Norse-Icelandic Skaldic poetry of the Scandinavian middle ages: an electronic edition 11/9/01 11:24 AM Norse-Icelandic Skaldic poetry of the Scandinavian Middle Ages: An electronic edition Margaret Clunies Ross: Introduction I am one of five General Editors of a project to re-edit, in hard copy and electronic format, the corpus of Old Norse-Icelandic skaldic poetry from the Scandinavian Middle Ages. Besides myself, Margaret Clunies Ross, from the University of Sydney, my colleagues are Professors Kari Gade, Indiana University and Edith Marold, University of Kiel, Dr. Guñrún Nordal, University of Iceland and Dr. Diana Whaley, University of Newcastle on Tyne. We also have an international advisory committee comprising the Directors of the two Arnamagnaean Institutes in Iceland and Denmark, Professor Vésteinn Ólason and Mr. Peter Springborg, Professor Roberta Frank of Yale University, Professor Kevin Kiernan, from the University of Kentucky, and Dr. Peter Robinson, of de Montfort University. Tarrin Wills is an ARC-funded Research Associate on the project, with special duties in connection with the electronic edition. Several of the editors have obtained, or are in process of obtaining, research grants to produce this edition. I have myself had ARC funding for 5 years now, and wait to hear whether I have been successful in obtaining another grant. The Australian Academy of the Humanities has adopted the project for endorsement by the Union académique internationale (UAI) in Brussels, and the UAI has this year accepted the project as no. 60 in its list of supported projects. There are now 41 scholars from 10 countries (Australia, Canada, Denmark, Germany, Iceland, New Zealand, Russia, Sweden, United Kingdom and the United States) involved in the project as contributing editors of parts of the skaldic corpus. Our aim is to complete the project by 2007. All editors have now received their allocated verses or poems and have submitted preliminary samples of their work to one of the five General Editors in order to ensure that proper procedures have been followed. The General Editors issued an Editors’ Manual in July 2000, whose purpose is to ensure that all contributing editors conform to the same methodology and procedures when preparing their editions. What is this project about and why is it important? In the European Middle Ages, most European cultural groups learnt to commit their literature and culture to writing using the Roman alphabet for the most part and inscribing their vernacular texts upon prepared animal skins. Medieval Icelandic writings are among the richest medieval European vernacular written records and the most numerous, even though unfortunately they are not as well known as they should be, given their significance to European cultural history. Many people are aware of medieval Icelandic saga literature; fewer probably know that Icelanders (and Norwegians) produced a large amount of a complex poetry that we call skaldic after the Old Norse word for a poet, skáld. In origin, skaldic poetry seems to have been an elite oral court poetry, practised at the courts of regional kings and earls in ninth-century Norway. The first skalds were almost certainly Norwegian, but, after the settlement of Iceland in c. 870-930, Icelanders gradually assumed a special role as court poets and practitioners of the skaldic art. Although skaldic verse continued to be composed in a courtly context, its uses in Iceland (where there were no courts) extended to non-courtly contexts, and, after the conversion of the Scandinavians to Christianity, skaldic poetry became a vehicle for Christian sentiments and doctrinal issues that in other medieval cultures were expressed in prose or through different verse conventions. Skaldic poetry had a 500-year history and survived a major religious and cultural change. Aside from its centrality and longevity in Scandinavia, skaldic poetry is of potential interest to scholars from many disciplines: medieval literature, orality-literacy studies, linguistics, poetics, myth studies, history, and http://skaldic.arts.usyd.edu.au/papers/drrh2001.html Page 1 of 6 Norse-Icelandic Skaldic poetry of the Scandinavian middle ages: an electronic edition 11/9/01 11:24 AM the history of religions, to name only a few. Until now, access to this corpus has been restricted to a small number of scholars of Old Norse, who could read both the medieval language and modern Scandinavian languages. One of the major aims of the editors of this new edition is to make skaldic poetry accessible to more scholars and students. We intend to produce an English translation of our texts, together with clear notes and references to the poetry’s social and cultural contexts of production and recording. As Tarrin will explain, most of the poetry that has come down to us is embedded in prose texts. It has taken the General Editors several years to work out the methodology for this project. We are now at a stage when editors will be working on their editions, and we hope that conferences such as this will enable us to get the best possible input into our working methods. Tarrin Wills: The electronic edition The aim of the project is to produce an electronic edition which can form the basis of both a printed and electronic text. The project is similar in some ways to other projects which present medieval texts in an electronic form: there will be the manuscripts; transcriptions of them; and an edited text with textual and critical apparatus. It differs from other projects in that most projects either cover a number of texts in a single manuscript (or do not concern themselves much with original documents); or they present a single text with multiple originals. The present project, however, will present a great many texts from a great many manuscripts. The scope of the project consequently raises some important issues. I wish to discusss some of these issues in this paper, as it is possible that they have some relevance to other projects or future projects. These issues relate to the organisation of the project and to the encoding of material related to the interpretation of the poetry. Problems Organisational issues present a few problems for the project. These relate mainly to its scope. The corpus consists of more than 5000 stanzas, with each stanza usually consisting of 8 lines. These verses are mostly found within larger prose works, and are in most cases separated from each other (as lausavísur or ‘loose verses’) by stretches of prose. The verses are contained in more than 370 manuscripts. Usually, each verse is found in a prose text which has multiple witnesses. The group of verses in a poem or prose text will consequently have its own stemma (transmission history). Locating all the stanzas in all the manuscripts is one of the main problems. Some of this work was done in the 1950s by assistants to Prof. Jón Helgason, a prominent scholar in the field, and most of this has already been entered into a database. The manuscripts are held in more than 15 institutions, although most are located in collections in Reykjavík, Copenhagen and Stockholm. Most of these institutions do not have facilities for digital photography, and those that do are only in the early stages of digitising their collections (so far, fewer than ten manuscripts in the present project have been digitised, mostly by the National Library of Iceland and the Arnamagnæan Institute in Reykjavík). The majority of manuscripts have been photographed and printed on black and white film. There is also a problem in that few of the institutions have electronic catalogues of their collections. Perhaps most importantly, the project is divided between 41 editors. This is necessary because of the scope and difficulty of the material being edited. Obviously, not all of the experts in skaldic poetry have much knowledge or interest in producing material in an electronic form. Whatever system is put in place for producing the edition, it cannot rely on all editors to produce their material electronically; at the same time, the material produced by editors must be transferrable to a manipulable electronic form without interfering in the http://skaldic.arts.usyd.edu.au/papers/drrh2001.html Page 2 of 6 Norse-Icelandic Skaldic poetry of the Scandinavian middle ages: an electronic edition 11/9/01 11:24 AM editorial process. The interpretation of the material poses a number of problems which affect the production of an electronic edition. Skaldic poetry is a highly complex and difficult verse form: [Example from Hallfreñar saga] Highly complex metrics: skaldic verse usually incorporates rhyme and alliteration. The most common form, dróttkvætt, has one or two groups of four lines, with usually six syllables and always three stresses in each line. Odd lines have internal half-rhymes on two of the stressed syllables including the third, and even lines have full rhymes in the same distribution. Each pair of lines has alliteration on three stressed syllables, usually with two alliterating syllables in the odd and one in the even lines. A highly complex word-order is to some extent demanded by the metrics. Groups of words are usually broken up within half-stanzas, with kennings and other groupings separated over two or more lines. Complex poetic diction ( kenningar and heiti): kennings constitute a system of reference to certain entities by nominal periphrasis. For instance, a boat may be called a horse of the sea, a horse being a mode of transportation, and a boat being a mode of transportation on the sea. Kennings can contain other kennings within them: thus a raven might be called the seagull of the din of shields, where din of shields is a battle, and the seagull of battle is a raven. Other mythological/legendary references: certain things may be alluded to by reference to certain myths. A common example is to call poetry dwarves’ drink, from a Scandinavian myth where poetry is created in the form of mead by dwarves. The complex word-order is of course linked closely to the metrics; however, it tends to separate various interpretative elements such as the kenningar which contain at least two parts. The verse word-order makes the marking up of such elements virtually impossible, and so a prose word order must be established. It is the amount and types of interpretative material which poses the biggest challenge for the electronic compilation of the edition. For each verse, there is the following textual and interpretive material: Verse text in lines Textual apparatus Annotations to the text Translation with explanation of kennings & other references Prose word order with kennings tagged Diplomatic texts MS references Edition references Account of the prose context General notes on the verse (Possibly) Metrical tagging Solutions for organisational issues As mentioned previously, probably the most difficult aspect of the project is the number of editors working on it. Because so much interpretative material is required, most of the work on the edition must be performed by experts in the editing and interpretation of skaldic poetry. In order to make the electronic edition useful, it must http://skaldic.arts.usyd.edu.au/papers/drrh2001.html Page 3 of 6 Norse-Icelandic Skaldic poetry of the Scandinavian middle ages: an electronic edition 11/9/01 11:24 AM go beyond the requirements of a printed edition — material to be manipulated in any way must be machinereadable, and that means being marked-up in consistent and unambiguous ways. There is an important issue here: how are we to make the most of the traditional editorial and interpretative skill and experience of scholars who nevertheless have little skill in producing electronic texts or experience in contributing to electronic projects? The possibilities of electronic media are exciting for many of us, but it often requires of editors to re-think important parts of the editing process. Many editors, however, have been producing editions for printed media for many years. For this project, most of that which needs to be machine-readable is a range of concrete and implied references: the textual apparatus and annotations cross-referenced with the text; a range of bibliographic and manuscript references; and the references of the kenning system, some of which are embedded inside each other. All of this requires editors to be very diligent in the way they produce their texts. Traditional editions allow for a great deal of flexibility in the way they reference such information, but such referencing is often not easily translatable into machine-readable form. For example, in the textual apparatus of a print edition, one could express certain sorts of features of the text (some examples which have arisen already: ‘ætt: 'æ' apparently corrected from 'a' 61, eitt 53, ort M’; ‘... 756 (seig in the margin) ...’; these can be encoded in TEI, but this requires some interpretation on the part of the encoder). To be put in machine-readable form, however, it must be standardised. This requires some intervention in the editorial process. This should be avoided: it is impossible in an edition of this scale to check every non-standard feature which has to be altered for the mark-up. It is obviously far more desirable to identify potential problems and make editors aware of them beforehand. Even with such guidelines, editors often just proceed how they always have. The process of convincing them to change their habits has been slow. It has been achieved to some extent by showing the editors the possibilities of a machine-readable text, which seems to have worked in making them realise the potential of producing the texts in an electronically manipulable form. I produced the following document for a symposium in May: [Example: Ragnarsdrápa stanza 3] Another major problem is that we are reliant on manuscript collections which do not provide their collections or catalogues digitally. The National Library of Iceland and the Arnamagnæan Institute in Reykjavík have digitial photography facilities, and we are currently developing a list of manuscripts to be scanned by them. Other manuscripts are only available digitally through scans of high-quality, albeit black-and-white, photographs. Solutions for interpretative material In order to deal with the various interpretative layers, the mark-up must be separated into two parts: firstly, textual and metrical analysis which is based on the verse word order; secondly, analysis of kennings and translation which is based on the prose word order. The process is shown by the following diagram: [Editing process] Each part of the process is included, something not possible with print editions. Manuscripts are transcribed from the originals, and then, on the basis of a stemma, an edited text is produced with variant readings in the apparatus. This text is normalised. It forms the basis also of the annotations and any metrical information. The text is then re-ordered into a putative prose order, which can then form the basis of the analysis of kennings and the translation. The link between the verse and prose orderings is the hardest to encode, as each contains information embedded as part of its structure; it could, however, be done on a word-by-word basis, or phrasehttp://skaldic.arts.usyd.edu.au/papers/drrh2001.html Page 4 of 6 Norse-Icelandic Skaldic poetry of the Scandinavian middle ages: an electronic edition 11/9/01 11:24 AM by-phrase basis. Alternatively, the link between the orderings could be left in a state which is not machinereadable. Such an option, however, would not allow for analysis and/or searches to be of both orderings. I am fairly confident that we can automate the re-ordering so that the link between prose and verse ordering is maintained. Almost all of the features listed above can be encoded using the guidelines of the Text Encoding Initiative. The two main types of data which cannot be encoded are stemmas and some features of diction, particularly kennings. Stemmas are basically hierarchical trees representing the ‘family’ relationship of extant and putative manuscripts. As XML and SGML have basically the same structure of parent/child/sibling relationships, stemmas can easily be encoded in those languages: [Stemma in XML] Kennings One of the aims of the project is to produce reference material, among which is a searchable index of kennings. Because kennings are often embedded within other kennings, and because they have implicit referents, they need to be marked up appropriately. If it is done properly, the index will include all parts of kennings as well as whole extended kennings. Fortunately, kennings have a fixed structure, with two elements (base-word and determinant) and an implicit referent; either element can be the referent of another kenning. Such a structure is superbly suited to a hierarchical mark-up such as in XML. [Example of kenning markup] < k e n n i n g r e f e <bw>stave</bw> <bw>icicle</bw> <det>axe-</det><bw>tumult</bw> </kenning> </det> </kenning> </det> </kenning> Of course, some kennings do not fit so neatly into this structure, but they can be accomodated in other, albeit less elegant, ways. Also, the XML elements introduced here are not part of any standardised XML application. However, similar features of diction are found in other literary forms and the inclusion of such elements in TEI is desirable. Getting editors to produce such mark-up is difficult in most cases. I am therefore working on a web-based script to allow such mark-up to be produced automatically: [Kenning script] Editors will eventually be able to mark up kennings and other complex structures by using such web interfaces. Conclusions http://skaldic.arts.usyd.edu.au/papers/drrh2001.html Page 5 of 6 Norse-Icelandic Skaldic poetry of the Scandinavian middle ages: an electronic edition 11/9/01 11:24 AM Understanding skaldic verse requires a great deal of interpretative material, for which we are reliant on experts in this type of verse. The editing of such verse involves quite complex manipulations of the texts in question. If it is to be manipulable in an electronic way, it must be encoded with all the referencing systems in a machinereadable form. Such encoding requires both some new XML solutions, and the co-operation of the editors. If it is done properly, the possibilities are enormous: not only a highly usable, detailed and interactive edition, but also an image library of all the sources; and reference material on the diction, lexicon and metrics of this important but complex poetic form. [Return to skaldic homepage] © Tarrin Wills and editors 2000, 2001 http://skaldic.arts.usyd.edu.au/papers/drrh2001.html Page 6 of 6