renaissance authors are almost opaque because of their overcomplicated artistry.

And so on & so forth. Perhaps within a limited number of genres, machine translation could be useful, but not across all. Generic differences must be taken into account.

3. Reliability: It was pointed out to me that machine translation programs for modern languages exist to help people with tongues that could otherwise be learned through aural/oral means. This can't be done with ancient languages. There will always be some means of double checking the output of a translation program for a modern language like French. There exists no such way to reliably check the output for Latin/Greek - texts have been studied for centuries without consensus on interpretation/translation. So it is my humble opinion that using machine translation to decipher untranslated texts risks completely misconstruing the text if there is no paradigmatic translation existing. And if there is, then why use machine translation?

Surely, the risk of misconstruing the text exists even for machine translation of modern languages. And I'm sorry to say that, given the results I've seen for machine-translated German, French, Spanish, and Russian, I can't image the Latin/Greek results will be very intelligible at this point in time. Perhaps one day, but we're quite far off from that point.

Perhaps machine translation could work for inscriptions, since those are very limited in genre and scope, and tend to adhere to more basic rules of grammar. Generally abbreviations used in inscriptions are fairly standard too, which is not the case in medieval palaeography. (I don't see automatic transcription of Latin manuscripts - unless such transcription were limited to a few consistent hands or to a specific scriptorium where abbreviations were standardized) in the near future because of the multiplicity of abbreviations for meanings and vice versa. There's just too much variation.

4. Audience: Who would use machine translation programs for ancient languages? At the risk of sounding dogmatic, a classicist should never use such a program for the purposes of scholarship. For the reasons above, to do so would be irresponsible. I venture to say that machine translation often would give much impression of, for lack of a better phrase, "what text X is about." So, the program is really left for particular kinds of historians & archaeologists (and other related fields where the ancient languages may be one-year or so requirement). [Note: I don't mean to imply that all historians/archaeologists are trained without much concern for languages - but it must be acknowledged that some are.] And who would they rely on for accurate translations? Other people who have had one year of Latin?

Or is this being suggested for non-scholars? I'm not even sure a student reading Tacitus or Aristotle would be able to use a machine translation to check his/her own work...

Okay, I'm ready for the inevitable onslaught!

• Login or register to post comments

I think it would be fairly straight-forward to set up a proof of concept of this with some number of the many free transcriptions already available online. While the overlay of tags and commentaries on high-res images of manuscripts would have to wait, a purely textual version should be easy to do. Any suggestions for some of the first texts for the experiment?

That's what, in essence, this Canterbury Tales website attempts, except that it's not dynamic and collaborative. It's clearly been a labor-intensive work, but it seems fairly static and not as expansive as it could be. Librarius has 2007 as their copyright. I wonder if they'd be willing to open source the work they've already done.

I am going to research the necessary software to implement something like this, but with more open participation, and then host it on the web somewhere, either on my own domain or, if I can, on HASTAC or some place similar.

• Login or register to post comments

Hi Michael and Cathy,

You might take a look at the Collaborative Rubaiyat, an installation of our eComma annotation engine that we put together as part of an exhibition at the Harry Ransom Center here at UT Austin.

It presents multiple editions of FitzGerald's translation in a simple diff-like format, allows free tagging (and here, for another example), threaded commentary, word cloud searches, etc.

We're still working on adding functionality and simplifying the installation and setup process, but we're planning to release the software under the AGPL later this semester.

• Login or register to post comments

Angela, I immediately thought of Mappamundi, too. I attended a lecture Prof. Heng gave last night, in fact, in which she discussed the vision for the future of Mappamundi. Project HESTIA sounds like a natural partner. Even though it's classical rather than medieval, the concerns for mapping and visualization are so close that it would seem a shame if there weren't, at the very least, some mutual awareness between the projects.

Further, the question of how to represent a different conception of space is a fascinating one. We can't just overlay a Google world map with medieval maps and routes if we want to attempt to capture not only the "real" geography, but also the medieval experience of that geopgraphy. Maps raise a new set of questions about how to represent them as digital texts, questions I think we've just barely touched on so far in this discussion.

• Login or register to post comments