Published in: Literary and
Linguistic Computing, Vol. 17, No. 1, 2002. © Oxford
University Press.
A previous version was presented at What’s
all the Hype in Hypertext About? A Humanities Computing Colloquium.
University College Dublin, 10-11 March 2000.
In all fields of education, the computer has come to stay. Unfortunately, much of the hype in educational information technology is due to a narrow focus on technical, material, administrative and service aspects, while the introduction of innovative scholarly methods in curricula are less often taken into account. The fact that students are surfing the web seems to be a cause for enthusiasm for many educators and software developers, but it is questionable if this development is leading toward any appreciable increase in humanities students' use of dedicated humanities software (such as, for instance, concordancing packages) compared to, say, ten or twenty years ago.
In the 1970s and 1980s, humanities scholars interested in computing usually learned a programming language quite thoroughly. They first used punch cards as a tedious input medium, later monochrome character-based displays. Today's humanities students have access to machines with memories roughly a thousand times bigger and equipped with colour displays, but they do not necessarily have better skills in programming or data analysis to tackle relevant research questions in their fields. Still, at several universities, new courses are being introduced that teach the new technologies.
Many universities have embarked on integrating web-based courses into the curriculum. Dedicated web servers, such as WebCT, support web-based courses by offering prepackaged course material, providing preprogrammed tests, tracking student progress, allowing students to write notes to the material, and so on. Rarely, however, do these tools incorporate computational support for the creative analysis of a scholarly problem. Also, traditional teaching in humanities subjects is increasingly being supplemented by discussion groups on the web and so-called MOOs or multi-user domains which allow interaction in a virtual space. Also here, the focus is mostly on tools which may be useful for communication but which are not supporting specific scholarly methods relevant to the subject matter.
On the one hand, it seems that many universities' strategic priorities on computing in the humanities are concentrated on packaging and delivery of traditional teaching materials. Information and communication technologies thus act as a gift wrapper, with colorful multimedia and hypermedia, often in an attempt to make education more efficient and more in line with current fashion. There is of course nothing wrong with new ideas on how a subject is to be taught, as long as they start from the question, what is to be taught, rather than the reverse. Furthermore, the question, what is to be taught, should be approached in the light of a reflection on why we teach and learn humanities.
On the other hand, academic staff involved in humanities computing are increasingly promoting curriculum innovation. As a consequence, we can fortunately observe ample occasions where academic staff engages in reflections and discussions on the design of humanities computing studies. These reflections and discussions are evident in scholarly publications and at scholarly meetings, particularly those of the ACH-ALLC, often through the presentation of concrete plans for new degrees in humanities computing (Hockey et al., 1997, 2001; Unsworth and Butler, 2001; Nowviskie and Unsworth, 1999).
From September 1996 through October 2000, a consortium of European universities carried out a SOCRATES/ERASMUS thematic network project called Advanced Computing in the Humanities (ACO*HUM). Although limited in terms of which humanities subjects were covered in depth, the project addressed far-reaching questions such as the following (Orlandi et al., 1999):
In that context, it is not superflous to consider the various disciplines related to humanities computing. Being an interdisciplinary concept by definition, humanities computing is not exclusively a matter of the humanities, although most of the relevant teaching are situated there. But even the humanities do not represent a clearly delineated group of disciplines. The extent of humanities as a collective term for a group of disciplines is one that varies somewhat with the different academic and cultural traditions in different places, as well as with the various names given to the faculties and schools.
A case in point is linguistics, which at many European universities is usually situated in humanities faculties, close to literature, and indeed often at the same department, while at other universities, notably American ones, linguistics is often situated outside the humanities. Another aspect to be considered is the place of computer science, which cannot be ignored, as it provides important premises, and indeed advances, for humanities computing. At some institutions, computational linguistics is carried out in computer science departments, while at other institutions, computational linguistics is situated in linguistics and literature departments, at the heart of humanities faculties. A survey carried out by ACO*HUM (De Smedt et al., 1999, pp. 97-102, 138-154) brought these and other important differences between various European institutions to light.
What then are the computational methods that students in the various humanities disciplines learn? In linguistic studies, some computational methods in curricula consist of formal linguistic modeling (parsers, formal grammars, etc.) while others deal with quantitative empirical studies (corpus linguistics). In literary and textual studies, some scholarly methods involving computational approaches are text encoding, edition philology and statistical approaches in formal stylistics. Students in some history departments learn to use data analysis methods for designing and mining electronic archives. Scholars in history of art learn to work with, for instance, image recognition and classification.
Although this brief list of examples is only illustrative, it does raise some questions. Given that the different humanities each have their own objects of study as well as their own goals and perspectives, we can ask: is there a single canon of computing methods that every humanities student must learn? On the one hand, experience has proven that it is a pointless effort to teach computing to a humanities student without integrating it into the student's domain of expertise. On the other hand, it is an equally pointless effort to teach computing methods in any discipline without an understanding of what computing itself is all about. The question then is, how far should the teaching of humanities computing differ from the teaching of computer science? At this point, it is useful to distinguish between various categories of humanities computing. These could be tentatively given the following names:
Humanities Computer Literacy. A large number of courses at European universities are dedicated to the provision of basic computational skills for Humanities students. These are usually geared towards specific disciplinary needs: For instance, a student of Russian needs to know how to write, display and print Cyrillic; this is even today not an altogether trivial task. As long as the use of computers is related to skills only, they do not influence the way in which scientific results are gained. At this level we are simply talking about the application of tools.
Humanities Computer Applications. A much smaller number of courses, and a substantial number of research projects, use strongly computer science-based methods (like database technology, applied to an information analysis of some specific problem area). Alternatively, they use computation intensive methods (like statistics) to gain scientific results, which could not be gained without the tools employed. At this level, therefore, we talk about the application of methods.
Humanities Computer Science. An even smaller number of courses and projects, finally, deal with the study and the development of computational methods themselves, aiming at their improved understanding. It might be expected that their results are relevant for a wide range of disciplines.
One current program offered at Groningen, for instance, is strongly based on formal methods in logic, language and cognition. Its curriculum includes natural language processing and language technology, corpus linguistics, logic and logic programming (Prolog), formal and mathematical methods, information analysis, cognitive modeling and neural networks, and statistics. By way of contrast, a quite different program in humanities computing offered at Bergen focuses on cultural aspects of information and communication technology. Its curriculum is geared towards historical and technical developments as well as social consequences of information and communication technology, use of the Internet, multimedia and hypermedia, a cultural and critical perspective, digital culture and computer games, practical computer use and programming, and statistical and quantitative methods.
It would be presumptuous to say that one of these two approaches is better than the other. It is significant, however, that the overlap between these instances of humanities computing programs is small and that the two programs cater to rather different types of knowledge and skills. Although in name not discipline-specific, they nevertheless differ clearly as to the profiles of their student audiences. While the Groningen program is strongly geared to students of language and logic, the one at Bergen seems more appropriate for students interested in media, culture, and aesthetics.
Of course, it is good to have a certain variety between programs at different institutions and in different countries, since variety provides richness. Specializations are, however, mostly useful at advanced levels of study. At an introductory level, widely different approaches with strong biases tend to form niches and may discourage student mobility. One would expect that programs in humanities computing at the bachelors and even taught masters levels would aim at developing some common foundation of knowledge that is applicable in a wide range of humanities disciplines.
The need for a balance between humanities-wide methods and their discipline-bound applications is explicitly recognized in the following two examples of planned study programs. The MA in Digital Humanities which is planned at the University of Virginia (Unsworth and Butler 2001) aims at providing students with experience in recognizing and articulating problems in humanities computing and working collaboratively to solve them. The design of the degree is motivated by the convergence of many aspects of our cultural heritage in the digital melting pot, and will prepare students to meet the cultural need of managing this migration. Although the motivation for the program is truly humanities-wide, the course of study includes discipline-specific electives that provide each student with in-depth course work in the student's particular subject area.
Another new program is the MA in Humanities Computing at Alberta, which emphasizes interdisciplinarity and cooperation rather than niche formation. The masters degree is offered with a large range of specializations: Applied Linguistics, Arts and Design, Chinese Literature, Classics, Comparative Literature, Drama, East Asian Studies, English, French, German, History, Italian, Japanese Literature, Latin American Studies, Linguistics, Music, Philosophy, Political Science, Religious Studies, Russian, Spanish, and Ukrainian. This shows not only that the term 'humanities' is very broadly read by the thirteen participating departments, but it could offer a real opportunity for intense interdisciplinary cooperation (Unsworth and Butler, 2001).
Another problem is that teaching of some computational methods, notably statistics and quantitative methods, seem to lag significantly behind the research advances made in relevant fields. Some instances that could be mentioned are the fields of stylistics and dialectology, where many members of the teaching staff are still staunchly resisting the adoption of computational methods. Nevertheless, great progress has recently been made in applying quantitative computational techniques in these fields, not only in terms of research results, but also in visualizing these results in insightful ways that can benefit students (e.g. Nerbonne et al., 1999).
On a similar note, it has to be mentioned that advances in computation need not be a burden on teaching in the traditional disciplines, but can be a help in practical instruction. A case in point is the well-known computer program Tarski's World (Barwise & Etchemendy, 1993) which was awarded with the 1997 Educom medal. The program allows the student in philosophy and logic to construct and test logical sentences relating to a world represented by simulated 3D objects on a screen. By doing this kind of exercise, students learn how to write well-formed sentences in the language of logic and to determine the truth of such sentences. Barwise & Etchemendy argue that it is advantageous for the student to quickly identify and correct his or her own misconstruals of the language, rather than waiting for classroom help or, as happens all too commonly, completing the entire course without the problem being noticed.
Despite the deceptive simplicity of Tarski's World, the game-like nature of the simulation allows a flexibility which is hardly possible to achieve in a book. Students can evaluate given sentences with respect to any given world. Conversely, students can construe a world that makes a set of sentences come true. In addition, students can make infinitely many new sentences to express facts about a world, and test these. The creative nature of the exercises is found enjoyable by students and stimulates the understanding of logical reasoning. Briefly, here we have a program which is not simply a replica of a book with some multimedia giftwrap. This learning tool is truly powerful: it has an underlying engine that embodies the laws of logic itself. Computation, in this case, contributes to making students of philosophy and logic understand how logic works by means of a simulation game.
As far as linguistics is concerned, similar developments are closing the gap somewhat between traditional linguistics curricula and computational linguistics. Among the available CL tools usable in education are several systems for the automatic analysis (or parsing) of sentences (De Smedt et al., 1999). Such tools not only perform sentence analysis but also presents the sentence structures visually in graphical ways which linguists are used to, including representations in the form of hierarchical tree structures and feature-value matrices. The educational benefits are considerable compared to students doing grammar on paper. Without the need to develop programming skills, students using these tools quickly learn the relation between a sentence and its syntactic structure as defined by a grammar. It is therefore hardly surprising that such tools are not only used by computational linguists, but have found their way into general linguistics curricula.
If various scholarly tools become more refined, accessible and user friendly, it would not be unreasonable to imagine that, in a foreseeable future, literature students will routinely use on-line concordances and automated keyword extraction to verify an analysis of the role of single women in 18th century French literature; that history students will, without hesitation, use on-line archives and censuses to tackle an assignment about the relation between age, marital status and profession in 19th century Norway; and that students in history of art will simply use automated comparative visual search techniques to track the borrowing of motives across national boundaries in 17th century painting (e.g. Vaughan, 1997). Will humanities computing then slowly seep into the humanities disciplines to the point where it is no longer recognized as something special?
The use of tools takes place on two levels. First, at the survival level, beginning scholars learn to identify the formal requirements for their fields of interest. They are quickly confronted with an increasing amount of information that cannot be searched or retrieved other than by using tools which inherently refer to formal methods. Second, at the basic level, they have to identify and understand the co-incidence of tools and problems to be solved by them, in that order. As an added value, they learn which tools should be applied to specific kinds of problems, what skills are additionally needed and how to acquire them.
At an advanced level, the application of methods requires an abstraction of the material of the investigation as well as of the questions to be answered. Formal procedures provide for the deductive explanation of underlying structures and processes which go beyond the treatment of individual cases. The added value of formalization at this level is a deeper, more abstract understanding of the field, in which the application of a method to a problem is mediated by a formal analysis of the problem.
For instance, the student of computational linguistics has to go beyond the level of the general linguistics student who needs to learn the relation between a sentence and its structure. If the computational linguistics student also knows how the method for establishing such a relation works, then an insight into underlying structures and processes is in order, and this in turn creates further requirements for pedagogical approaches. For example, a visual representation of the step-by-step growing search space of a top-down parser referring to a left-recursive rule can help the student understand the limitations of that particular parsing mechanism (Black et al., 1999).
At an expert level, the investigation is no longer application oriented, but adds value by developing new methods for the explanation of human activities in some way. Usually this will take the form of new models, the workings of which need new algorithms. This development of new methods provides an added value by affecting the other levels, leading to a spiral of progress in the understanding of the field.
Although the development of new methods should be left to the experts, we should still consider if knowledge at the other two levels needs to be taught only to a small group of dedicated humanities computing students, or if it should be incorporated in all humanities curricula. My very formulation of this question may suggest that I would prefer the last option, and indeed I do, although I think it would be a point for discussion as to how far one can go in this.
In any case, it is a legitimate question, in how far humanities computing is a scientific field or a technology. Various study programs reflect different positions on this question. The postgraduate MA in Electronic Communication and Publishing, based in the School of Library, Archive and Information Studies at University College London, has taken in students since 1997 (Hockey et al., 2001). The program is quite successful in attracting students and delivering graduates to employers in the publishing industries. Although the program's theoretical component bears a clear relation to humanities, the practical instruction mainly covers relevant IT systems to create and manage web sites. The MA, therefore, seems to represent in the first place a technical and professional study rather than a scholarly discipline. It would therefore fall outside of the humanities not so much due to its subject matter but due to its approach.
In contrast, the long established M.Phil in History and Computing at the University of Glasgow is a one-year taught postgraduate course providing training in the application of computer-based methods to substantive problems in history (Hockey et al., 2001). Instruction in this course emphasises historical interpretation as much as methods and techniques. Here, the focus is on how to use the computer as a scholarly tool for analysis and modeling in order to enhance historical interpretation and contribute to key historiographical debates. Technical matters are subordinate to methodological ones. The M.Phil does not deliver IT professionals but rather humanities scholars.
This brings to the forefront another aspect of the debate, namely, on the one hand, how far the integration of computing in humanities education has a positive effect on employability of graduates, and on the other hand, whether this implies that the humanities are on the road from liberal studies towards professional studies. While the former seems to be a case and clearly would be a benefit, the latter does not seem a necessary consequence and is unlikely to receive unequivocal support. Universities still mostly pride themselves as providing an academic rather than a professional education. While computing tends to enhance the applicability of approaches, the new methods are equally important for making purely academic advances in the field.
The telescope was invented in 1608 and was initially thought useful in war. Galileo obtained one, improved it a little and used it to challenge existing ideas about the solar system. Although a magnificent new technology in itself, the telescope was hardly a scientific tool until Galileo used it to create new knowledge. Today, if we wish to give humanities a credible future, we cannot be content to teach students to use computers merely to access and communicate existing knowledge. Those humanities computing programs who take a purely instrumental approach to computing miss the spirit of Galileo.
New knowledge is being produced in traditional humanities subjects using a variety of computational methods that were not necessarily conceived for the humanities. But now, humanities scholars use, for instance, statistical techniques that for a long time were predominant in the social sciences, learning models that have their roots in psychology, neurology and artificial intelligence, and Monte Carlo modelling that was initially used to predict the fate of neutrons moving through uranium. Advances in humanities computing will be measured by the degree in which we succeed in stimulating students to make creative use of computational methods in order to discover new knowledge, not merely by the amount of web pages that students read and write.
Still, it is necessary to consider both what and how humanities computing should be taught. Even at an introductory level, textbooks are hardly effective tools for learning humanities computing beyond the level of absorbing facts. Students need hands-on experience, experimentation and exercise. In an ideal world, students would have access to the same scholarly tools as researchers in their disciplines. Thanks to the web, this is, in fact, technically possible today, but practical and organizational problems remain.
For one thing, curricula are changing too slowly in what are perhaps the most paper-based areas of study. Unless we want humanities computing to remain a niche, computing methods must be explicitly recognized as a way to transform traditional humanities disciplines. Furthermore, it must be recognized that rapid developments in humanities computing require a conscious effort to update the competencies of the teaching staff. In other words, the teachers have to be taught, through training and retraining programs consisting, for instance, of short courses and workshops, conference attendance, on-line forums and helpdesks, and access to on-line pedagogical resources and documentation.
Supporting actions that target these logistical needs are indeed worth mentioning. Especially in the UK, there has been a long tradition of support centres for humanities computing, in particular the Arts and Humanities Data Service (AHDS) and the various organizations that it incorporates. An example of a newer, international and network-based initiative is the Joint European Website for Education in Language and Speech (JEWELS) started by ELSNET in 2000. JEWELS is intended as a permanent, growing catalog of courses and educational resources and tools, partly supervised by editors.
As more and more study programs in humanities computing are created, the reflection and discussion on useful designs only increases. The ACO*HUM project has perhaps been the first major international project entirely dedicated to investigating the status quo, the needs and opportunities in humanities computing education. As a direct result of stimuli from this project, the ALLC, ACH, ELSNET, EACL and other organizations have already taken an interest in relevant educational matters. It is to be hoped not only that more efforts will be undertaken in the near future, but also that higher education institutions themselves will take the consequences to heart and prepare to implement useful strategies.
Black, W.J., Hill, S. and Kassaei, M. (1999) Intranet learning tools for NLP. In Rosner, M. (Ed.) Computer and Internet Supported Education in Language and Speech Technology. Proceedings of the Workshop, EACL '99, Bergen, 12 June 1999.
JEWELS. Joint European Website for Education in Language and Speech. Available: http://www.hltcentral.org/jewels
De Smedt, K., Black, W., Van den Bosch, A., Lavid López, J.,
Mc Kevitt, P. and Way, A. (1999a). European studies on
computational linguistics. In De Smedt, K., Gardiner, H., Ore,
E., Orlandi, T., Souillot, J. & Vaughan, W. (eds.) Computing
in humanities education: A European perspective. University
of Bergen: HIT centre, pp. 89-154.
De Smedt, K. (1999) Introduction. In De Smedt, K., Gardiner, H., Ore, E., Orlandi, T., Souillot, J. and Vaughan, W. (eds.) (1999b). Computing in humanities education: A European perspective. Bergen: University of Bergen, HIT centre, pp. 1-12.
Hockey, S., Short, H., Ore, E., McCarthy, W. and Rockwell, G. (1997). ALLC special session: Humanities computing in the graduate curriculum. ACH-ALLC'97, Queen's University, Kingston, Ontario, 3 June 1997. Available: http://www.qucis.queensu.ca/achallc97/papers/allcsession.html
Hockey, S., Brown, I., and Spaeth, D. (2001) MA Programmes for Humanities Computing and Digital Media. Session, ACH-ALLC2001, New York University, 13-16 June 2001. Available:http://www.nyu.edu/its/humanities/ach_allc2001/papers/hockey/
Nerbonne, J., Heeringa, W. and Kleiweg, P. (1999). Edit Distance and Dialect Proximity. In: Sankoff, D. and Kruskal, J. (eds.) Time Warps, String Edits and Macromolecules: The Theory and Practice of Sequence Comparison. Stanford, CA: CSLI Publications, pp. v-xv.
Nowviskie, B. and Unsworth, J. (eds.) (1999) Is humanities computing an academic discipline? An interdisciplinary seminar. University of Virginia, Autumn 1999. Available: http://jefferson.village.virginia.edu/hcs
Ore, E., Short, H., Ballam, A., Broady, D., Burnard, L., Burr, E., Lee, S., Opas, L. L., and Rommel, T. (1999). European studies on textual scholarship and humanities computing. In De Smedt, K., Gardiner, H., Ore, E., Orlandi, T., Souillot, J. and Vaughan, W. (eds.) Computing in humanities education: A European perspective. University of Bergen: HIT centre, pp. 63-88.
Orlandi, T., Burnard, L., Buzzetti, D., De Smedt, K., Kropac, I., Souillot, J. and Thaller, M. (1999). European studies on formal methods in the humanities. In De Smedt, K., Gardiner, H., Ore, E., Orlandi, T., Souillot, J. and Vaughan, W. (eds.) Computing in humanities education: A European perspective. University of Bergen: HIT centre, pp. 13-62.
Unsworth, J. and Butler, T. (2001) A Masters Degree in Digital Humanities at the University of Virginia. Session, ACH-ALLC2001, New York University, 13-16 June 2001. Available: http://www.nyu.edu/its/humanities/ach_allc2001/papers/unsworth1/
Vaughan, W. (1997). Computergestützte Bildrecherche und Bildanalyse.
In Kohle, H. (ed.) Kunstgeschichte Digital. Berlin: Dietrich Reimer,
pp. 97-106.