Medieval texts - Contemporary media. The art and science of editing in the digital age

2833Saibene001-154DEF copia 13:Layout 1 27/10/09 11:51 Pagina 1

Medieval texts – Contemporary media:The art and science of editing in the digital age



Edited by Maria Grazia Saibene and Marina Buzzoni


Cover illustration: Sandro Botticelli, La primavera, 1482 (particular, Mercury extending hiscaduceus to stop the wind).

Published with the contribution ofthe Italian Ministry of Education, University and Research

(PRIN-06: “Composizione, trasmissione e instabilità del testo germanico medievale:nuovi critri ecdotici e modelli di edizione”).

© Ibis, Como – Pavia 2009www.ibisedizioni.itFirst edition: September 2009ISBN 978-88-7164-283-3


9 Preface

Medieval texts – Cotemporary media: The art and science of editing in the digital age

13 Matthew James Driscoll, Marking-up abbreviations in Old Norse-Icelandicmanuscripts

35 Marina Buzzoni, «Uuarth thuo the hêlago gêst that barn an ira bôsma»:Towards a scholarly electronic edition of the Hêliand

57 Odd Einar Haugen, An Apology for the text that never was: Reconstructing TheKing’s Mirror

81 Giuseppe Brunetti, Old English poetry on the Web: A lexicographic edition93 Paul Gabriele Weston, Digital Text Archives: A librarian’s approach

113 Maria Grazia Saibene, The Wanderer. Text, Intratext, Intertext: Editing OldEnglish Elegies

131 Roberto Rosselli Del Turco, Il Progetto Vercelli Book Digitale: codifica e visua-lizzazione di un’edizione diplomatica grazie alle norme TEI P5


Preface

The essays collected in this volume began life as Workshop papers. TheWorkshop, which took place in Pavia in June 2008, explored the interactionbetween Information and Communication Technology (ICT) systems and thephilological analysis of medieval texts. Turning the papers into book essays hasbeen a long and challenging process and we are grateful to the contributors fortheir expertise, patience and constant support in pursuing a common goal.

The result of this joint undertaking is this volume which focuses on the cur-rent updates in the preparation of digital editions of medieval texts. Three essays(Buzzoni, Haugen, Saibene) investigate from the philological point of viewboth the theoretical and practical issues encountered in the creation of digitaltexts. It is generally agreed that handwritten documents demand an editioncapable of taking into account the peculiar character of each witness. It is notsimply a matter of transcribing texts electronically; each witness within a multi-faceted textual tradition must be individualized. Two articles (Driscoll, RosselliDel Turco) address specific mark-up problems connected respectively with themechanisms of encoding abbreviations and digitizing manuscript images.Brunetti’s contribution provides an overview of his lexicographic edition of OldEnglish poems on the Web, while Weston’s paper investigates the potentialitiesof new electronic tools for conversion into digital form in the light of long-termpreservation of a cultural heritage.

Throughout the book the theoretical implications for both present andfuture research are taken into account, and the current e-edition projects hereillustrated offer an international perspective of editorial scholarship in the elec-tronic medium, providing also original contributions in the field of digitalphilology.

Major problems linked to the material production of scholarly editions arealso dealt with, directly and indirectly: for example, the use of TEI-conformantXML to facilitate electronic document exchange, the problems involved inobtaining high-quality images of manuscripts and, last but not least, the highcost of digitization, as well as the all too frequent lack of professional computingcentres capable of providing technical support and website maintenance in thefield of the humanities.

The collection is, we believe, coherent and, not surprisingly, many essaysinteract. It was not however our primary objective to achieve standardization of

9


these diversified contributions; the differences between them opportunelyreflect the peculiarities of each and every specific area of research.

The 2008 Workshop was attended by many enthusiastic and passionatelydevoted doctoral students. It is to them that we dedicate this book in the hopethat in this field they will in the future give evidence of the same enthusiasm andintellectual energy.

Pavia, June 2009

Maria Grazia Saibene and Marina Buzzoni


MATTHEW JAMES DRISCOLLArnamagnæanske Samling, Nordisk Forskningsinstitut, Københavns Universitet

MARKING-UP ABBREVIATIONSIN OLD NORSE-ICELANDIC MANUSCRIPTS

Abstract. Although it is often presented as a fairly straightforward matter, theexpansion of abbreviations involves a good deal of interpretation, not to say guess-work, on the part of the editor. In fact, there are many cases where it may be far fromcertain how an abbreviation should be expanded. Generally, where justification isgiven at all, abbreviations are said to have been expanded as ‘a service to the reader’.As with other services, one would like at the very least be aware when one is beingdone one, and also have the option of declining. The mechanisms for encodingabbreviations and their expansions described in this essay, employing TEI-confor-mant XML, allow one, for the first time, to do precisely that.

1. Introduction

The use of abbreviations is a characteristic feature of medieval Latin manu-scripts and those of most European vernacular traditions.1 The practice, whichwas intended both to spare the scribe the labour of writing words which, due totheir frequency generally or in a particular text, could easily be understood in anabbreviated form, and in order to save parchment and ink, derives from antiquity.In Roman times there were three systems of abbreviation in use: the notæ juris,which were used extensively in legal documents, the Tironian notæ, a system ofshorthand signs developed by Cicero’s secretary Tiro, and the nomina sacra, con-tracted forms of the words for ‘God’ and the name ‘Jesus Christ’, a practice bor-rowed by the early Christians from Hebrew. The use of abbreviations in Latin

13

1 On abbreviations in antiquity and the middle ages generally see B. BISCHOFF,Paläographie des römischen Altertums und des abendländischen Mittelalters, Berlin, Schmidt, 1979,pp. 202-223. The standard work on Latin abbreviations remains A. CAPPELLI, Lexicon abbre-viaturarum: Dizionario di abbreviature latine ed italiane, Milano, Hoepli, 19616.


manuscripts increased until the 12th century, after which it began to fall off. Latinpractice was taken over more or less wholesale in manuscripts written in the ver-nacular languages, but in general the use of abbreviations was never as great in thevernacular as it had been in Latin. An exception to this are Old Norse-Icelandicmanuscripts, which both in terms of frequency and variety of abbreviationsexceed even Latin practice. There were also several Icelandic innovations, such asthe use of small capitals and dotted letters to indicate geminate consonants.2

Abbreviations are customarily divided, with varying degrees of granularity,3

into a number of types, generally with regard to the means through which theabbreviation is achieved. The four basic types are:

1. Suspension, where only the first letter or letters of a word are written, gen-erally followed, and frequently also preceded, by a point or, occasionally,with a superscript stroke.

2. Contraction, where the first and last letters are written, normally with asuperscript stroke, or, less commonly, a point or points.

3. Superscript letters, a superscript vowel normally representing that vowelpreceded by r or v, a superscript consonant that consonant preceded by a.

4. Special signs (‘tittles’) or brevigraphs, sometimes originating in a letter orcombination of letters but no longer recognisable as such.

Taken to a slightly higher level of abstraction, however, abbreviations may besaid to fall into two distinct groups: in the first, either part of the word is writtenout and the rest omitted, the omission often, but not always, being indicated bysome sign or mark, or a special sign is used to represent a particular lexical item;in the second, a superscript letter or sign is used instead of a specific combina-tion of graphemes.4 Abbreviations of the first type, comprising the suspensions,

14

MATTHEW JAMES DRISCOLL

2 On the use of abbreviations in Icelandic manuscripts see e.g. HREINN BENEDIKTSSON,Early Icelandic script as illustrated in vernacular texts from the twelfth and thirteenth centuries,Reykjavík, Handritastofnun, 1965, pp. 85-94. A useful survey of abbreviations in OldNorse-Icelandic manuscripts is found in O.E. HAUGEN, Paleografi, in O.E. Haugen (ed.),Handbok i norrøn filologi, Bergen, Fagbokforlaget, 2004, pp. 175-214 (207-210).

3 L. A. CHASSANT, Dictionnaire des abréviations latines et françaises, Paris, Jules Martin,18845, distinguishes between eight types, KR. KÅLUND, Palæografisk atlas, Ny serie: Oldnorsk-islandske skriftprøver c. 1300-1700, København, Gyldendal, 1907, between seven and A.CAPPELLI, Dizionario…, between six. The four-fold division presented here derives fromHREINN BENEDIKTSSON, Early Icelandic script…, p. 85.

4 The numerous examples where the n in the combination ng is abbreviated through theuse of the nasal stroke over the preceding vowel, e.g. «þīg» (=þing, ‘assembly’), or one half of


contractions and some of the brevigraphs, may thus be said to have a specificlexical reference, and are in practice restricted to a fairly limited number of lexi-cal items; those of the second, comprising the superscript letters and tittles andthe remainder of the brevigraphs, have a specific graphemic reference, i.e. repre-sent the same combination of graphemes regardless of the lexical item in whichthey occur.

One of the most common signs of abbreviation is the superscript stroke orbar, which can be used both to indicate the suppression of one or more nasalconsonants (m or n) and as a more general mark of abbreviation in suspensionsand contractions. Although in appearance there is no discernible differencebetween the two signs, in terms of their function they are quite distinct. The let-ter h with a stroke («ħ»), for example, represents the word hann (‘he’), but herethe stroke cannot be said to ‘stand for’ the characters ann in the same way as itstands for n when the same word is abbreviated «han-», where the stroke func-tions in the same way as the 2-shaped sign which always stands for the combina-tion ur, wherever it appears. The stroke in «ħ», on the other hand, can representother letters in other inflectional forms of the same word, for example onu in thedative singular honum (‘him’), abbreviated «ħm», and an in the genitive hans(‘his’), abbreviated «ħs». Finally, although less common, the barred-h can alsorepresent forms of the feminine pronoun hón (‘she’), «ħi», for example, for henni(‘her’ dat.sing.) or «ħar» for hennar (‘her’ gen.sing.). The barred-h and similarcharacters can thus be said to function as brevigraphs, since they can representvarious inflectional forms of the same or related words.

It is customary in traditional scholarly printed editions to expand – orresolve, as the practice is sometime called – abbreviations, i.e. to supply the let-ters which have been omitted, or represented through other means, by thescribe. In some cases this is done silently, but in more diplomatic editions thesupplied letters are generally typographically distinct from the others, for exam-ple rendered in italics or placed within round brackets.5 Owing to the limita-tions of the printed page, however, this method can only ever succeed in mak-

15

Marking-up abbreviations…

a diphthong is abbreviated, e.g. «t++utt» (=trautt, ‘loath’) makes it quite clear that they refer tographemes, rather than phonemes.

5 In the editions of Old Norse-Icelandic texts produced under the auspices of theArnamagnæan Institutes in Copenhagen and Reykjavík abbreviations are generally expand-ed silently, with the exception of suspensions using a point, which are given in round brack-ets. The reasoning behind this is that there is greater degree of uncertainty as to the preciseform intended with suspensions than with other types of abbreviation.


ing reasonably clear what has been supplied by the editor, while the actual formfound in the manuscript must in many cases be left to the reader’s imagination.

2. Marking-up abbreviations in electronic texts

The Text Encoding Initiative’s Guidelines for Electronic Text Encoding andInterchange 6 provide mechanisms for marking up abbreviations and their expan-sions, allowing one both to register what is actually in the manuscript and topresent this in a form more palatable to the modern reader. In previous releasesof the Guidelines two elements were available for this purpose:

<abbr> (abbreviation) contains an abbreviation of any sort.<expan> (expansion) contains the expansion of an abbreviation.

Although these two elements are, like <sic> and <corr>, regarded as mirrorimages of each other, so-called «Janus tags», they did not in fact come into beingsimultaneously. In the earliest version of the Guidelines, P1, released in 1990, theelement for tagging abbreviations was called <abbrev>. There was no correspon-ding element for tagging expansions, but <abbrev> carried an optional @fullattribute, which one could use to give the expanded form of the abbreviation, aswell as a @type attribute, which could be used to «classif[y] the abbreviationusing terms such as title, initials, acronym, degree». The example given is:

<propname type=person><abbrev type=title>Dr.</abbrev><abbrev type=initials>M.</abbrev> Deegan</propname> is theResearch Officer of the <abbrev full=’Computers inTeaching Initiative’ type=acronym>CTI</abbrev>Centre for Literature and Linguistic Studies.

It seems clear both from this example, the only one given, and from the listof suggested values for the @type attribute, that the authors of the Guidelineswere not at this point thinking of the sort of abbreviations found in ancient,medieval and early-modern manuscripts and inscriptions, but rather those com-monly used in modern languages.

The <expan> element was introduced in TEI P2 (1992). At the same time<abbrev> was shortened to <abbr> (in agreement, possibly coincidentally, with

16


6 The TEI Guidelines are available electronically online at <http://www.tei-c.org/Guidelines/P5/> (accessed 20 December 2008).


HTML, itself then just coming into being) and the @full attribute replaced by@expan. The same example – which is still found in the Guidelines – nowappeared:

<abbr type=title>Dr.</abbr> <abbr type=initial>M.</abbr>Deegan is the Director of the <abbr expan=’Computers inTeaching Initiative’ type=acronym>CTI</abbr> Centre forTextual Studies.

There is, in addition, another example, one taken from a medieval manu-script:

Ex<abbr type=brevigraph expan=’per’ Resp=PG>&per;</abbr>ience, thogh noon auctoritee

To illustrate the Janus-like nature of the pair the same example is also givenfavouring <expan> over <abbr>:

Ex<expan type=brevigraph abbr=’&per;’Resp=PG>per</expan>ience, thogh noon auctoritee

Thus, from the very beginning, there were two fundamentally different waysin which these two elements could be used, depending, on the one hand, onwhether one regarded the entire word as constituting ‘the abbreviation’ (and itsexpansion), e.g. «Dr.» = «Doctor», or, on the other, whether one perceived ‘theabbreviation’ as consisting only of the mark or sign which indicated the suppres-sion of one or more letters in the word, and, correspondingly, ‘the expansion’ asthe letters supplied in the process of expansion, e.g. the symbol «p-» = «per». Bothhave always been possible, and both widely practised.

To return to our previous example, «ħ», for hann (using the Unicode characterU+0304 for the bar or «combining macron»7), could always be marked up either:

<abbr>h̄</abbr> // <expan>hann</expan>

or

h<abbr>̄</abbr> // h<expan>ann</expan>

17


7 I am following here the recommendations of the Medieval Unicode Font Initiative; see<http://www.mufi.org> (accessed 20 December 2008). An alternative to using Unicodecharacters directly in this way would be to employ the new <g> element (for gaiji, theJapanese term for a non-standardized character or glyph); see chapter 5 of the TEI Guidelines,«Representation of Non-standard Characters and Glyphs».


Recognising this discrepancy in usage as a potential problem, two further ele-ments were introduced in TEI P5, officially released on 1 November 2007:<am> (for ‘abbreviation marker’), which «contains a sequence of letters or signspresent in an abbreviation which are omitted or replaced in the expanded formof the abbreviation», and <ex> (for ‘editorial expansion’), which «contains asequence of letters added by an editor or transcriber when expanding an abbre-viation». At the same time the Guidelines now also recommend that use of<abbr> and <expan> be restricted to «the whole of an abbreviated [or expand-ed] phrase or word».8 This allows for a much clearer, and more nuanced, mark-up of abbreviations and their expansions, especially when these elements arecombined with <choice>, also an innovation in P5, the purpose of which is toserve as a wrapper for alternative encodings of the same textual feature.9 Onecan now, for example, do the following:

<choice><abbr>h<am>̄</am></abbr><expan>h<ex>ann</ex></expan>

</choice>

Here it is made completely explicit what the form of the abbreviation is,which part of it is omitted or replaced when it is expanded and what letters aresupplied in the process of expansion. The addition of these new elements doesnot entirely solve the original problem, however, in that one can also just as eas-ily do the following:

h<choice><am>̄</am><ex>ann</ex></choice>

Or, indeed, the following:

<abbr>h<choice><am>̄</am><ex>ann</ex></choice></abbr>

This mark-up, it could be argued, offers the best value for money in terms ofregistering all the relevant information with a minimum of verbosity. It would

18


8 TEI Guidelines, section 3.5.5; cf. section 11.3.2: «The content of the <abbr> elementshould usually include the whole of the abbreviated word, while the <expan> element shouldinclude the whole of its expansion.» Although the wording is slightly ambiguous – what ismeant by «the whole of its expansion» – the meaning should be clear enough: both <abbr>and <expan> should be used to tag whole words or phrases.

9 See section 3.4 of the Guidelines, «Simple Editorial Changes». The <choice> elementwas introduced in preference to the so-called ‘Janus-tag’ mechanism referred to above, large-ly as a consequence of the abolition of attributes containing plain text.


10 There are other cases where there appears to be a discrepancy in the way abbreviationswhich involve letters are treated, i.e. whether the letters are perceived as being there or not.When expanding the tironian nota (⁊) or ampersand (&) – a practice for which there seemslittle justification in any case – the entire expansion is treated as supplied, e.g. italicised.When expanding the barred-p, as we saw above, all three letters, per, are generally treated assupplied, even thought the p is as palpably present as is the h in «h-». A less obvious case is thesign resembling the numeral 4 () used to represent the termination –rum (both in Latin andOld Norse-Icelandic), which is in fact a round r with an oblique curve through the leg. Here,too, it is not uncommon when expanding to treat all three letters as supplied, even thoughthe r is arguably already there. One could also argue that superscript letters representingthemselves plus another letter are also actually there and do not need to be supplied.

also allow one in a very simple way to distinguish between different types ofabbreviations by means of the @type attribute, either for statistical analysis or ifthey are to be treated in varying ways in their expanded forms, as is sometimesthe case with suspensions, the expansions of which are placed in round brackets,while other types, if not expanded silently, are italicised.

Many a user of the TEI, not least the novice, will thus still be left wonderinghow best to proceed. Which of the two ways – tagging whole words or parts ofthem – is to be preferred? The most reasonable answer would appear to be thatwhile both have their justification, one or the other may be more appropriate incertain cases. The mark-up employed in this last example works well in this par-ticular case, and would work equally well in many if not most others; clearly notin all cases, however. It seems counterintuitive to treat certain types of abbrevia-tions on anything other than the whole-word level. In English and other lan-guages, for example, «p.» is a common abbreviation for the word «page» (Lat.pagina or the equivalent in other languages), but, as with the bar in «h-», the dotcannot be said to ‘stand for’ the letters age; rather we have an entire word repre-sented by its initial letter – the siglum as it was known in Latin – followed by apoint signalling that this is an abbreviation. The siglum can, moreover, be dou-bled to indicate the plural: «pp.». This was common practice in Old Norse-Icelandic too, particularly in kinship terms, where a point was frequently alsoplaced before as well after the sigla: «.bb.» for bræðr (‘brothers’), «.ss.» for synir(‘sons’) etc. It seems nothing short of perverse to maintain that in such a case thefirst letter somehow really is ‘there’10 whereas the second and attendant dot(s)are not, but are rather represent the suppressed letters. And yet this is preciselywhat we do when we expand such abbreviations: «.ss.», for example, would in aprinted edition typically appear in expanded form as «s(yner)». Here, the newmark-up possibilities offered by P5 would allow us to treat the abbreviation andits expansion in slightly different ways, for example like this:

19



<choice><abbr>.ss.</abbr><expan>s<ex>ynir</ex></expan>

</choice>

rather than:

<choice><abbr><am>.</am>s<am>s.</am></abbr>

<expan>s<ex>ynir</ex></expan></choice>

or:

<abbr><choice>

<am>.</am><ex/>

</choice>s

<choice><am>s.</am><ex>ynir</ex>

</choice></abbr>

At the same time, however, in the case of abbreviations with a graphemic ref-erence, where there is generally a one-to-one correspondence between theabbreviation sign and its expansion, this correspondence is lost, or at leastblurred, if treated on the whole-word basis, not least where there is more thanone abbreviation within a single word. Consider the following mark-up for theword konungarnir (‘the kings’), written «konūgarn»:

<abbr>konu<choice><am>̄</am><ex>n</ex></choice>garn<choice><am>͛</am><ex>ir</ex></choice></abbr>

As opposed to the following:

<choice><abbr>konu<am>̄</am>garn<am>͛</am></abbr><expan>konu<ex>n</ex>garn<ex>ir</ex></expan>

</choice>

20



While in some ways perhaps neater, the latter encoding, in addition to los-ing the connexion between the abbreviation and its expansion,11 also requiresthe repetition of eight characters which are otherwise unchanged. As theexpansion of abbreviations of this type essentially entails the substitution ofone set of characters for another it is arguably more logical to place within<choice> only those characters which are involved in the substitution, and notthose which remain the same—a choice between two identical options is nochoice.

3. A typology of abbreviations

In what follows I should like to propose a typology of Old Norse-Icelandicabbreviations and suggest ways of encoding them in light of the discussionabove. As with the TEI generally, there may be wrong ways of encoding things,but there is no single right way, so those presented here should not be taken asprescriptive. Where appropriate I give examples of both the ‘basic’ encodingusing the elements <abbr> and <expan> wrapped in <choice> and the alterna-tive encoding suggested above.

A. Abbreviations with a lexical reference

A.1. Suspensions

A.1.1. The first letter or letters of a word are written out and the remainderomitted, this omission being indicated by means of a point (or colon) set after,or both before and after, the letter or letters. This method was used both for anumber of generally common words, k. for konungr (‘king’), d. for drottinn(‘lord’) or dóttir (‘daughter’), s. for sonr (‘son’) or segir/sagði (‘says/said’) etc., andalso for words, especially proper names, which appear repeatedly in a text andare thus only understandable in context. Whole phrases, for example legal for-mulae, if frequently repeated, could also be represented by the first letters ofeach word.

21


11 One could, of course, make the connexion explicit by using the global linking element@corresp.


A.1.2. The first letter or letters are written on the line and the remainderomitted, a superscript stroke, sometimes with a curl, indicating the omission.This form of abbreviation is largely restricted to certain common words.

22


k. k(onungr)‘king’

<choice><abbr type=”A.1.1”>k<am>.</am></abbr><expan>k<ex>onungr</ex></expan>

</choice><abbr type=”A.1.1”>k<choice><am>.</am><ex>onungr</ex></choice></abbr>

Ol. Ol(afr)‘Ólafr’

<choice><abbr type=”A.1.1”>Ol<am>.</am></abbr><expan>Ol<ex>afr</ex></expan>

</choice><abbr type=”A.1.1”>Ol<choice><am>.</am><ex>afr</ex></choice></abbr>

ħ hann‘he’

<choice><abbr type="A.1.2">h<am>̅</am></abbr><expan>h<ex>ann</ex></expan>

</choice><abbr type="A.1.2">h<choice><am>̅</am><ex>ann</ex></choice></abbr>

ҟ skal‘shall’

<choice><abbr type="A.1.2">sk<am>̅</am></abbr><expan>sk<ex>al</ex></expan>

</choice><abbr type="A.1.2">sk<choice><am>̅</am><ex>al</ex></choice></abbr>

þ– þeim‘them’

<choice><abbr type="A.1.2">þ<am>̲</am></abbr><expan>þ<ex>eim</ex></expan>

</choice><abbr type="A.1.2">þ<choice><am>̲</am><ex>eim</ex></choice></abbr>

The letter thorn (þ) with a stroke through the descender was used in manu-scripts from the mid-13th century onwards to represent þeir (‘they’), or occasion-ally þess (‘its’) but came increasingly to be used for þeim (‘them’).


Particularly in younger manuscripts, the word þess (‘its’) is frequently abbre-viated using a double stroke through the ascender. This is probably a develop-ment from the ligatured contractions mentioned below.

A.1.3. Only the initial letter is written on the line, and the letter immediatelyfollowing it is written superscript. This form of suspension is essentially alsorestricted to certain common words, in particular the two shown here.Superscript letters used in this way should be distinguished from those whichrefer to a specific combination of letters, discussed below in section B.1.2.

A.2. Contractions

A.2.1. The initial and final letters of a word are written on the line, but someor all of the intervening letters are omitted, the omission being indicated bymeans of a superscript stroke or bar, either straight or with a curl, which is setover short letters and generally passes through the ascenders of any tall letters.Points are also possible – following, preceding and, very rarely, between the let-ters – but are less common.

23


titil‘to’

<choice><abbr type="A.1.3">t<am>ͥ</am></abbr><expan>t<ex>il</ex></expan>

</choice><abbr type="A.1.3">t<choice><am>ͥ</am><ex>il</ex></choice></abbr>

hohon‘she’

<choice><abbr type="A.1.3">h<am>ͦ</am></abbr><expan>h<ex>on</ex></expan>

</choice><abbr type="A.1.2">h<choice><am>ͦ</am><ex>on</ex></choice></abbr>

þ–– þess‘its’

<choice><abbr type="A.1.2">þ<am>̿</am></abbr><expan>þ<ex>ess</ex></expan>

</choice><abbr type="A.1.2">þ<choice><am>̿</am><ex>ess</ex></choice></abbr>


Note that in some cases the letters supplied in the process of expansion comebefore, or both before and after, the abbreviation mark. In these cases there willnot be a one-to-one correspondence between the <am> and <ex> elements, andso the alternative mark-up in which these two are wrapped in <choice> tags isless appropriate.

Certain genitive forms, in particular konungs (‘king’s’), hans (‘his’) and þess(‘its’), are sometimes abbreviated in such a way that the first and last letters arecombined as ligatures (k plus tall s etc.), with a bar to indicate that they are con-tractions:

A.2.2. The initial letter is written on the line and the final letter is writtensuperscript (a practice still common with ordinal numbers, «1st», «2nd» and soon). This type is especially common with forms of the word maðr (‘man’) andthe verbs fara (‘go’), hafa (‘have’), taka (‘take’) and vera (‘be’).

24


ħm honum‘him’

<choice><abbr type="A.2.1">h<am>̅</am>m</abbr><expan>h<ex>onu</ex>m</expan>

</choice><abbr type="A.1.2">h<choice><am>̅</am><ex>onu</ex></choice>m</abbr>

l– skal‘shall’

<choice><abbr type="A.2.1">s<am>̅</am>l</abbr><expan>s<ex>ka</ex>l</expan>

</choice><abbr type="A.1.2">s<choice><am>̅</am><ex>ka</ex></choice>l</abbr>

ml–i mælti‘spoke’

<choice><abbr type="A.2.1">ml<am>̅</am>i</abbr><expan>m<ex>æ</ex>l<ex>t</ex>i</expan>

</choice>

kk–ia kirkia‘church’

<choice><abbr type="A.2.1">kk<am>̅</am>ia</abbr><expan>k<ex>ir</ex>kia</expan>

</choice>

ƙ– konungs‘king’s’

<choice><abbr type="A.2.1"><am></am></abbr><expan>k<ex>onung</ex>s</expan>

</choice>


A.3. Brevigraphs with a specific lexical reference

A.3.1. The nomina sacra, i.e. Ihc for Jesus and Xpc for Christus. Originally thenomina sacra are contractions: in Ihc for Jesus, for example, the Latin letters reflectthe form, but not the phonological value, of the original Greek uncials IHC(=IES). Other inflectional forms are also found: Ihm for Jesum, Ihu for Jesu etc.

A.3.2. The Tironian nota for et, probably the most common of all abbrevia-tions (although entirely absent from some early manuscripts); it had variousforms, the earliest resembling a 7.

A.3.3. Runic letters. The f- and m-runes are occasionally used to represent thewords fé (‘cattle, wealth’) and maðr (‘man’), their names in the runic alphabet.The latter is frequently found with a superscript (Latin) letter indicating aninflectional ending, i for manni (‘man’ dat.sing.), a for manna (‘man’ gen.pl.) etc.;this corresponds entirely to contractions mentioned in section A.2.2.

25


Jħc Iesus‘Jesus’

<choice><abbr type="A.1.2"><am>Jh̅c</am></abbr><expan><ex>Iesus</ex></expan>

</choice>

⁊ ok‘and’

<choice><abbr type="A.3"><am>⁊</am></abbr><expan><ex>ok</ex></expan>

</choice>

uouoro‘[they] were’

<choice><abbr type="A.2.2">u<am>ͦ</am></abbr><expan>u<ex>oro</ex></expan>

</choice><abbr type="A.2.2">u<choice><am>ͦ</am><ex>oro</ex></choice></abbr>

mamanna‘men’(gen.pl.)

<choice><abbr type="A.2.2">m<am>ͣ</am></abbr><expan>m<ex>anna</ex></expan>

</choice><abbr type="A.2.2">m<choice><am>ͣ</am><ex>anna</ex></choice></abbr>


B. Abbreviations with a specific graphemic reference

B.1. Supralinear letters and signs (tittles)

B.1.1. Supralinear signs (tittles). These are normally written over another char-acter, or sometimes, depending on their shape, slightly to the right. They cannotoccur word-initially or, for the most part, in pairs.

The nasal stroke, indicating the suppression of one or more nasal consonants;it can occur both medially and finally.

The pi- or omega-like sign (originally a superscript a), representing ra or, lessfrequently, va, ja or ar.

26


ⲯ maðr‘man’

<choice><abbr type="A.3"><am>ᛘ</am></abbr><expan><ex>mađr</ex></expan></choice>

han– hann‘he’

<choice><abbr type="B.1.1">han<am>̅</am></abbr><expan>han<ex>n</ex></expan>

</choice><abbr type="B.1.1">han<choice><am>̅</am><ex>n</ex></choice></abbr>

þīg þing‘assembly’

<choice><abbr type="B.1.1">þi<am>̅</am>g</abbr><expan>þi<ex>n</ex>g</expan>

</choice><abbr type="B.1.1">þi<choice><am>̅</am><ex>n</ex></choice>g</abbr>

fra‘from’

<choice><abbr type="B.1.1">f<am></am></abbr><expan>f<ex>ra</ex>s</expan>

</choice><abbr type="B.1.2">f<choice><am></am><ex>ra</ex></choice></abbr>


The 2-like sign, representing ur (or its mutated form yr); it sometimes resem-bles the infinity symbol (∞).

A sign resembling the numeral 9, used to indicate us (or its mutated form ys);it is principally used finally, in which case it can appear on the line, but can alsobe used medially.

The zigzag-shaped sign, representing er or ir, occasionally also re (esp. withp), eir (esp. in the word þeir ‘they’), or simply r (esp. with e to represent the 3rdperson sing. form of the verb ‘to be’); it is sometimes also used as general markof abbreviation in suspensions, particularly where one of the suppressed lettersis an r.

27


ua uera‘[to] be’

<choice><abbr type="B.1.1">u<am>͛</am>a</abbr><expan>u<ex>er</ex>a</expan>

</choice><abbr type="B.1.1">u<choice><am>͛</am><ex>er</ex></choice>a</abbr>

þra þeirra‘their’

<choice><abbr type="B.1.1">þ<am>͛</am>ra</abbr><expan>þ<ex>eir</ex>ra</expan>

</choice><abbr type="B.1.1">þ<choice><am>͛</am><ex>eir</ex></choice>ra</abbr>

agt fagurt‘beautiful’

<choice><abbr type="B.1.1">fag<am></am>t</abbr><expan>fag<ex>ur</ex>t</expan>

</choice><abbr type="B.1.1">fag<choice><am></am><ex>ur</ex></choice>t</abbr>

magn magnus‘Magnús’

<choice><abbr type="B.1.1">magn<am></am></abbr><expan>magn<ex>us</ex></expan>

</choice><abbr type="B.1.1">magn<choice><am></am><ex>us</ex></choice></abbr>


B.1.2. Supralinear letters. Pretty much any letter, including accented charac-ters and ligatures, could appear superscript as an abbreviation, normally repre-senting itself preceded by r or v if a vowel, or by a if a consonant. Some of themore common ones are listed here:

Superscript a, representing va.

Superscript e, representing re or ve.

Superscript i, representing ri, ir or vi (esp. in the word því, ‘because’).

28


sa sva‘so’

<choice><abbr type="B.1.2">s<am>ͣ</am></abbr><expan>s<ex>va</ex></expan>

</choice><abbr type="B.1.2">s<choice><am>ͣ</am><ex>va</ex></choice></abbr>

depa drepa‘[to] kill’

<choice><abbr type="B.1.2">d<am>ͤ</am>pa</abbr><expan>d<ex>re</ex>pa</expan>

</choice><abbr type="B.1.2">han<choice><am>ͤ</am><ex>re</ex></choice>pa</abbr>

oid ofrid‘strife’ (lit.‘un-peace’)

<choice><abbr type="B.1.2">of<am>ͥ</am>d</abbr><expan>of<ex>ri</ex>d</expan>

</choice><abbr type="B.1.2">of<choice><am>ͥ</am><ex>ri</ex></choice>d</abbr>

þiat þviat‘because’

<choice><abbr type="B.1.2">þ<am>ͥ</am>at</abbr><expan>þ<ex>vi</ex>at</expan>

</choice><abbr type="B.1.2">þ<choice><am>ͥ</am><ex>vi</ex></choice>at</abbr>


Superscript o, representing or or ro.

Superscript c, representing ik(k) or ek(k).

Superscript d or ð, representing eð or uð (the latter especially in the word guð‘god’).

Superscript m, representing um.

29


gc geck‘went’

<choice><abbr type="B.1.2">g<am>ͨ</am>pa</abbr><expan>g<ex>eck</ex></expan>

</choice><abbr type="B.1.2">g<choice><am>ͨ</am><ex>eck</ex></choice></abbr>

md med‘with’

<choice><abbr type="B.1.2">m<am>ͩ</am></abbr><expan>m<ex>ed</ex></expan>

</choice><abbr type="B.1.2">m<choice><am>ͩ</am><ex>ed</ex></choice></abbr>

tignm tignum‘noble’(dat.pl.)

<choice><abbr type="B.1.2">tign<am>ͫ</am></abbr><expan>tign<ex>um</ex></expan>

</choice><abbr type="B.1.2">tign<choice><am>ͫ</am><ex>um</ex></choice></abbr>

bog borg‘fort, castle’

<choice><abbr type="B.1.2">b<am>ͦ</am>g</abbr><expan>b<ex>or</ex>g</expan>

</choice><abbr type="B.1.2">b<choice><am>ͦ</am><ex>or</ex></choice>g</abbr>

ábott ábrott‘away’

<choice><abbr type="B.1.2">áb<am>ͦ</am>tt</abbr><expan>áb<ex>ro</ex>tt</expan>

</choice><abbr type="B.1.2">áb<choice><am>ͦ</am><ex>ro</ex></choice>tt</abbr>


Superscript n, representing in(n) or an(n) (particularly in forms of the encliticdefinite article).

Superscript r, representing ar.

Superscript t, representing at or it, or, later, að or ið.

B.2. Linear brevigraphs

B.2.1. A semicolon-like sign used to represent ed or eð; the later cursive formcan resemble the numeral 3 or the letter z. In Latin writing it could also stand,with b, for the ending -bus and with q for the enclitic -que. Its use in Old Norsemanuscripts is for the most part restricted to the word með (‘with’) although thesign is sometimes also used with the letter s, representing the Latin word sed; ina text otherwise in the vernacular this is normally expanded heldr (‘beit’).

30


ma manna‘men’(gen.pl.)

<choice><abbr type="B.1.2">m<am></am>a</abbr><expan>m<ex>ann</ex>a</expan>

</choice><abbr type="B.1.2">m<choice><am></am><ex>ann</ex></choice>a</abbr>

ur uar‘was’

<choice><abbr type="B.1.2">u<am>ͬ</am></abbr><expan>u<ex>ar</ex></expan>

</choice><abbr type="B.1.2">u<choice><am>ͬ</am><ex>ar</ex></choice></abbr>

dyrt dyrit‘the animal’

<choice><abbr type="B.1.2">dyr<am>ͭ</am></abbr><expan>dyr<ex>it</ex>pa</expan>

</choice><abbr type="B.1.2">dyr<choice><am>ͭ</am><ex>it</ex></choice></abbr>

m med‘with’

<choice><abbr type="B.2.1">m<am></am></abbr><expan>m<ex>ed</ex></expan>

</choice>


B.2.2. A sign resembling an inverted c or sometimes the numeral 9 (in whichcase it is identical to the superscript sign for us) which stands for con and (inLatin) com. In Old Norse, as in Latin, it is found only in initial position. It ismost commonly used in Latin loan words, but can also be found in native wordsas well, in particular kona (‘woman’) and konungr (‘king’).

B.2.3. A sign resembling the numeral 4 (in fact a round r with a oblique curvethrough the leg) used to represent the termination -rum; it is especially commonafter the letters ð/d and o. In Latin this mark could also function as a generalmark of abbreviation, but such use is rare in Old Norse manuscripts.

Note that here, and in the following example, one can treat the abbreviationand the expansion slightly differently – in keeping with the idea that the r in thiscase and the p in one below actually are present and therefore do not need to besupplied – using the first method but not the second, where there must be adirect correspondence between the two.

B.2.4. The letter p with a straight stroke through the descender indicates per,while a curved stroke or flourish extending through the descender represents pro,and p with a superscript bar pre; although occasionally found in native words,these are most commonly used in Latin loan words such as prófeti (‘prophet’).

31


ↄa kona‘woman’

<choice><abbr type="B.2.2"><am>ͻ</am>a</abbr><expan><ex>kon</ex>a</expan>

</choice><abbr type="B.2.2"><choice><am>ͻ</am><ex>kon</ex></choice>a</abbr>

od odrum‘others’(dat.pl.)

<choice><abbr type="B.2.3">od<am></am></abbr><expan>odr<ex>um</ex></expan>

</choice><abbr type="B.2.3">od<choice><am></am><ex>rum</ex></choice></abbr>

<abbr type="B.2.1">m<choice><am></am><ex>ed</ex></choice></abbr>


B.3 Small capitals and dotted letters were used in Icelandic manuscripts toindicate geminate consonants. It is not standard practice to expand small capitals(although some editors have done so), while dotted letters normally are treatedlike other abbreviations.

4. Conclusion

It is, as has been mentioned, customary to expand abbreviations in scholarlyeditions of medieval texts; this can either be done silently, i.e. with no distinc-tion between the letters supplied by the editor and those present in the manu-script, or the supplied letters can be kept distinct from the rest, usually beingrendered in italics or placed within round brackets. While certainly preferable tothe former, the latter practice, in attempting to do two things simultaneously,unfortunately succeeds fully in neither.

Although it is often presented as a fairly straightforward matter, the expan-sion of abbreviations involves a good deal of interpretation, not to say guess-work, on the part of the editor. The common verbs segja (‘say’) and mæla(‘speak’), to take the most obvious example, are frequently represented throughsuspensions, with no indication as to whether the present or preterit forms areintended. Both are usually perfectly possible, and it is not unlikely that, if asked,the scribe wouldn’t have known – or wouldn’t have cared – which form theabbreviation stood for either, but the editor must decide. There are many othercases where it may be far from certain how an abbreviation should be expanded.

32


ar persar‘Persians’

<choice><abbr type="B.2.4"><am></am>sar</abbr><expan>p<ex>er</ex>sar</expan>

</choice><abbr type="B.2.4"><choice><am></am><ex>per</ex></choice>sar</abbr>

hoġr hoggr‘[he] strikes’

<choice><abbr type="B.3">hog<am>̇</am>r</abbr><expan>hog<ex>g</ex>r</expan>

</choice><abbr type="B.3">hog<choice><am>̇</am><ex>g</ex></choice>r</abbr>


It is standard practice when expanding abbreviations to do so in keepingwith the normal orthographic practice of the scribe in question.12 Thus, if thescribe normally uses e to represent the unstressed front vowel /I/, rather than i,the zigzag-shaped tittle ( ) will be expanded er rather than ir. If the scribe usesboth forms, as most scribes do, the editor will normally expand consistentlyusing the form which most frequently occurs when the word is written in full.The situation can thus easily arise where a scribe has written er three times andir twice, but otherwise used the tittle. This would then be expanded, perhapsseveral hundred times, to er, giving an entirely false impression of the distribu-tion of the two forms in the resulting text.

Expanding abbreviations can also result in impossible forms, or at least formsthe scribe would not have used had he written the word in full. To take a simpleexample: from the earliest times the round r () was used after o and, gradually,other round letters, especially d (), ð and þ,13 and it is customary in more diplo-matic editions to reproduce the round form where it occurs. If the scribe writesthe word þeir (‘they’) in abbreviated form using a round r after the þ («»), aswas common, the expanded form would be «þei», which the scribe would nev-er have written, as i is not a round letter. Another example is the word síðan (‘lat-er’), written «si», where a scribe, the better to attach the superscript stroke,might use þ instead of ð (or d), which, if the manuscript was from anytime afterthe mid-13th century,14 is what he would presumably have written had he notchosen to abbreviate. The expanded form «siþan» would, therefore, be mislead-ing at best; the form «siþan» a direct falsification.

Editors have, of course, been aware of this. R. I. Page comments in the intro-duction to his edition of Gibbons saga that the silent expansion of abbreviations(other than suspensions employing a dot) «may give a to some extent falseimpression of a text, a type of falsification not detectable from the present edi-tion». Half a century earlier, in his edition of Rómverjasaga, Rudolf Meissnerchose not to indicate the expansion of the many abbreviations, as they «nur einepalæographische Bedeutung haben», with the exception of the ending ir/er, «weilin der ausgeschriebenen Form ein fast regelloses Schwanken herrscht». To

33


12 Cf. e.g. R.I. PAGE (ed.), Gibbons saga, Editiones Arnamagnæanæ B.2, Copenhagen,Munksgaard, 1960, p. xxxii: «Abbreviated forms are expanded as far as possible in accordancewith the scribe’s normal spelling».

13 HREINN BENEDIKTSSON, Early Icelandic script…, pp. 47-4914 STEFÁN KARLSSON, The development of Latin script II: in Iceland, in O. Bandle et al. (eds.),

The Nordic languages: An international handbook of the history of the North Germanic languages,Berlin/New York, de Gruyter, 2002, pp. 832-840 (835).


choose one or the other of the forms, he says, «würde ein falsches Bild von derOrthographie der hs. geben».15

Generally, where justification is given at all, abbreviations are said to havebeen expanded as ‘a service to the reader’. As with other services, one would likeat the very least be aware when one is being done one, and also have the optionof declining. The mechanisms for encoding abbreviations and their expansionsdescribed here allow one, for the first time, to do precisely that.

34


15 R. MEISSNER (ed.), Rómverjasaga (AM 595, 4°), Berlin, Mayer & Müller, 1910, pp. 2-3.


MARINA BUZZONIUniversità Ca’ Foscari, Venezia

«UUARTH THUO THE HÊLAGO GÊSTTHAT BARN AN IRA BÔSMA»:

TOWARDS A SCHOLARLY ELECTRONIC EDITIONOF THE HÊLIAND

Abstract. This contribution focuses on how the electronic medium can adequate-ly convey the often disregarded differences amongst the witnesses of a ninth-centuryOld Saxon poem known as Hêliand. After dealing with the most relevant peculiaritiesof the manuscript tradition both from the linguistic and the cultural point of view, itwill be shown that the inner mouvance which characterizes this work can be bettercaptured through a «fluid edition». In fact, such an edition will permit users to choosebetween visualizing only one text scenario or several. Technically, what is being builtup is a hyper-textual environment and a hyper-textual way of using the electronicedition, based not on a static but on an interactive model.

1. Introduction

Despite the traditional presentation of the Hêliand as a single work – an ideawhich is both induced and implemented by the most common editions of theOld Saxon poem –,1 both the two major witnesses of the text (mss. M and C)and the fragments (V, P/L, and S)2 differ in many respects. To quote only a fewof them, they differ codicologically, linguistically, as well as geographically.3

35

1 Cf. in particular B. TAEGER, Heliand und Genesis, herausgegeben von O. Behaghel, 10.,überarbeitete Auflage von B. Taeger, Tübingen, Max Niemeyer Verlag, 1996 [19841].

2 As the graphic under 1.1. shows, P and L share some features with C, while S is connect-ed with M (B. TAEGER, Heliand…, pp. XVIII-XXIV).

3 The confusing nature of the Hêliand data has led many scholars to the speculation thatthe manuscript evidence defies conclusive dialect identification.


The aim of this paper is twofold. On the one hand, it focuses on the histori-cal and ideological motivations which are assumed to be at the basis of the afore-mentioned differences, and on the other hand, it explores the possibility of usingan electronic medium to adequately convey them. To achieve this purpose, theeditor will need to do more than add hypertext connectivity to existing modelsof presentation of editorial data. Rather, he/she will need to find new means ofvisualization as a prelude to a much greater challenge: the making of what maybe called a «fluid edition», able to capture the inner mouvance of the Old Saxonpoem.

1.1. Manuscript tradition

The text of the Hêliand – a ninth-century alliterative paraphrase of the life ofChrist –4 has come down to us in a nearly complete form in two manuscripts: M(Monacensis, Cgm. 25, preserved in the Bavarian Staatsbibliothek though comingfrom Bamberg, where it was discovered in 1794) and C (Cotton Caligula A.vii,preserved in the British Library). Four more fragments transmit short passagesof the text: namely V (Codex Palatinus lat. 1447, discovered by K. Zangemeisterin 1894 and now housed in the Vatican Library; it contains lines 1279-1358); P(formerly preserved in the University Library of Prague, now in Berlin at theDeutsches Historisches Museum, PA R 56/2537; it contains lines 958b-1006a), S(the Straubing fragment, currently held in the Bavarian Staatsbibliothek; in poorcondition, only lines 351-360, 368-384, 393-400, 492-582, 675-683, 693-706, 715-722 are preserved) and – last but not least – the newly discovered leaflabelled as L (the Leipzig fragment, found in 2006; it contains lines 5823-5870a).5 The recto side of the leaf is reproduced in Plate 1.

36

MARINA BUZZONI

4 The total amount of lines is nearly 6.000.5 For an overview of the manuscripts – with the exception of L – see B. TAEGER,

Heliand…, pp. XVIII-XXIV. On the S-fragment see: P. SCARDIGLI et al., Un nuovo testimone peril Heliand, «Romanobarbarica», III, 1978, pp. 271-289; B. BISCHOFF, Die StraubingerFragmente einer Heliand-Handschrift, «PBB», CI, 1979, pp. 171-180. On the newly discoveredL-fragment see: H.U. SCHMID, Ein neues ‘Heliand’-Fragment aus der UniversitätsbibliothekLeipzig, «Zeitschrift für deutsches Altertum und deutsche Literatur», CXXXV/3, 2006, pp.309-323; I. RAUCH, The Newly Found Leipzig Heliand Fragment, «Interdisciplinary Journal forGermanic Linguistics and Semiotic Analysis», XI/1, 2006, pp. 1-17; H. SAHM, Neues Licht aufalte Fragen. Die Stellung des Leipziger Fragments in der Überlieferungsgeschichte des ‘Heliand’,«Zeitschrift für deutsche Philologie», CXXVI, 2007, pp. 81-98; H.U. SCHMID, Nochmals zumLeipziger ‘Heliand’-Fragment, «Zeitschrift für deutsches Altertum und deutsche Literatur»,CXXXVI/3, 2007, pp. 376-378.


The considerable number of diverse witnesses which have survived is evi-dence that the effect of the Hêliand in its day may well have been much greaterthan has been acknowledged.6

As for the relationship between the witnesses, most of the scholars date M tothe middle or late ninth century7 and C to the tenth century.8 Both are supposedto derive from an earlier manuscript of which they are later copies. M, which

37

«Uuarth thuo the hêlago gêst that barn an ira bôsma»

Pl. 1: Fragment L, recto side (http://www.ub.uni-leipzig.de)

6 H. SAHM, Neues Licht…, pp. 89-98.7 According to Bischoff M dates back to around 850. B. BISCHOFF, Die Schriftheimat der

Münchener Heliand-Handschrift, «PBB», CI, pp. 160-170.8 Among the first to deal with this topic was R. PRIEBSCH, The Heliand Manuscript Cotton

Caligula A.VII, Oxford, Clarendon Press, 1925. See also N.R. KER, Catalogue of Manuscriptscontaining Anglo-Saxon, Oxford, Clarendon Press, 1957, p. 172.


begins at line 85 and contains many lacunae, has usually been taken as theguide-text since it is very close in time to the alleged composition date (around830),9 and is the one most consistently «Saxon». M has thus always been thebase manuscript in editorial practice. However, some scholars, followingRathofer’s hypothesis,10 prefer the later English witness C on the grounds that itis more faithful to the uncorrected inconsistencies of the dialect of the presumedoriginal, and also because it has better preserved the ancient structural divisionof the Hêliand into single units called fitts. V is derived from one further sourceand thus it represents an autonomous branch of the tree, while as for P/L11 thereis no final unanimity among scholars on its/their origin though the prevalentopinion tends to connect P with C (rather than with M) due to the presence of ashared binding «error».12

The stemma codicum can be illustrated as follows:13

38

MARINA BUZZONI

9 On the thorny issue of the date of composition, mainly based on speculations that restupon external evidence contained in a twofold Latin document printed by Flacius Illyricus in1562 (the Praefatio in librum antiquum lingua Saxonica conscriptum, in prose, and the Versus depoeta et interprete huius codicis, in verse), see B. TAEGER, Heliand…, pp. XXXIII-XXXVIII.

10 J. RATHOFER, Der Heliand: theologischer Sinn als tektonische Form; Vorbereitung undGrundlegung der Interpretation, Cologne, Böhlau, 1962.

11 In all probability L is a leaf originally contained in the P-codex, as the same ruling andductus seem to suggest.

12 V. 980a: hêran heƀencuning M; herran heƀencuning P; herren heƀencuning C.13 Adapted from B. TAEGER, Heliand…, p. XXIV.


39


2. The Annunciation to Mary and the Incarnation

Though M is given a primary role in textual criticism, a closer look at themanuscripts reveals deep discrepancies between M and C. In light of this, andsince their communicative effect in what may be called «their different receptionbackground» appears to be sharply divergent, it is not too forward to considerthem as different «texts».

Take, for example, the developments and implications of the «Annunciationto Mary and the Incarnation» episode described in the Fourth Song of theHêliand. Mary’s consent to Angel Gabriel’s Annunciation appears expanded toinclude the authors’ own constant message – «don’t doubt» – to the allegednewly converted Saxons. While this message is perfectly coherent with thesocial context from which M springs, it sounds odd in C, being the latter a man-uscript which was copied in the South of England in the second half of thetenth century.14

Lines 286-288 read as follows:

uuerđe mi aftar thînun uuordun, al sô is uuilleo sî, / hêrron mînes; nis mi hugi tuîfli, / neuuord ne uuîsa. (IV, 286-288)

«Be it done unto me according to your word, whatever should be the will of mylord; my heart and mind are not in doubt, neither in word nor in deed.»

The Hêliand poets15 then tell of the incarnation, announcing it with an epicintroductory line: So gifragn ik «I have heard it told». Mary’s acceptance of theproposal (mid leohtu hugi «with light-hearted mind», mid gilôƀon gôdun «in goodfaith», mid hluttrun treuun «with transparent loyalty») is followed by a somewhatshocking turn of phrase: Uuarth (thuo) the hêlago gêst that barn an ira bosma (IV,291-292).16 In orthodox theological terms, it is the Second Person of the Trinity,

14 R. PRIEBSCH, The Heliand Manuscript…, pp. 9-11; N.R. KER, Catalogue…, p. 172[n. 137].

15 The hypothesis that the Hêliand is the result of «teamwork» was put forward forciblyand convincingly in H. HAFERLAND, War der Dichter des ‘Heliand’ illiterat?, «Zeitschrift fürdeutsches Altertum und deutsche Literatur», CXXXI, 2002, pp. 20-48. This hypothesis wasagreed upon in M. BUZZONI, Re-writing discourse features: speech acts in Heliand, in The Gardenof Crossing Paths: The Manipulation and Rewriting of Medieval Texts, edited by M. Buzzoni – M.Bampi, Venezia, Cafoscarina, revised edition 2007, pp. 139-161.

16 Contrary to Taeger (see B. TAEGER, Heliand…, p. 17), I wouldn’t insert any commaafter gêst since the insertion leads to a different reading of the sentence, namely «the holy spir-it came, the baby was in her womb», which is too difficult to reconcile with the whole text.


the Son, who becomes man in Jesus Christ, not the third one, the Holy Spirit.Though the authors may have misunderstood the relevant passage of the Creedof Nicea: et incarnatus est de Spiritu Sancto ex Maria Virgine «He was made flesh bythe Holy Spirit», they may also have been misled by their own purpose to estab-lish both the difference and the superiority of Christ to Germanic divinities,«whose origin and birth were entirely within the world from cosmic naturalforces».17 Most of the scholars consider this passage as proof which exhibits theanti-Arian, docetist bias of the poem, further enhanced by the insistence on theidea that Christ is from heaven above, from the «meadows of the sky», and hisorigin lies in the Spirit (i.e. the strength) of God. In other words, Christ is shownto be higher than the highest Germanic divinities. According to Murphy, thevery structural rearrangement of the whole Annunciation scene serves the samepurpose.18 The sequence of events in one of the most important Hêliand sources,the Old High German translation of Tatian’s Diatessaron, is as follows:

1. Annunciation to Mary – the Incarnation;2. Visitation of Mary to Elisabeth;3. Birth of John the Baptist;4. The book of the family tree of Christ;5. Joseph’s doubt and faith.

The structure of the Hêliand has been greatly rearranged and simplified tocreate a perfect symmetry between the figures of Mary and Joseph in order toimplement the idea that faith supersedes doubt:

1. Annunciation to Mary, who reacts with no doubt, but faith;2. Incarnation of the Holy Spirit;3. Annunciation to Joseph, who reacts with doubt, then with faith.

The relevant point here is whether and how the over-orthodox view con-veyed through this rearranged structure is presented in the subsequent parts ofthe two manuscripts.

While M displays one and the same attitude, the Cotton manuscript – copiedwithin the syncretic cultural context provided by the insular background –shows something different. Let us first consider the verbal text and then turn tothe so-called «paratextual» elements.

40

MARINA BUZZONI

17 G.R. MURPHY, The Saxon Savior. The Germanic Transformation of the Gospel in the Ninth-Century Heliand, New York-Oxford, Oxford University Press, 1989, p. 45.

18 Ibidem.


After the incarnation of the Holy Spirit, «almost as though the pregnancy ofMary were too sublimely spiritual for her to notice it, she has to be reminded byFate when the time has come for her to give birth».19 In ms. M the word for theworkings of Fate, giscapu (l. 336), appears along with the modifier godes. In oth-er words, the text presents Fate «simply» as a messenger of God’s will. In ms. C,instead, godes is not found, as if it were useless, and this attaches a slightly lessorthodox flavour to the same episode. Since this reading seems also to break theregularity of the alliterative pattern (to be found only in later poems), it has beensuggested that it should be considered closer in fidelity to the original version –a sort of lectio difficilior. This is similar to what appears in line 619: friđugumonobezt. Tho sprak im eft that folc angegin, where C omits that folc thus breaking thealliterative pattern between the half-lines.

Another and – to my mind – even more convincing evidence of the ‘patina’of antiquity that permeates the C-version of the Hêliand can be found in the lin-guistic layer. I will skip the phonetic and phonological analysis, since a hugeamount of studies have come to the plausible conclusion that C shows manyfranconian dialect features.20 Rather, I will briefly focus on the linguistic varia-tion which is encoded in other parts of the grammar (namely, syntax) and inprosody. In particular, the pronominal system of C points at older uses than theM-version, as shown by the frequency of linguistic phenomena like the doublyfilled complementizer (es. l. 298b: He afsof that that siu habda barn undar iru lit.«He could see that that she had a child in her womb»),21 the lack of an explicitpronoun either in anaphoric or expletive contexts (ll. 92b-93: gern uuas he suîđo,/ that he ø thurh ferhtan hugi fremmean môsti «He was very happy and carried [it]out with a devout mind»; l. 3138b: ‘gôd is ø hêr te uuesanne […]’ «This is a goodplace to live»),22 the so-called «case attraction» in the relative clauses (ll. 267b-268: Endi ni kumid, / thes uuîdon rîkeas giuuand, thes he giuualdan scal «No end will

41


19 Ivi, p. 46.20 See, among others, T. KLEIN, Phonetik und Phonologie, Graphetik und Graphemik des

Altniederdeutschen (Altsächsischen), in Sprachgeschichte. Ein Handbuch zur Geschichte der deutschenSprache und ihrer Erforschung, herausgegeben von W. Besch, A. Bette, O. Reichmann andS. Sonderegger, 2. vollständig neu bearbeitete Auflage, Berlin-New York, de Gruyter, 2000,2. Teilband, pp. 1248-1252.

21 Examples of the same construction to be found only in C are in lines 1046a (them the),1550b (thu thi), 2063b (thero the), 2075 (thero the), 3402b (thia the), 3427a (thia that), 3828(thes the).

22 Cf. M, ll. 92b-93: gern uuas he suîđo, / that he it thurh ferhtan hugi frummean môsti;l. 3138b: ‘gôd is it hêr te uuesanne […]’.


come of the broad Kingdom over which he will rule»).23 All of these featuressuggest that the C-text displays a fairly good amount of linguistic evidencewhich can be considered older than the evidence transmitted by M.

John Jeep, in his recent study on the rhetorical significance of the alliterativetradition in the Hêliand, has come much to the same conclusion: the usage ofalliterating word-pairs in ms. C seems to reflect an older stage.24 For instance, ifwe focus our attention on the common word-pair land : liud(i), employed todescribe what God has created («the land and the people»), we realize that in ms.M (l. 2888a) this pair appears in a slightly different form, namely as land endiliudskepi:

C: land endi liudM: land endi liudskepi

Liudskepi seems to have been a younger coinage by the hand of M,25 as it isnever recorded as such in the other three occurrences within the Hêliand itself(ll. 354a, 4373a: ia land ia liudi; l. 2287a: landes endi liudio) nor does it appear inany other Old Saxon texts.26

In order to draw a partial conclusion, it can be maintained that: • from the linguistic point of view, the C-manuscript seems to reflect an

older stage in the development of Old Saxon (and not just a differentdiastratic variety as shown in the previous studies whose analysis waslimited to phonetics and/or the lexicon);

• this peculiarity deserves proper consideration while editing the text;• a traditional linear apparatus risks hiding such «complex» diachronic

42

MARINA BUZZONI

23 Examples of the same construction to be found only in C are in lines 1370b, 2768b,3268a, 3344a, 4636b, 5016b. On this subject see also J.D. SUNDQUIST, Case Attraction andRelative Clause Variation in the Old Saxon Heliand, 13th Germanic Linguistics AnnualConference, Pennsylvania State University, State College, PA (April 2007),<http://web.ics.purdue.edu/~jsundqui/>.

24 J. JEEP, The rhetorical significance of the alliterative tradition in the Heliand, in New Insightsin Germanic Linguistics, III, edited by I. Rauch – G.F. Carr, New York…Oxford, Peter Lang,2002, pp. 107-130, in particular pp. 117-119.

25 Perhaps on the basis of an Old High German model, since liutscaf appears in theAbrogans twice. (Cf. T. STARCK – J.C. WELLS, Althochdeutsches Glossenwörterbuch, Heidelberg,Winter, 1990, p. 381).

26 Cf. F. HOLTHAUSEN, Altsächsisches Wörterbuch, 2nd edition Köln-Graz, Böhlau, 1967, p.48; M.R. DIGILIO, Thesaurus dei saxonica minora. Studio lessicale e glossario, Roma, Artemide,2008, p. 237.


features, mainly because of its word-oriented (rather than sentence-ori-ented or text-oriented) structure.

Furthermore, it has been shown that:• from the theological point of view M conveys an over-orthodox reli-

gious message, while C seems to be less concerned with conveying sucha message.

2.1. The M-manuscript neums

It is most striking that in M the passage which reports Joseph’s reaction to theAnnunciation (ll. 310ff.) is accompanied by a huge amount of co-occurringparatextual elements: a sign in the margin made of two parallel dashes, an inten-sification of stresses and corrections, and – last but not least – the presence ofneums, denoting musical notation (Plates 2a and 2b).27 This means that thispassage of M was not only read, but also actually used, and perhaps even «per-formed», during the Middle Ages. One may wonder what the purpose of such anotation should have been. A plausible answer is to help the performer use aproper «tonus evangelii» in order to enhance M’s didactic message, namely that,just like Joseph, every Saxon should not give in to doubt.28

43


Pl. 2a: Ms. M, f. 5r; ll. 288-302 (taken from B. TAEGER, Ein vergessener handschriftlicher Befund…)

27 B. TAEGER, Ein vergessener handschriftlicher Befund: Die Neumen im Münchner ‘Heliand’,«Zeitschrift für deutsches Altertum und deutsche Literatur», CVII, 1978, pp. 184-193.

28 See above, section 2, and M. BUZZONI, Re-writing discourse features…, p. 158. See also


Furthermore, palaeographic studies of the M-manuscript by Bischoff haveled him to connect its handwriting style with Corbie.29 This hypothesis seems tofind confirmation in the insertion of neums according to what is supposed tohave been the early Corbie style, that is neums «in campo aperto», without anylines on the page.30 Thus the paratextual elements, which do not appear in anypaper editions of the Hêliand, are unavoidable hints for the reader to give coher-ence to the text (provided that we consider a text as a «communicative event»and not simply as a product). In this respect an electronic edition can proveextremely helpful: it seems that the new systems of data analysis permit us to seepatterns and relationships which were always there, though never before accessi-ble. In turn, we could use the explicatory power of the new medium to allowreaders to discover these relationships by themselves.31

44

MARINA BUZZONI

Taeger’s conclusion: «So würde man einem Schluß auf halbliturgische Verwendung des Textswohl Wahrscheinlichkeit zubilligen können». (B. TAEGER, Ein vergessener handschriftlicherBefund…, p. 190). Another possibility is that this passage is endowed with a legal meaning,perhaps regarding the condition of women when about to be repudiated, though thisassumption needs further investigation, which would lead us far beyond the aims of the pres-ent article.

29 B. BISCHOFF, Die Schriftheimat der Münchener Heliand-Handschrift, «Beiträge zurGeschichte der deutschen Sprache und Literatur», CI, 1979, pp. 161-170.

30 Cf. B. BISCHOFF, Paläographie des römischen Altertums und des abendländischenMittelalters, 2. Auflage von Erich Schmidt, Berlin, 1986, p. 225 and pp. 229-232.

31 «Indeed, the ideal reader of an ideal edition would behave much as does acomputer game player: seeing puzzles in the materials presented, rearranging them to seekways into the puzzle, trying out different solutions […] seeing the results, using what islearnt to frame yet further hypotheses, to create yet further readings». P. ROBINSON,Where We Are with Electronic Scholarly Editions, and Where We Want to Be, «Jahrbuch fürComputerphilologie», V, 2004, pp. 123-143 (127).

Pl. 2b: Ms. M, f. 5r; ll. 309b-314 (taken from B. TAEGER, Ein vergessener handschriftlicherBefund…)


2.2. The C-manuscript Biblical Illustrations

Ms. British Library, Cotton Caligula A.vii displays at the very beginning octotabulae dipictae of uncertain origin [ff. 3-10 (formerly ff. 5-10)]. The whole folioilluminations, whose parchment appears to be insular, run as follows:

(f. 3) Annunciation of the Virgin Mary;(f. 4v) Visitation;(f. 5) Birth of Christ;(f. 6v) Angels and the Shepherds;(f. 7) Massacre of the Innocents;(f. 8v) Presentation in the Temple;(f. 9) Adoration by the Magi;(f. 10v) Baptism of Christ by John the Baptist.

According to the traditional view32 the pictures were probably painted inGermany in the 12th century and prefixed (with others that have perished)either to a Psalter33 or to a Latin copy of the Gospel.34 Sir Robert Cotton him-self has long been considered responsible for the association of the pictures withthe Hêliand.35 Yet, no definite proof of this has ever come to light; so the illumi-nations themselves and their role within the C-manuscript need to be reconsid-ered. Indeed, both Dr. Waagen in his Treasures of Art in Great Britain (publishedin 1854) and, more recently, Karl Hoffmann36 have observed the strongByzantine influence in the style of the paintings. This is particularly evident inthe use of golden grounds, in the brownish flesh tones and in the proportions ofthe figures (e.g. long and elegant bodies, small feet). Though Byzantine in style,many details link the illuminations with the insular post-half-tenth-centuryart:37 for example, if we focus our attention on the «Nativity» (Plate 3a), thescene bears an evident similarity with the famous «Liverpool ivory» (c. 950,

45


32 H. WANLEY, Librorum Veterum Septentrionalium… Catalogus Historico-Criticus …, vol. IIof G. HICKES, Linguarum Veterum Septentrionalium Thesaurus Grammatico-Criticus etArchaeologicus, Oxford, 1705, p. 225; G.F. WARNER, Illuminated Manuscripts in the BritishMuseum…With Descriptive Text, series I-IV, London, British Museum, 1903 [plate 17].

33 So Warner. 34 So Wanley.35 G.F. WARNER, Illuminated Manuscripts…; B. TAEGER, Heliand…, pp. XIX-XX [n. 7].36 K. HOFFMANN (ed.), The Year 1200. A Centennial Exhibition at the Metropolitan Museum

of Art, vol.1, Catalogue, New York, Metropolitan Museum of Art, 1970, p. 291 [n. 284].37 Cf. D. TALBOT RICE, English Art. 871-1100, Oxford, Oxford University Press, 1952, in

particular pp. 179-182.


Plate 3b) as far as the bed and the pillows under the Virgin’s head are concerned.Furthermore, they are both stylistically close to a rendering of the scene whichappears in the Benedictional of St. Æthelwold (975-80, Plate 3c) where theVirgin is reclining on a tilted bed, with Joseph seated at the foot. Other detailsinclude the crib, which is shown on the nearside of the bed, and the ox and don-key that warm the baby with their breath.

46

MARINA BUZZONI

Pl. 3a: «Nativity» , Cotton Caligula A.vii, f. 5r (by permission of the British Library)


47


Pl. 3b: The «Liverpool ivory» (cf. D. TALBOT RICE, English Art…, plate 37)

Pl. 3c: Benedictional of St. Æthelwold, British Library Add. MS 49598, f. 15v<http://www.imagesonline.bl.uk>


As for the «Annunciation» scene (Plate 4a), the similarity with theBenedictional of St. Æthelwold (Plate 4b), where the Angel who appears to theVirgin has large gaudy wings, is once more evident. The Angel’s right hand israised in benediction and his left wing is leaning to the right, just as in the sceneof the Cotton manuscript. A clear difference is represented by the image of thedove, which is completely absent not only in the Benedictional scene, but also inall of the previous and coeval insular representations of the Annunciation, sinceit first appears in the 13th century as the result of the introduction into Britainof continental models.38

48

MARINA BUZZONI

Pl. 4a: «Annunciation», Cotton Caligula A.vii, f. 3r (by permission of the British Library)

38 D. TALBOT RICE, English Art…, in particular pp. 162-163.


The C-manuscript Annunciation scene can thus be seen as a sort of precur-sor of later continental models. The presence of the dove in the illustration ofCotton Caligula A.vii, which – with the only exception of this particular – isiconographically closely parallel to the late tenth-century Benedictional of St.Æthelwold, makes the scene more consistent with the Hêliand poets’ view of theincarnation as being that of the Holy Spirit rather than the Son. This does notmean that the painting was influenced by the Old Saxon poem; it simply meansthat in all probability both the English copy of the Hêliand, and the pictures wenow find prefixed to it, do originate from the same cultural milieu. Which kindof milieu then? A milieu that shows interest in international contacts, winking atthe Continent (and in particular at the Saxon kingdom) not disregarding an

49


Pl. 4b: Benedictional of St. Æthelwold, British Library Add. MS 49598, f. 5v<http://www.imagesonline.bl.uk>


overt link with the Byzantine world. This may well correspond with the laterOttonian Age (it should also be noted that the Ottonian rulers of Germany arein primis Saxon dukes). During this period, the Byzantine Princess Theophano(wife of Otto I and mother of Otto III) and Bishop Bernward of Hildesheimwere responsible for great developments in art, and they exercised a deep influ-ence also in Britain mainly through the protagonists of the monastic revival, i.e.St. Dunstan, St. Æthelwold himself and St. Oswald, all of whom were in closetouch with the continent.39 Keeping good relations with the Ottonian Empirewould result in an increase in trade and wealth, in an age of great internal tur-moil, which saw the renewal of «Danish» incursions. So, it is really tempting toassume that the model of the English copy of the Hêliand came to the Britishisles along with the commodities mentioned in Aelfric’s dialogues as importedby England from the Continent (namely brocade, silk, gold, ivory, wine, oil,spices and glass). Once released from the motherland, the text of the poem owesits fortune not so much to its religious message as to its inherent prestige. Thesame prestige that has probably brought the manuscript into the hands of KingCnut the Great, if we are to trust Franciscus Junius who, while transcribing theHêliand from Cotton Caligula A.vii, described the text as «euangelica historia[…] in usum Canuti Regis adhuc imbuendi primis religionis Christianae pri-mordiis».40

3. Final remarks

The analysis carried out in this study has contributed to highlighting the fol-lowing differences within the Hêliand tradition. From the linguistic point of viewit has been shown that the Cotton Manuscript, though younger, seems to pre-serve an older version of the poem. From the cultural point of view it seems rea-sonable to suppose that, while ms. M displays an orthodox bias and was probablyused as a sort of didactic tale – with the precise intent of both preventing monksfrom yielding to the folly of heterodox thought and teaching them how to dif-fuse the True Christian Message among pagan or newly converted peoples –,41

ms. C owes its existence mainly to political reasons. Whether the illuminations

50

MARINA BUZZONI

39 Cf. Ivi, pp. 20-25.40 Cf. K. DEKKER, Francis Junius (1591–1677): copyist or editor?, «Anglo-Saxon England»,

XXIX, 2000, pp. 279-296.41 Cf. M. BUZZONI, Re-writing discourse features…, in particular pp. 156-158.


were added later or not is still open to debate. Nonetheless, they are in all prob-ability the product of the same multi-cultural milieu as the one from which theHêliand itself springs.

A further point to be considered is how an electronic edition can be prof-itably used to represent the «added value» of this multi-layered text, that is itshistorical communicative effect. This issue will be tackled in the followingAppendix.

51



APPENDIX: The Electronic Hêliand Project: technical aspects(by Francesca Anzalone and Filippo Caburlotto)42

At the start of the project, there were two main problems, namely usabilityand tagging. Nowadays one of the hardest elements in using an electronic edi-tion is the difficulty of interaction with instruments, often caused by the com-plexity of the layout. The objective was to build a user-friendly layout or, in oth-er words, a screen interface that could be completely «usable».43 Above all,usability means a cross-platform, cross-browser, and plug-in-free solution, whichshould also include a lack of restrictions and a user run layout. But how to reach

52

MARINA BUZZONI

Pl. 5: The modal windows <http://venus.unive.it/mbuzzoni/heliand.html>

42 Francesca Anzalone and Filippo Caburlotto (Netlife s.r.l. Software House) are theProject Manager and the Scientific Coordinator of «Archivio dannunziano» <www.archivio-dannunzio.it>, an online interactive and searchable archive which collects the completeworks of Gabriele d’Annunzio, together with the relevant bibliography.

43 F. ANZALONE, F. CABURLOTTO, Comunicare in rete l’usabilità, Milano, Lupetti, 2002;J. NIELSEN, Designing Web Usability, Indianapolis, New Riders Publishing, 2000.


this objective? The template is based on a series of click-and-drag resizable win-dows, which can be activated or deactivated by the user, so that he/she can freelychoose the material to view and in which order, according to his/her own inter-ests. The operations of sorting, dragging and activating the windows requires avery easy task management system, similar to the one users are already familiarwith, since it is present in many of the most popular operating systems.

Technically, the modal windows were developed using a Java/Ajax OpenSource Framework which can build up a multi-layered structure and fill in thewindows with data stored in a database. Nowadays it is a typical web-orientedsolution to split the management of the layout from the context one. In thisway, it is possible to obtain different forms of representation for the same itemsimply by switching on or off a specific set of instructions. It was decided toallow the user the possibility of switching so that he/she could choose whichwindows to activate, or move around or minimising them according to his/herpreferences. The windows are not isolated items; they are connected by hyper-links. Thus, by clicking on a word in the main window (the one at the top-leftof the screen) the user can activate other windows such as, for example, theone containing the image of a manuscript, or the one providing its transcrip-tion. Thus, what is being built up is a hyper-textual environment and a hyper-textual way of using the electronic edition, based not on a static but on aninteractive model. The main principles followed are text mobility, on the onehand, and flexibility of text representations, on the other. All of the windowswork in this way except for the «Search» one. In fact, the Search instrument ismore powerful than all the other ones. It is structured in such a way as to allowthe user to search using words, part of words, or tags. Furthermore, the usercan choose some words and look for them highlighting some «blank» spacesbetween them, for example the search undar *** gifrumida gives as a possibleresult the line undar mancunnea / mâriða gifrumida. The focus of the work wasto provide the users with the widest range of possible research tools and solu-tions. A search can either be free (through the simple insertion of words ortags), or driven.

Another means of data visualization is based on a drop-down menu, where-by the user can choose from a set of fixed solutions that are dynamically gener-ated by the Ajax application looking at the tags and the explanations stored inthe database. It was possible to provide this option because the decision was tak-en to use a relational database in which the texts split from the tagging systemswere stored. The Java interpreter re-builds the code and presents it to the uservia the correct style sheet. This is the second main technical aspect of the system.One of the major difficulties using the TEI-P5 tagging system is the overlapping

53



hierarchies problem.44 The approach chosen for solving this problem was togrant the users access to more than one tagging structure. Basically, what hap-pens is that the text is stored in the database using some flags that tell the Javainterpreter where to put the tags. In so doing, it is possible to obtain more thanone tag for any instance, and represent them in different ways by using differentstyle sheets. A fluid approach to the final layout permits users to choose betweenvisualizing only one tag scenario or several. In the latter case, different taggingsolutions are represented with different colours. The advantage of this approachis that the text can be fully tagged while avoiding most of the problems withhierarchies and overlapping, thus obtaining a valid XML document (or evenmore than one) in the majority of cases.

Up till now only the front-end aspect, that is the layout for the final user, hasbeen considered. However a back-end panel was also developed. Scholars can

54

MARINA BUZZONI

44 It goes beyond the aims of this appendix to deal with the overlapping problem. Forreferences about the ongoing debate on this thorny issue, see: C. MAH – J. FLANDERS –J. LAVAGNINO, Some Problems of TEI Mark-up and Early Printed Books, «Computers and theHumanities», 31, 1997, pp. 31-46.

Pl. 6: Window mobility <http://venus.unive.it/mbuzzoni/heliand.html>


55


insert data from a dynamic admin(istration) panel by simply uploading them viaa web form, or via a form in which they can copy and paste texts from any XMLeditor. Furthermore, considering character set problems, a WYSIWYG editorwas developed which can help with the insertion of special characters, images,hypertextual links, etc. This enabled us to separate the scientific proceduresfrom the technical ones in order to allow scholars to work on the texts withouthaving a specific technical know-how.


ODD EINAR HAUGENUniversitetet i Bergen

AN APOLOGY FOR THE TEXT THAT NEVER WAS:RECONSTRUCTING THE KING’S MIRROR

Abstract. The great majority of Medieval Nordic works have been preserved in anumber of manuscripts that normally differ with respect to orthography. Thesemanuscripts are frequently incomplete, lacking smaller or larger parts of the work. Asa consequence, the editor is faced with the challenge of constructing the text of thework from sources that have different orthographies and cover different parts of thework. The present article addresses this issue with respect to the editing of the OldNorwegian 13th-century work Konungs skuggsjá (‘The King’s Mirror’), using a tripar-tite typology of text editions (monotypic, synoptic and eclectic). The eclectic editionby Rudolf Keyser, Peter Andreas Munch and Carl Richard Unger (1848) is comparedwith the monotypic edition by Ludvig Holm-Olsen (1945), and the strengths andweaknesses of these editions evaluated. A compromise between these editions is sug-gested, in which the diplomatic orthography of a monotypic edition can be combin-ed with the uniform orthography of an eclectic edition.

1. A state of loss

An Old Norse manuscript may look almost as fresh today as it was some800 years ago, but the vagaries of time have taken their toll on many manu-scripts. The majority have probably been lost, and of those that survive, manyare damaged or incomplete. The Old Norwegian manuscripts seem to havefared worse than the Old Icelandic ones, which to some extent is due to the factthat the Norwegian language changed more over time than the Icelandic lan-guage. As a consequence, Norwegian manuscripts became less accessible to theNorwegians than the Icelandic manuscripts were to the Icelanders, and so werenot kept in as great numbers.1 The law manuscripts are probably the best-pre-

57

1 See E.F. HALVORSEN, Ble det lest litteratur i Norge i middelalderen?, «Saga och Sed», 1982,pp. 128–140.


served of the Norwegian manuscripts but even among them, the degree of frag-mentation is high. Of the Landslǫg of King Magnús Hákonarson (reigned1263–1280), 39 fairly complete medieval copies have survived, in addition to31 fragments of varying size.2 For other literary genres, the picture is bleaker.Stefán Karlsson has estimated that in addition to law manuscripts and land reg-isters («jordebøker»), no more than 54 manuscripts have survived of thosewhich were known to have existed in Norway in the 12th and 13th centuries.3

It is impossible to give any exact estimate of how many Norwegian manuscriptshave been lost, but the collection of fragments in the National Archive(Riksarkivet) in Oslo may give some indication as to the ratio of fragments toonce-existing manuscripts. Today, there are about 500 fragments of OldNorwegian manuscripts in the National Archive, most of which belong to lawmanuscripts. It has been estimated that these fragments are the remains ofabout 100 manuscripts. Since the fragments are quite small – usually less thana page – the degree of loss in the corpus of Norwegian manuscripts can be seento have been very high indeed.4 A similar ratio of loss applies to the Latin man-uscripts in the National Archive: there are about 5,000 fragments that oncebelonged to around 1,200 manuscripts.5

In a selection of six 13th-century Norwegian works I recently made, itturned out that four of the six main manuscripts of these works were incom-plete, preserving only a part of each respective work.6 These four works were theOld Norwegian homilies in AM 619 4° (c. 1200–1225), Strengleikar in DG 4–7(c. 1270), Konungs skuggsjá in AM 243 b α fol (c. 1275) and Barlaams ok Josaphatssaga in Holm perg 6 fol (c. 1275). The two remaining manuscripts that con-tained whole works were that of the Legendary saga of Óláfr inn helgi Haraldsson

58

ODD EINAR HAUGEN

2 M. RINDAL, Laws. 3. Norway, in Ph. Pulsiano (ed.), Medieval Scandinavia: An Encyclopedia,New York, Garland, 1993, p. 386.

3 STEFÁN KARLSSON, Islandsk bogeksport til Norge i middelalderen, «Maal og Minne», 1979,pp. 4–6. Of these manuscripts, Stefán believed that at least 31 were of Icelandic provenance.

4 TH. EKEN, Gammelnorske membranfragmenter i Riksarkivet, Oslo, Selskapet til utgivelse avgamle norske håndskrifter, 1963, p. xiii. (Corpus Codicum Norvegicorum Medii Aevi,Quarto Series, vol. 3.)

5 Ibidem. It might be noted that the majority of the Latin manuscripts were produced out-side Norway.

6 O.E. HAUGEN, A Quarrel of the Ancients and the Moderns: On the merits of Old and NewPhilology in the editing of Old Norse texts, in F. Ferrari – M. Bampi (eds.), On Editing OldScandinavian Texts: Problems and Perspectives, Trento, Università degli Studi di Trento, 2009(in press).


in DG 8 II (c. 1225–1250), and the Landslǫg of King Magnús Hákonarson inAM 305 fol (c. 1300).7 Although this selection may not be representative, thehigh degree of fragmentation tallies nonetheless with what we know about thestate of surviving Old Norwegian manuscripts.

The editing of a work which is incompletely preserved in the surviving man-uscripts is a difficult task, irrespective of whether the editor identifies with tradi-tional textual criticism or with the so-called new philology. The editor may, ofcourse, restrict himself or herself to a single codex optimus and edit this as itstands. If there are no other manuscripts this is the only solution. In most cases,however, several manuscripts of the work may be extant. The editor is thenfaced with the challenge of documenting the textual variation in the manu-scripts and, as far as possible, piecing together a coherent and uniform text of thework. This is a demanding task for works that have been preserved in a numberof complete manuscripts and even more demanding when the surviving manu-scripts are incomplete.

In this article, I will outline the editorial options when the manuscript mate-rial is incomplete, using Konungs skuggsjá as a case study. The codex optimus ofKonungs skuggsjá, AM 243 b α fol (ill. 1), lacks around 20 % of the text. This is ahigher figure than that in the case of the three other works mentioned above,but it is not exceptionally high. Fortunately, some of the younger manuscripts ofKonungs skuggsjá preserve the complete text of the work. However, when turningto the 22 manuscripts which have independent critical value (according to therecension by Ludvig Holm-Olsen in his 1952 doctoral dissertation), none ofthese manuscripts carry the complete text.8 This means that the constitution ofthe text of Konungs skuggsjá is a challenging task.

2. Editorial strategies

When planning a new edition, an editor is faced with two important ques-tions. First of all, he has to decide how the text should be rendered, ranging from

59

An apology for the text that never was

7 The sigla and dating of these manuscripts follow the index volume of Ordbog over detnorrøne prosasprog, København, Den Arnamagnæanske Kommission, 1989. The siglum AMhere refers to manuscripts kept in the Arnamagnæan Collection (Den ArnamagnæanskeSamling) at the University of Copenhagen, while the DG siglum refers to manuscripts in theDelagardian Collection (Delagardieska Samlingen) in Uppsala University Library.

8 L. HOLM-OLSEN, Håndskriftene av Konungs skuggsjá, København, Munksgaard, 1952, ch.IV, cf. the table on pp. 117–121. (Bibliotheca Arnamag næana, vol. 13).


60

ODD EINAR HAUGEN

Ill. 1. Konungs skuggsjá in the Norwegian codex optimus, AM 243 b α fol, p. 2 (c. 1275).Reproduced in 55 % of original size. Photograph by Susanne Reitz and Elin L. Pedersen, TheArnamagnæan Collection, Copenhagen, 2008. The script is protogothic, typical of Norwegianmanuscripts in the second half of the 13th century.


61


a very close transcription on the one hand, to a fully normalised version on theother. Then, in the case of works being preserved in more than one manuscript,he has to decide how the text should be selected, either by following a singlemanuscript (usually the one which is regarded as being the best manuscript), orby constructing the text from more than one source. The typology presentedhere is a somewhat simplified version of that presented in my article Constitutiotextus,9 and while it is based on the traditions prevalent in Old Norse philology,I believe it has wider validity.

2.1. Text reproduction

There is no single way of reproducing a text from a manuscript into print.Some editors choose to render the text by means of an extremely close tran-scription, with no or minimal editorial intervention. Other editors representthe text with some simplifications and regularisations. Yet other editors presentthe text in a fully normalised orthography, including modern punctuation.While linguists and language historians prefer texts that follow the sourceclosely, literary scholars and historians often prefer to work from a normalisedversion. Some scholars are interested in the actual manuscript while others areconcerned with the work behind the manuscripts. Several types of textualreproduction can be identified along this axis of textual fidelity. In my articleConstitutio textus, I suggested five levels, while Guðvarður Már Gunnlaugssonhas gone into even greater detail more recently.10 In the context of this article,I believe three levels may be identified as prototypical. These levels have alsobeen selected for use by the Medieval Nordic Text Archive in the MenotaHandbook.11

9 O.E. HAUGEN, Constitutio textus. Intervensjonisme og konservatisme i ut gjevinga av norrønetekster, «Nordica Bergensia», VII, 1995, pp. 69–99.

10 GUÐVARÐUR MÁR GUNNLAUGSSON, Stafrétt eða samræmt? Um frædilegar útgáfur ognotendur þeirra, «Gripla», XIV, 2003, pp. 197–235. (Stofnun Árna Magnússonar á Íslandi,Rit 60.)

11 O.E. HAUGEN (ed.), The Menota Handbook: Guidelines for the encoding of Medieval Nordicprimary sources, v. 2.0, Bergen, 2008, <http://www.menota.org/guidelines> (date of access: 30January 2009).


The closest way of reproducing a text is the photographic facsimile. Modern-day facsimiles are digital, in colour and often with an impressive resolution.Some photographs may in fact be easier to read than the actual manuscript.There are manuscripts which have been written so clearly and are so well pre-served that they can be read by anyone with a knowledge of the script and thelanguage. Many other manuscripts, especially vernacular ones, are less easy toread. While a photographic facsimile is an extremely useful way of reproducinga text, the text as such is not searchable. Transcriptions are therefore necessary,and it is with these that the following typology is concerned. To illustrate thepossibilities, a short passage from Konungs skuggsjá as preserved in AM 243 b αfol will be used as an example (ill. 2).

62

ODD EINAR HAUGEN

Ill. 2. Slightly modified detail of Konungs skuggsjá in AM 243 b α fol, p. 2 (fol. 1v), col.B, l. 24–30 (cf. ill. 1). Reproduced in 100 % of original size. Note the «f» in the margin forfilius, i.e. sunr, ‘son’.

2.1.1. Facsimile levelAt the facsimile level, the transcriber keeps to the text as it is presented in

the manuscript as closely as possible, not making any interpretations other thanthose that are strictly necessary. The manuscript is typically transcribed line byline, using the same inventory of characters as found in the source, copying anyabbreviation characters exactly as they are, and not changing capitalisation orpunctuation. Transcriptions at this level are faithful, but may be difficult to


read and hard to cite.12 The first transcription in ill. 3 is at the facsimile level.Note that the line and word divisions of the original have been copied, thatcharacters are reproduced in some paleographical detail, and that abbreviationsare reproduced with no interpretation (cf. the horizontal bars in lines 3 and 5).A few editions of this type were produced in the second half of the 19th centu-ry, but they disappeared after the introduction of the photographic facsimileeditions in the late 19th century. Thanks to modern computer and font tech-nology there seems to be a revival of some sort currently, especially in conjunc-tion with the transcription at this level being offered as one of two or even threeparallel transcriptions.13

63


æ� ��ı at ec em

nu alettazta �� Sunr

aldre �a �umz ec at �ara lan�da mæ�al

��ı at ec tretumz eıgı til hır�ar leıta

�� enec hæ��a e� anna��a man�a ı�u

a� . Nu er �ætta �ı mın nema �er

læggıt annat ra� tıl �ı� mer.

Sunr. Mæ� �vi at ec em nu alettazta �� aldre �a fysumz ec at

fara lannda mæ�al �vi at ec treystumz eigi til hir�ar leita fy� enec

hæf�a se� anna�a manna si�u a�r. Nu er �ætta fysi min nema �er

læggit annat ra� til vi� mer.

Sonr: «Me� �ví at ek em nú á léttasta aldri, �á f�sumk ek at fara

landa me�al, �ví at ek treystumk eigi til hir�ar leita fyrr en ek hef�a

sét annarra manna si�u á�r. Nú er �etta f�si mín, nema �ér leggit

annat rá� til vi� mér.»

M

Ill. 3. Transcriptions of the text in ill. 2. The first is on the facsimile level, the second onthe diplomatic level, and the third on the normalised level. The erasure of approximately 5characters in line 2 is indicated by .

12 This is much less of a problem now than it was only a few years ago, since The UnicodeStandard v. 5.1 has accepted many medieval characters, especially in the new range LatinExtended-D, <http://www.unicode.org> (date of access: 30 January 2009). Moreover, theMedieval Unicode Font Initiative provides a recommendation for the encoding of a largenumber of other characters as well as making fonts with these characters freely available,<http://www.mufi.info> (date of access: 30 January 2009).

13 See O.E. HAUGEN, Parallel views: Multi-level encoding of Medieval primary sources,«Literary and Linguistic Computing», XIX, 2004, pp. 73–91.


2.1.2. Diplomatic levelA diplomatic transcription is often defined as an exact copy of the original,

nothing added, nothing removed. However, the typical diplomatic edition inter-prets and simplifies the source in several respects compared to the facsimile level.Firstly, it usually normalises the character inventory so that the characters repro-duced are restricted to those that presumably have individual phonological value.At the diplomatic level, the transcriber usually does not distinguish between char-acter variants such as the Carolingian and the Insular form of «f», or the straightand the round form of «r». Secondly, abbreviations are expanded, either silently ordisplayed typographically by means of italics. Thirdly, punctuation and capitalisa-tion may be regularised so that personal and place names are more easily identifi-able. The second transcription in ill. 3 is at the diplomatic level. Compared withthe facsimile level transcription, the character inventory has been reduced, abbre-viations have been expanded, and line divisions are no longer copied. Many OldNorse editions belong to this level, such as the great majority of editions by NorskHistorisk Kjeldeskrift-Institutt, and also many of the Arnamagnæan editions.

2.1.3. Normalised levelAt the normalised level, the orthography of the source is regularised. This

level is to a great extent exclusive to Old Norse, i.e. Old Norwegian and OldIcelandic, texts, since there is little regularisation in other European vernaculareditions. For Old Norse, however, there is a long-standing tradition of regularis-ing the orthography, partly on the basis of Modern Icelandic orthography.During the 19th century, this orthography was fixed through grammars and dic-tionaries, and in the 20th century, the Íslenzk fornrit series established a clearstandard. The great majority of textbooks adhere to this standard, albeit withsome modifications. The regularisation addresses a number of aspects includingthe marking of vowel length (by acute accent), consonant length (gemination),and unstressed vowels (only i, a and u), as well as introducing the capitalisationof proper names and sentence-initial words, and what is essentially a modernpunctuation (including question marks and quotation marks, for example). Thethird transcription in ill. 3 is at the normalised level, which is easily readable forthose who have studied elementary Old Norse (even if the meaning may turnout to be less easily grasped).

2.2. Selecting the text

If there is more than one manuscript of a work, the editor has to make aselection from the available manuscripts, either keeping to the text of a single or

64

ODD EINAR HAUGEN


a few manuscripts, or constructing the text from a greater number of manu-scripts. From the point of view of the text constitution (constitutio textus), whichis the term used by Paul Maas,14 I would identify three focal types of editions:monotypic, synoptic and eclectic.

2.2.1. Monotypic editionsA monotypic edition is based on a single manuscript, often the best manu-

script, the codex optimus. The best manuscript may not necessarily be the oldestmanuscript: it may rather be the best preserved manuscript, or a combination ofboth. Due to the high degree of fragmentation of Old Norwegian manuscripts,one may have to use manuscripts that are not complete. Selecting a single man-uscript will always be done at the expense of representing textual variation thatmay exist, but it has the advantage of allowing the editor to represent the sourcein great detail. Even if a normalised rendering is possible with a single manu-script, diplomatic or even facsimile renderings are rather more common.

2.2.2. Synoptic editionsThe synoptic edition juxtaposes several versions of the text, from two or

more manuscripts. There is no upper limit to the number of manuscripts, butfor practical reasons there are seldom more than four different versions of thetext in a printed edition. An electronic edition may offer many versions in asmany windows, but that is strictly not a synoptic edition. For an edition to besynoptic, the texts have to be aligned, which typically is done on the two-dimen-sional spread of pages in a printed edition. Due to transpositions of passages oftext, or instances where the text is supplemented or deleted, this alignment canbe difficult. Johann Jakob Griesbach found it fairly easy to make a synoptic edi-tion of the gospels of Matthew, Mark and Luke, but when he added the gospelof John, he ran into difficulties with the alignment of the texts.15 There are anumber of synoptic Old Norse editions, some of which are synoptic throughoutwhile others only present certain passages of the text synoptically.16

65


14 In his Textkritik, Leipzig, Teubner, 1960, p. 5.15 J. J. GRIESBACH (ed.), Synopsis Evangeliorum Matthaei, Marc et Lucae, Halle, Curtius, 1774

(the three synoptical gospels); Synopsis Evangeliorum Matthaei, Marci et Lucae una cum iisJoannis pericopis, 2nd ed., Halle, Curtius, 1797 (all four gospels).

16 One of the earliest examples of a synoptic Old Norse edition is the brief edition ofSigurd Ranesønss Process, edited by G. Storm, Kristiania, Malling, 1877, while an early full-length edition is Fóstbrœðra saga, edited by Björn K. Þórólfsson, København, Jørgensen,1925–27. The majority of Old Norse editions from the 17th and 18th centuries present the


2.2.3. Eclectic editionsThe eclectic edition is an edition which aims to reconstruct the original text

of the work, in parts or preferably as a whole. This has been the aim of most edi-tors of Classical texts, which typically are preserved in much younger, post-Carolingian manuscripts. It has also been the aim of many Old Norse editors,since no originals of Medieval Nordic works from the period before untilaround 1400 are thought to be extant.17 Medieval copies of a work almost with-out exception deviate from each other and since this deviation is distributedamong the copies (although not necessarily evenly), it follows that a single man-uscript can seldom be a faithful representation of the lost original. The eclecticedition thus tries to remedy this situation by reconstructing the text of the orig-inal work as far back as the extant manuscripts allow. Several Old Norse editionsfrom the second half of the 19th century are eclectic, e.g. the edition of Laxdœlasaga by Kristian Kålund.18

2.3. An editorial landscape

The typology discussed here can be displayed along two axes, as shown in ill.4. The horizontal axis represents the degree of reconstruction while the verticalaxis represents the diversity of the manuscript material. While manuscript tradi-tions with few copies or only a single manuscript favour monotypic editions,larger textual traditions which have a higher degree of variation tend to favour asynoptic edition. If the editor wishes to produce an edition with less interpreta-

66

ODD EINAR HAUGEN

Old Norse text with accompanying translation into Latin or one of the Scandinavian lan-guages, or sometimes both, but these do not qualify as synoptic editions, since they do notpresent two or more versions of the same text in the same language.

17 Although Flateyjarbók (GKS 1005 fol) is a compilation of older sagas (of which theoriginals have been lost) it may be argued that this is a case of an original, and definitely anoriginal compilation – it is hardly likely that such a huge manuscript, the parchment forwhich required the skin of 113 calves, was produced in more than a single copy. Flateyjarbókwas written in the years 1387–1394 by two priests who name themselves, Jón Þórðarson andMagnús Þórhallsson. A convincing example of an original and indeed a not fully finishedoriginal is the translation of Barlaam och Josaphat in Nådendals klosterbok, Holm A 49 (c.1440), which contains both a draft and a partly finished fair copy (renskrift) of the text. Cf.MARIA ARVIDSSON, Den fornsvenska översättningen av legenden om Barlaam och Josaphat i Holm A49, in Karl G. Johansson – Maria Arvidsson (eds.), Barlaam i nord, Oslo, Novus, 2009, pp.153–176. (Bibliotheca Nordica, vol. 1).

18 K. KÅLUND (ed.), Laxdæla saga, København, Møller, 1889–1891. (Samfund tilUdgivelse av gammel nordisk Litteratur, vol. 19).


tion, he or she will move towards the monotypic or synoptic solution, while aneditor who prefers to build the text synthetically from several manuscripts willmove towards producing an eclectic edition.

67


There is a tendency to co-variation along the two axes in this model, in thesense that monotypic and synoptic editions tend to be on the diplomatic or evenfacsimile level, while eclectic editions almost always are on the normalised level.An edition which is based on a single manuscript is free to choose its level of rep-resentation. Some monotypic editions have in fact regularised orthography, suchas most editions of the Eddic poems, while other editions favour a closer tran-scription of the source. For an eclectic edition, the orthography is usually regu-larised; the alternative would be to render the text in an orthography alternatingaccording to the different sources. For Old Norse texts, the regularised orthogra-phy reflects the linguistic state of the first half of the 13th century. For works dat-ing from the 13th and the 14th centuries, the difference between the orthogra-phy of the manuscript and the regularised orthography may not be that large.There are, however, several medieval works that are only known to us in relative-

Ill. 4. A two-dimensional editorial landscape.


ly late or even post-medieval copies, and even if regularising their orthography ispossible, the difference between the young orthography of the manuscripts andthe reconstructed orthography of the now-lost original may be considerable.

Even if large manuscript traditions with considerable textual variation favoura synoptic edition, the variation between the extant manuscripts may sometimesbe too large to represent. The four gospels of the New Testament have alreadybeen mentioned as an example where aligning the gospel of John with the threeother gospels has proved difficult. An Old Norse parallel would be the transla-tion of the gospel of Nicodemus, Niðrstigningar saga. This translation is pre-served in four medieval manuscripts, but the degree of textual variation as wellas fragmentation of each manuscript is so high that Carl Richard Unger chose topresent all four texts sequentially, i.e. one after the other, in his Heilagra mannasögur.19 This is not, therefore, a synoptic edition, but rather a sequential mono-typic edition. In general, the larger the manuscript material, the more difficult itis to construct a single, eclectic text, and the greater the textual variation, themore difficult it is to align texts in a synoptic edition.

The horizontal axis in the editorial landscape reflects not only the diversityof the manuscript material but can also be interpreted as an axis reflecting theopposition between the traditional or old philology and the so-called new ormaterial philology. The new philology will be found on the left-hand side of thedivide and old philology on the right-hand side.20

3. The King’s Mirror: Konungs skuggsjá

Konungs skuggsjá (‘The King’s Mirror’) has long been considered the jewel inthe crown of Old Norwegian literature. It was probably written in Bergen in the1250s for the benefit of the sons of King Hákon Hákonarson (d. 1263), Hákonungi (‘the young’) Hákonarson (d. 1257) and Magnús Hákonarson (d. 1280),later known as lagabøtir (‘the Lawmender’). The King’s Mirror is modelled oncontemporary European works, but contains much native Norwegian material.

68

ODD EINAR HAUGEN

19 C.R. UNGER (ed.), Heilagra manna sögur, 2 vols., Christiania, Bentzen, 1877; Niðrstig -ningar saga is printed in vol. 2, pp. 1–20.

20 There is a considerable amount of literature on the distinctions between the old vs. thenew philology; for a couple of recent discussions within Old Norse philology, see M.J.DRISCOLL, The words on the page: Thoughts on philology, old and new, and O.E. HAUGEN, Stitchingthe text together: Documentary and eclectic editions in Old Norse philology, both in J. Quinn – E.Lethbridge (eds.), Creating the Medieval Saga: Versions, Variability and Editorial Interpretations ofOld Norse Saga Literature, Odense, Syddansk Universitetsforlag, forthcoming in 2010.


It has been preserved in around 60 manuscripts, many of which are youngerIcelandic paper manuscripts. This makes the King’s Mirror apparently one ofthe most popular, or at least intensively-transmitted, works in Old Norse litera-ture. Of Norwegian works, only the Landslǫg by Magnús Hákonarson has beenpreserved in more manuscripts.21 The oldest and most complete Norwegiancodex of the King’s Mirror is AM 243 b α fol from c. 1275 (ill. 1 above). In hisdoctoral dissertation, Ludvig Holm-Olsen made an exhaustive recension of allknown manuscripts.22 Holm-Olsen’s recension led to a stemma in which AM243 b α fol (hovedh.) is located very close to the archetype (ill. 5). Three otherNorwegian manuscripts, RA 58A (α), RA 58C (β) and NKS 235 g 4° (γ), alsobelong to the oldest part of the extant textual tradition, but only fragmentsremain of these manuscripts.23 Unfortunately, AM 243 b α fol is far from com-plete. It once had 86 leaves, but only 68 have survived to the present, meaningthat around 20% of the text is lacking in this manuscript.

69


Ill. 5. The stemma of Konungs skuggsjá, from the revised edition of Konungs skuggsiá byLudvig Holm-Olsen, Oslo, 1983, p. xiv.

21 As mentioned above, there are 39 fairly complete medieval manuscripts of the Landslǫgand 31 fragments, making a total of 70 medieval manuscripts, assuming that all fragmentsbelong to different manuscripts. Gustav Storm estimated that in the 16th century, there wereat least 80 manuscripts of the law, see Om Haandskrifter og Oversættelser af Magnus LagabøtersLove, Christiania Videnskabsselskabs Forhandlinger 1879, Christiania, Dybwad, p. 11.

22 L. HOLM-OLSEN, Håndskriftene av Konungs skuggsjá… 23 The siglum RA refers to the National Archive (Riksarkivet) in Oslo, and NKS to the

New Royal Collection (Den nye kongelige samling) in the Royal Library (Det KongeligeBibliotek), Copenhagen.


In his recension of the Konungs skuggsjá manuscripts, Holm-Olsen concludedthat 22 manuscripts have independent value, i.e. they are not copies of other,known manuscripts. None of these independent manuscripts is complete. Inorder to construct the complete text of Konungs skuggsjá, several of the independ-ent manuscripts must be used, including the Icelandic ones, especially AM 243a fol and AM 243 e fol.

The first edition of Konungs skuggsjá was published by Hálfdan Einarsson andJón Eiríksson in Sorøe in 1768.24 Like many other 18th-century editions, theSorøe edition uses facing pages to present the Old Norse text in parallel withtranslations into Danish and Latin. The Old Norse text is based on some of theyounger Icelandic manuscripts, but is supplemented by a critical apparatus withreadings from older manuscripts. After the publication of the Sorøe edition, thetext was edited by the triumvirate Rudolf Keyser, Peter Andreas Munch andCarl Richard Unger in 1848, by Oskar Brenner in 1881, by Finnur Jónsson in1920 and finally by Ludvig Holm-Olsen in 1945.25 In this article, the 1848 edi-tion by Keyser, Munch and Unger and the 1945 edition by Holm-Olsen will bediscussed in some detail, beginning with the latter edition, which remains themost recent edition of Konungs skuggsjá. Holm-Olsen had planned a critical edi-tion based on his 1952 recension, but did not find the time to make this editionamong his many other obligations.26

3.1. The 1945 edition of Konungs skuggsjá

The 1945 edition was produced by Ludvig Holm-Olsen before he had actu-ally performed the recension of the manuscripts. However, AM 243 b α fol had

70

ODD EINAR HAUGEN

24 HÁLFDAN EINARSSON – JÓN EIRÍKSSON (eds.), Kongs-skugg-sio: Utlögd a daunsku og lat-inu, Sorøe, Lindgren, 1768.

25 R. KEYSER – P.A. MUNCH – C.R. UNGER (eds.), Speculum regale: Konungs-skuggsjá:Konge-speilet, Udgivet efter Foranstaltning af det akademiske Collegium ved det kongeligenorske Frederiks-Universitet, Christiania, Werner, 1848; O. BRENNER (ed.), Speculum regale:Ein altnorwegischer Dialog, Nach Cod. Arnamagn. 243 Fol. B und den ältesten Fragmentenherausgegeben, München, Kaiser, 1881; FINNUR JÓNSSON (ed.), Konungs skuggsjá: Speculumregale, Udgivet efter håndskrifterne af det Kongelige Nordiske Oldskriftselskab, Kjøbenhavn,Gyldendal, 1920; L. HOLM-OLSEN (ed.), Konungs skuggsiá, Utgitt for Kjeldeskriftfondet,Oslo, Dybwad, 1945 (Gammelnorske tekster, vol. 1); revised edition Oslo, Kjeldeskrift -fondet, 1983 (Norrøne tekster, vol. 1).

26 He first served as dean and then as rector of the University of Bergen. On Holm-Olsen’s plans for a new edition, see the preface to the revised edition of Konungs skuggsjá,Oslo, Kjeldeskriftfondet, 1983, p. ix.


long been established as the Norwegian codex optimus, and a couple of theyounger Icelandic manuscripts, especially AM 243 a fol and AM 243 e fol, wererecognised to be important supplemental textual witnesses. In the 1945 edition,which is strictly diplomatic, Holm-Olsen followed AM 243 b α fol whereverpossible, filling the missing parts with text mainly from AM 243 a fol and AM243 e fol. As a consequence, the text of the work switches back and forthbetween Icelandic and Norwegian orthography. 27 An example of this change isshown in ill. 6, where the first part, ending with «stutta skyrtu» in line 7 of theextract, is in Norwegian orthography while the remaining text, beginning with«þijna og lijn», is in Icelandic orthography. The first part is based on theNorwegian codex optimus, AM 243 b α fol (c. 1275), and the second part on aconsiderably younger Icelandic manuscript, AM 243 e fol (c. 1500–1550). Inthe stemma reproduced in ill. 5 above, AM 243 b α fol is referred to as «hov-edh.» (‘principal manuscript’) and AM 243 e fol as «e», both belonging to the«B» branch of the tradition.

71


27 This way of organising the edition was not unprecedented. Oscar Brenner in his 1881edition and Finnur Jónsson in his 1920 edition had done essentially the same.

Ill. 6. Extract from ch. 30 in the edition of Konungs skuggsiá by Ludvig Holm-Olsen, Oslo,1945, p. 45, l. 32 – p. 46, l. 5. The edition switches from Norwegian to Icelandic orthogra-phy three words before the end of line 7 in the extract. In this passage, the father advises hisson on how to appear at the royal court.


From the outset, the 1945 edition was simply a collection of paper slips inthe Old Norwegian Dictionary (Gammalnorsk Ordboksarkiv) in Oslo. Eachword in the text was listed on a slip with lemma and grammatical form and a lineof context. By arranging these slips in the correct order, the edition could bepieced together from the beginning to the end, and this was in fact what hap-pened. This is a most unusual way of producing an edition, and can to someextent explain the very close reading of the manuscript. It also explains theunusually wide dimensions of the pages of the edition, since the width reflectedthe column width of the paper slips. The text was based on the 1920 edition byFinnur Jónsson, and checked against the facsimile edition by George T. Flomand photostat copies of the Icelandic manuscripts.28 Since the work was doneduring the Second World War, it was not practically possible to proofread theedition against the original manuscripts in Copenhagen. The selection of vari-ants was based on the material in Finnur Jónsson’s edition, but re-evaluated byHolm-Olsen in his revised 1983 edition. The text of the manuscripts, however,has not been changed in the revised edition.

The solution chosen by Holm-Olsen for dealing with the problem of frag-mentary manuscripts in the 1945 edition was later followed by Magnus Rindalin his edition of Barlaams ok Josaphats saga.29 This edition, too, follows the codexoptimus from beginning to end, supplying missing passages from youngerIcelandic manuscripts. The codex optimus, Holm perg 6 fol, contains about 95 %of the text, so the amount of supplied text is less than for Konungs skuggsjá, butstill considerable. Rindal has rendered any supplementary text in its Icelandicorthography, as did Holm-Olsen, but Rindal gives the supplementary passagesin an appendix rather than inserting them into the text of the main codex. Inboth cases, the orthographic uniformity of the work has to be construed by thereaders, and in the case of Rindal’s edition, the text must be read by switchingbetween the main text and the appendix. From a lexicographical or linguisticpoint of view this may be preferable, but not so for users who would like tostudy the text as representative of the work, whether it is Konungs skuggsjá orBarlaams ok Josaphats saga.

72

ODD EINAR HAUGEN

28 G.T. FLOM (ed.), The Arnamagnean Manuscript 243 Bα, folio at Det Kongelige Bibliotek,Copenhagen: The main manuscript of Konungs skuggsjá in phototypic reproduction with diplomatictext, Urbana (Ill.), The University of Illinois, 1915.

29 M. RINDAL (ed.), Barlaams ok Jospahats saga, Oslo, Norsk Historisk Kjeldeskrift-Institutt, 1981. (Norrøne tekster, vol. 4).


The 1945 edition of Konungs skuggsjá is a monotypic edition in that it followsthe text of the codex optimus in great detail and with minimal emendation. Thetext of the missing passages of the codex is supplied from other manuscripts, butthis part of the edition is also monotypic. The orthography of the supplementaryIcelandic manuscripts has been left as it is. The edition may thus be seen as asequential monotypic edition, belonging to the left-hand region of the editoriallandscape in ill. 4. This is not the whole story, however, for it might be askedhow the editor arranged the monotypic pieces of text in a meaningful order. AsHolm-Olsen states in the preface,30 this was done on the basis of the 1920 edi-tion of Finnur Jónsson, which primarily is based on AM 243 b α fol and AM243 e fol.31 If a work has been preserved only in fragmented manuscripts, acompletely monotypic edition is probably not possible. There has to be someelement of reconstruction, whether it is performed by the editor of the mono-typic edition or by previous, more eclectic-minded editors.

3.2 The 1848 edition of Konungs skuggsjá

What alternatives are there to a mixed-orthography edition of the type madeby Holm-Olsen and by Rindal? For 19th-century editors, the answer seemed tobe obvious: to edit the text in a uniform orthography. That was also the decisiontaken by Keyser, Munch and Unger in their 1848 edition of Konungs skuggsjá.There are good reasons for doing this. One is that it makes the text simpler toread, since readers only have to worry about a single orthography. Since theorthography typically is the standard Old Norse orthography known from gram-mars, dictionaries and introductory text books, this is also a great help for theunderstanding of the text. Furthermore, a uniform orthography can be seen asan approximation of the now-lost original, for it is highly unlikely that a workwas produced in variable orthography, apart from the minor variations that arefound in almost any manuscript. To what extent the normalised orthography isan approximation to the actual orthography of the original is a moot point, butfew scholars would dispute that standard Old Norse orthography is closer to theOld Norwegian language of the 1250s than the orthography of Icelandic 14th-or 15th-century copies.

73


30 L. HOLM-OLSEN (ed.), Konungs skuggsiá… (1983), p. ix.31 FINNUR JÓNSSON (ed.), Konungs skuggsjá… (1920), pp. 40–41.


The editors of the 1848 edition criticised the 1768 Sorøe edition for relyingon younger manuscripts rather than the older ones, and for using an orthogra-phy that was based on younger Icelandic manuscripts but not identical with anyof them. The 1848 editors preferred to use the oldest and best manuscripts, andfurthermore, to render the text in a uniform orthography suitable for the OldNorse language in what they perceived to be its best period.32 This period wasthe middle of the 13th century, which also was seen (and still must be seen) asa particularly expansive and culturally adoptive period in Norwegian history.Ill. 7 shows how the text was rendered in the 1848 edition; a comparison withthe corresponding passage in ill. 6 shows that the 1848 edition is much moreaccessible, albeit at the cost of being further removed from the manuscriptsthemselves.

The 1848 edition is typical of its time, but it should be remembered thatthis edition, like any other edition, must be understood in the context of othereditions of the same work. The editors point out that in spite of the manystrengths of the 1768 Sorøe edition, this edition was no longer readily available,and that due to its format, it was not a particularly handy edition for students.33

The 1848 edition was published as a supplement to the catalogue of universitylectures for 1848, and therefore had to be smaller and cheaper to produce thanthe typical 18th-century editions. The quality of the work is still highly com-mendable. Keyser and Munch had by that time built up a considerable editori-al expertise through the editing of the Old Norwegian laws, Norges Gamle Love,and several other works, and Unger later turned out to be the most prolific edi-tor of all Norwegian editors.

The 1848 edition could base its reconstruction of the work on the 1768 edi-tion, but the editors decided to make the orthography much more conservativethan that in the Sorøe edition. Both editions belong to the right-hand region ofthe editorial landscape in ill. 4, on the basis of their representation of the text ina uniform (if different) orthography and for the way in which they piece togeth-er a single work from a selection of texts.

74

ODD EINAR HAUGEN

32 «… en conseqvent Retskrivning, passende til det norröne Sprog i dets beste Periode,og ikke heller væsentlig afvigende fra den i de ældste benyttede Haandskrifter», R. KEYSER –P.A. MUNCH – C.R. UNGER (eds.), Speculum Regale…, p. xiii.

33 Ivi, p. xii.


4. The text that never was

The 19th-century editors may be criticised for having edited a text that nev-er was, as opposed to the text that actually has been preserved in the survivingmanuscripts. To be sure, there have been instances of excessive editorial inter-vention in the higher criticism of Eddic scholarship, with stanzas of somepoems, e.g. Hávamál and Atlakviða, being completely re-arranged, and littlerespect being paid to the documented textual transmission. There have alsobeen examples of younger manuscripts being discarded as having no critical val-ue on the basis of a questionable recension. That danger prompted GiorgioPasquali in his time to remind editors that younger manuscripts are necessarilynot of lesser quality – recentiores, non deteriores.34

However, this does not mean that the pendulum should swing to the other

75


Ill. 7. Extract from ch. 30 of the edition of Konungs skuggsjá by Rudolf Keyser, PeterAndreas Munch and Carl Richard Unger, Christiania, 1848, p. 66, l. 3–21. The orthographyhas been normalised by the editors.

34 G. PASQUALI, Storia della tradizione e critica del testo, 2nd ed., Firenze, Le Monnier, pp.43–108.


76

ODD EINAR HAUGEN

Konungs skuggsjá

En klæða bunaðe þinum hattat at þu ser klæddr allum goðum

ærm

þ n til hosna en æcki an-

mæð grœnum m oc þo goð klæðe oc sœmilegh en linklæðe þin

þijna og lijn klædi þijn el ætla iafnan godan mun stuttþijna en iri þui at eingi m hæuerskur n giera af hor e g þitt og hár sk ádur en þu gengur firi kon ir þeim sidum sem þá eru j hird er þu leitar til hird n þá eg jnnan hird þ sidur skorit stutt a en og kemt suo sijd m huer em huer

og sijd skor stutt firi n m skegg þ sidur þá at giora stutt skegg og snogguann og sijd jád skegg ersku og eigi er komi sa sidur at fegri sie e betur somi kongs herlidi at h n nu ber suo ikist buinn t kongs fund huortueggia at klæda bunadi og odrum hlutum og kemur þu j god tijma og jnngon m r-dur þu suo at hátta þ komu þinne at þier ckur lidligur þionustu m n þo at þid fá ir jnngon ecki

er leing en jnnann jnnar at stall a stoli hit leinog lát h giæta skickiu þinn

1

4

7

10

13

16

19

22

AM 243

AM 232 e fol

8 æ

2

———

pal

var


77


Ch. 30

-

-

-

1

4

7

10

13

16

19

22


78

ODD EINAR HAUGEN

Ill. 8 (pp. 76-77). Having one’s cake and eating it too. A specimen of a synoptic editionof Konungs skuggsjá with a monotypic text to the left and an eclectic text to the right. Theapparatus is divided in two, the first containing paleographical comments and the secondtextual variants. The translation at the bottom of the page is by Laurence Marcellus Larson,The King’s Mirror, New York, Twayne, 1917.

extreme, from the fanciful emendations and conjectures of high-spirited editorsto the uncritical veneration of any manuscript, whatever its quality. Can therein fact be a compromise between the reconstruction in many 19th-century edi-tions on the one hand, and the diplomatic approach found in many 20th-cen-tury editions on the other? In the editorial landscape laid out in ill. 4, the 1848edition of Konungs skuggsjá belongs to the right, while the 1945 edition falls tothe left. In a single-text edition, I do not think there is any compromisebetween these positions. The text must either follow one or more manuscriptsin their specific orthography or it must construct the text on the basis of sever-al manuscripts and render it in a uniform orthography. The exact nature of thisorthography may vary depending on the manuscripts, but for Old Norse man-uscripts it will usually be close to the standard orthography in Íslenzk fornrit orsimilar editions.

In ill. 8, I have tried to show how another type of compromise may bereached. This is a synoptic edition with a monotypic text on the left-hand page(based on the text in the 1945 edition, p. 45 l. 32 – p. 46 l. 10) and an eclectictext on the right-hand page, based on the 1848 edition, p. 66 l. 2 – l. 28). Forthe former text, I have made no changes to the 1945 edition other than makingthe column-width narrower and specifying in the margin which manuscript thetext is based upon. For the latter, I have changed the orthography slightly so thatit follows the standard as set down by the Ordbog over det norrøne prosasprog.35 Itwould have been possible to regularise the orthography according to the codexoptimus, AM 243 b α fol. This was in fact what Keyser and Unger did in theiredition of Barlaams ok Josaphats saga, in which the supplementary text taken fromyounger Icelandic manuscripts have been regularised to the East Norwegianorthography of the codex optimus, Holm perg 6 fol.36 However, since the ration-ale behind a normalised version is to make the text more accessible, I believethat the standard orthography should be chosen, unless there are compelling

35 Ordbog over det norrøne prosasprog, København, Den arnamagnæanske Kommission,1989ff. The major difference lies in the usage of accents for all long vowels, thus «ø» for «œ»and «æ» for «æ».

36 R. KEYSER – C.R. UNGER (eds.), Barlaams ok Jospahats saga, Christiania, Feilberg ogLandmark, 1851.


reasons to choose otherwise.The example in ill. 8 is confined to the two facing pages of a printed book.

In a digital edition of the text, this limitation is no longer crucial. However, thefact that a digital edition can align as many texts as there are manuscripts of thework should not tempt the editor to submerge users in an ocean of variants. Oneimportant task for the editor is to simplify and clarify the textual variation, dis-tinguishing between accidental variation and substantial variation. The acciden-tal variation should not be deleted, but the editor should consider carefully howmuch variation and how many versions of the text should be displayed simulta-neously. Assuming that Holm-Olsen’s recension of the manuscripts is correct,the critical apparatus to be displayed might be reduced to the substantial varia-tion in the 22 manuscripts thought to be of critical value. This is what Holm-Olsen did in the revised 1983 version, in which the variants from the 1945 edi-tion were replaced by a new selection, and no longer based on the 1920 editionby Finnur Jónsson.

In the preface to the revised 1983 edition, Holm-Olsen begins by pointingout that the 1945 edition should long have been replaced with an edition con-taining an exhaustive critical apparatus based on a re-evaluation of the manu-script material. He does not specify the exact nature of this edition, but I believethat he might have been sceptical of an eclectic text. The specimen in ill. 8 is arather conservative solution to the challenge of editing Konungs skuggsjá, but itmight prove to be a more realistic solution than a fully-fledged critical edition. Itis an edition which would find a position between the 1848 and the 1945 edi-tions, not too far from the centre of the editorial landscape set out in this article.It is not the text that never was, but a text that might have been.

79


* I would like to thank Dr. Emily Lethbridge, Cambridge, for her help with the Englishlanguage aspect of my article.


GIUSEPPE BRUNETTIUniversità di Padova

OLD ENGLISH POETRY ON THE WEB:A LEXICOGRAPHIC EDITION

Abstract. I give an overview of my lexicographic edition of Old English poems onthe Web. The texts (either singly, like Beowulf, or in groups, like the Elegies) areaccompanied by interlinear glosses, glossaries (Old English-Italian and Italian-OldEnglish), concordances by lemmas, lists of word forms (with their matching lemmas,homographs distinguished), lists of compounds and poetic words, of words govern-ing cases and/or clauses, statistics of lemmas and word classes, and graphs of word-class distribution of each poem (or group of poems) in relation to the corpus. Allthese are shown in different frames of the same window, together with a superglos-sary, that is a glossary of all the poems tagged so far (13,044 lines = 73,548 wordforms = 7,551 lemmas, over one-third of the whole corpus). The Old English-Italianglossary, the concordance, the list of word forms and the Italian-Old English glossaryare hyperlinked to each other and are linked to the line numbers of the poem. In thewindow of the hyperlexicon, instead, the links go in the opposite direction: one canclick any word anywhere in the poem (or group of poems) and visualise its gloss (lexi -cal information in context), and then the full entries of glossary (Old English-Italianand Italian-Old English), concordance and list of word forms.

I will give an overview of my Old English poetry on the Web, which I call alexicographic edition because its aim, to use an expression from Beowulf, is toopen up the word-hoard of the poems and display their lexicon in a variety ofrepresentations, from glosses to graphs – the glosses giving full morpho-syntac-tic information on each word in context, the graphs showing the distributions ofword classes across the corpus and its lexical space.1

81

1 Poesia antico inglese, ed. by G. BRUNETTI, <http://www.maldura.unipd.it/dllags/brunet-ti/OE/begin.htm>, 2006– (latest update: 25 August 2008).


1. XML markup

The representations are extracted from the following information embeddedin the poems themselves, word by word, in XML markup:

W word forml lemmac word classo homographys syntax (government of verbs, nouns, adjectives, prepositions)g grammar (morphology)t translation (Italian)

Thus the tagging<W l=”gefrignan” c=”v” s=”a+” g=”p1p” t=”apprendere”>gefrunon</W>

reads: the word (W) gefrunon is a form of the lemma (l) gefrignanits class (c) is the verb (v)it governs (s) an accusative and a complement clause (a+)its morphology (g) is preterit 1 plural (p1p), and in the context its Italian translation (t) is apprendere

The homography attribute, o, distinguishes homographs of the same wordclass,<W l=”metan” c=”v” o=”o1” s=”a” g=”p2p” t=”misurare”>mæton</W>

<W l=”metan” c=”v” o=”o2” s=”a” g=”p3s” t=”incontrare”>mette</W>

The attribute is used also to mark together instances of a noun with doublegender (which is tagged as c, and by default homographs with different gendersare different lemmas),

<W l=”sæ” c=”m” o=”o1” g=”as” t=”mare”>sæ</W>

<W l=”sæ” c=”f” o=”o1” g=”as” t=”mare”>sæ</W>

The tagging was done manually in the first stage; then a semi-automatic lem-matizer was developed for the purpose. It draws on a database of tagged words,and when a new word is proposed for tagging it shows all the ways in which theword has been tagged before, so that the right one can be chosen or a proximateone modified, or an entirely new tagging introduced. The new and modifiedtaggings are added to the database for immediate re-use.

82

GIUSEPPE BRUNETTI


The lemmatizer also validates new or modified tagging by cross-checkingclass and morphology.

Over one-third of the whole corpus has been tagged so far (August 2008),and edited as single poems (Andreas *, Ascension *, Battle of Maldon, Beowulf,Christ in Judgement *, Dream of the Rood, Elene, Fates of the Apostles, Genesis A,Genesis B, Guthlac A *, Judith, Juliana *, Phoenix) and groups of poems (AdventLyrics *, Bestiary, Charms *, Elegies, Gnomic Poetry, Minor Heroic Poems, Riddles *).Asterisked poems are not yet complete. The tagged corpus amounts to 13,044lines = 73,548 words = 7,551 lemmas.

2. Representations in windows and frames

The opening window gives the list of the poems: those that do not changecolours under the cursor are in preparation. Each poem (or group of poems) canbe opened in two windows, one with the hyperlexicon, the other with the lexi-cal listings and the statistics.

It is possible to pass from one window to the other, and also to return to theopening, or ‘main’, window.

2.1. The hyperlexicon window

The hyperlexicon is devised to give a profile of each word in the poemselected.

Figure 1. Window of Hyperlexicon

83

Old english poetry on the web

text concordance

gloss

glossary list of word forms

Italian-Old English glossary


Its text appears in the top left frame and is clickable. And by clicking anyword anywhere one can visualise its gloss underneath (lemma, class, morphol -ogy, syntax, Italian translation); and by clicking on the various parts of the glossone can have, in the other frames, the entries of glossary, concordance (by lem-mas), list of word forms and Italian–Old English glossary.

2.2. The listings window

The listings window has a three-tiered menu at the top that gives the options,with their targets in the five frames below (cross-labelled). The frames can be re-sized with the cursor.

Figure 2. Window of Listings

84

GIUSEPPE BRUNETTI

A

B1 B2

C1 C2

Atext

text with interlinear glossestext with Italian translation

statistics

B1glosses

glossarygovernment

B2concordancepoetic wordssuperglossary

C1glossary

list of word formsnotes

C2Italian-Old English glossary

compoundsabbreviations

2.2.1. GlossesThe glosses – either interlinear in the same frame with the text or by them-

selves with the text in a different frame – are a glossary in situ, with the lemmasin the sequential order of the words in the line and with full lexical informationaccording to context. The morpho-syntactic abbreviations can be expanded bypositioning the cursor on them.


2.2.2. GlossaryA full analytic glossary of all the lemmas and their forms, with the Italian

translations. The morpho-syntactic abbreviations are expanded by positioningthe cursor on them.

2.2.3. List of word formsEach word form (with variant spellings) is matched with its lemma, or lem-

mas in case of homography.

2.2.4. Italian–Old English glossaryA list of the lemmas used for the translation attribute t.

2.2.5. ConcordanceIt is by lemmas (with class and frequency), alphabetically ordered by forms.

Homograph lemmas of the same class have distinct entries.

Glossary, list of word forms, Italian–Old English glossary and concordanceare hyperlinked with each other. By clicking a lemma of the glossary in oneframe one can see its concordance in another, and the other way round; and sofor glossary and list of word forms, and glossary and Italian–Old English glos-sary. All the numbers in the four listings are linked with the lines of the poem,and by clicking them one can see the contexts.

2.2.6. Compounds and poetic wordsThe compounds (with class and frequency) are listed under the simple lem-

mas that make them, either as first or second part. The poetic words (thosefound only in poetry) are linked to the lines in which they appear.

2.2.7. GovernmentFive listings give information on the government (nominal and clausal,

valency and case) of verbs, nouns, adjectives and pre-/post-positions. Class andmorpho-syntactic abbreviations are expanded by positioning the cursor onthem.

2.2.8. SuperglossaryThis is the glossary of the corpus tagged so far (also the abbreviations of the

poems are here expanded by positioning the cursor on them).

85



2.2.9. Statistics and graphsSeveral frequency lists of lemmas are provided, ordered both alphabetically

and by rank, with comparison with the corpus: a general list of all the lemmasand lists of the four main lexical categories (nouns, verbs, adjectives andadverbs).

Word class distribution: numbers and percentages of total lemmas and formsfor each category; and the same for compounds and poetic words. The distribu-tion is also represented graphically, with the possibility of comparing poems.

In order to make poems of different lengths comparable, the statistic used isbased on the difference between theoretical and actual frequency. If a poem’stotal word forms are, say, a fifth of those of the corpus, the theoretical frequencyof its nouns is a fifth of the nouns of the corpus: the difference between actualand theoretical frequency shows whether the poem has an excess or a deficiencyof nouns – whether it is above, below or within the norm. The differences of thevarious word classes are then standardized to make their incidence comparableand show whether they are significant or not: they are if they exceed ±2 (twostandard deviations). All this gives a profile of the lexical style of the poems.

Table 1. Distribution of word classes, compounds and poetic words

86

GIUSEPPE BRUNETTI

n v a av np p pp c d nu e cp po

and 2.50 –1.27 1.06 –2.04 –4.61 0.37 3.18 –0.86 –2.66 –1.25 0.34 7.17 5.08

beo 0.92 2.28 0.53 1.10 13.45 –3.70 –5.40 –1.67 –2.71 –3.64 –0.89 13.50 13.61

best 0.73 –2.57 3.43 0.12 –3.70 –3.01 1.86 –0.13 3.57 –0.62 –0.73 2.51 0.74

dream –2.53 1.03 –0.10 1.11 –4.16 3.25 1.82 –1.77 1.05 –1.49 1.63 –3.92 –5.20

eleg –1.25 3.25 0.60 2.09 –7.50 2.33 –2.45 –0.32 –1.07 –3.57 2.58 1.21 –0.55

elen 2.79 –2.08 –1.82 –0.18 –3.23 –5.84 6.34 –2.63 5.80 1.01 1.94 0.49 0.43

genA 3.34 –3.61 –3.35 –1.77 8.68 0.58 –2.08 1.54 –3.03 9.71 –1.74 0.54 –1.35

genB –7.44 1.88 –5.16 1.25 –6.85 14.32 0.00 1.45 2.20 –0.94 1.54 –10.71 –11.16

gnom 3.12 3.32 4.76 –3.01 –9.06 –1.22 –1.20 –1.70 –5.60 –0.79 –1.59 –2.15 –1.03

guthA –1.63 1.10 –2.80 –2.56 –5.00 6.07 –0.53 1.55 3.07 –1.94 –0.52 –7.53 –10.70

heroic –5.31 –2.97 –2.01 –0.06 31.61 –2.36 2.55 2.65 –2.01 –0.40 –0.13 –1.11 –1.28

jud –1.36 –2.27 5.00 2.29 –2.98 –2.28 0.11 –1.51 6.03 –1.94 –1.10 3.46 5.81

mald –4.74 3.44 –4.04 2.01 2.43 1.90 1.40 0.09 2.01 –0.73 –0.19 –6.04 1.87

phoe 5.75 –4.29 4.91 –0.19 –8.83 –9.76 2.48 2.41 3.42 –2.73 –1.50 –1.82 –2.22

rid –1.88 2.91 0.23 –0.17 –9.46 4.59 –1.54 0.98 –2.18 4.08 –1.55 –5.44 –1.78


n=nouns, v=verbs, a=adjectives, av=adverbs, np=proper nouns, p=pronouns, pp=prepo-sitions, c=conjunctions, d=demonstratives, nu=numerals, e=interjections, cp=com-pounds, po=poetic words;

and=Andreas, beo=Beowulf, best=Bestiary, dream=Dream of the Rood, eleg=Elegies,elen=Elene, genA=Genesis A, genB=Genesis B, gnom=Gnomic Poetry, guthA=Guthlac A,heroic=Minor Heroic Poems, jud=Judith, mald=Battle of Maldon, phoe=Phoenix,rid=Riddles.

The data can be graphed in a number of ways. Here is a barplot comparingthe distribution of nouns, verbs, adjectives, adverbs, compounds and poeticwords in some of the poems:2

Figure 3. Distribution of lexical word classes, compounds and poetic words

87


2 This and the following graphs have been specially drawn for this presentation using theR language and environment (R Foundation for Statistical Computing, Vienna, Austria.ISBN 3-900051-07-0, URL <http://www.R-project.org>).


Notice the abundance of compounds and poetic words in Beowulf, and thecorresponding scarcity in Genesis B. A verbal style characterizes Maldon, withverbs and adverbs above and nouns and adjectives below zero (the norm). Withan abundance of nouns and adjectives Phoenix has a nominal style, whereasGenesis B is denominalized. Judith is rich in adjectives, and Gnomic Poetry innouns, adjectives and verbs.

And here is the distribution of the other word classes.

Figure 4. Distribution of proper nouns, function word classes, numerals and inter-jections

88

GIUSEPPE BRUNETTI


The denominalized Genesis B is highly pronominalized, whereas Genesis Aabounds in proper nouns (but less than Beowulf ) and numerals. The scarcity, inBeowulf, of pronouns, prepositions, conjunctions and demonstratives is a clue toan elliptic, asyndetic, paratactic style – partly like that of Gnomic Poetry andunlike that of Phoenix.

The lexical variables (=word classes, compounds and poetic words) can bethought of as the coordinates of the poems in a multidimensional space. The sta-tistic of multidimensional scaling is then used to reduce the dimensions to twoand give a plane representation of the distances between the poems mirroringthe distances in the original multidimensional space. This is the lexical space ofthe corpus.

Figure 5. Lexical space

89



The lexical variables are also graphed (in grey); they indicate the directionsfrom the centre in which they locate the poems according to their weight in eachof them. The denominalized Genesis B is dominated by pronouns, as are theMinor Heroic Poems by proper nouns (because of Widsith). The position ofBeowulf is determined, along the vertical axis, by compounds and poetic words,and by proper nouns along the horizontal axis. Nouns and adjective dominatePhoenix.

2.2.10. More dataMore data can be drawn from the frequency lists: for instance, the distribu-

tion of modal verbs, seen as part of the characteristic lexicon of a poem.

Figure 6. Distribution of modal verbs

90

GIUSEPPE BRUNETTI


Gnomic sculan dominates Gnomic Poetry, while cunnan is the characteristicmodal of Elene and magan of the Elegies. Genesis A and Genesis B are roughly sym-metrical. The heroic code of Maldon is appropriately modalized by willan andmotan – what is willed and what is allowed. The legend lists the modals in theirdescending order of frequency in the corpus.

Information on prepositions can be gleaned from glossaries and superglos-sary: which ones are only pre- (æt, ofer), or only post- (neah, togeanes), or both(on, to). Postpositions give rise to discontinuous phrases with one or more inter-vening words, or even verses.

Information on the valency of verbs is given in the government listing ofeach poem. Further processing of the tagged texts provides the general patternof the corpus. Its 1,598 lexical verbs can be divided into two classes: those thatare used always in one and the same valency (either with or without comple-ments) and those that alternate between two or more valencies.

Table 2. Verb complementation and valency alternations

91


valency lemmas forms

fixedverbs with zero complements 489 3057

verbs with one or more complements 564 1459

variable verbs with a varying number of complements 545 7859

tot. 1598 12375

The second class has about one-third of the lemmas and almost two-thirds ofthe forms. The verbs which have most alternations are biddan, (ge)cyðan, gedon,gelæran, gesecgan, geweorðan, hatan, scyppan, secgan, unnan (their complemen-tation varies from zero to three, as can be checked in the superglossary).

3. What next?

The work ahead is the extension of the corpus and the correction of the tag-ging (Beowulf has already been revised twice).

A development which I envisage at the moment is that of comparing differenteditions of the same poem; and I am working on a lexico-grammatical collation ofeleven editions of Beowulf, from Klaeber 3rd 1936 to Klaeber 4th 2008, with aview of producing an apparatus of variants of form, lemma, morphology, govern-ment and translation, and a representation of the lexicographic space of the poem.


PAUL GABRIELE WESTONUniversità di Pavia

DIGITAL TEXT ARCHIVES:A LIBRARIAN’S APPROACH

Abstract. The current digitisation landscape amply demonstrates that the library sec-tor should place the electronic conversion of its documentary heritage and the devel-opment of a strategy for achieving this high on its agenda. The exploitation of therichness of these bibliographic sources will necessitate revisiting fundamental ques-tions about the nature of library OPACs and the ways in which they function. Thefollowing paper investigates some of the potential and, indeed, challenges associatedwith using structured texts as components of an interoperability tool. The issue of along-term preservation strategy based on the use of appropriate metadata schemas isalso tackled.

We are currently witnessing a period of profound change, particularly withrespect to the way in which information circulates and texts are transmitted.Concepts such as the ‘digital age’, the ‘information age‘ and the ‘electronic age’are at the centre of the debate taking place among specialists engaged in an effortto provide definitions, circumscribe important areas, and identify futureprospects. Particularly in relation to the last item on this list – namely, the sce-nario that will be encountered by scholars, librarians and web users in the next15-20 years or so – each of the individuals who have contributed to the debatehas expressed his or her own view, and these views have sometimes been verydifferent. This situation explains why the librarian community currently appearsto be in a state of general confusion as regards the creation of digital content; thisconfusion has been caused by a lack of shared policies and the adoption of com-mon strategies. The following paper addresses the conversion into digital formof cultural heritage composed of bibliographic, documentary and other kinds ofresources held in institutional repositories.

93


It is altogether clear by now that digital technology – including the infra-structure designed to transmit and exchange data – has profoundly changed theway in which texts are accessed and used. These texts are, by now, significantlymore important to the majority of readers than books. In the course of the debateon the contemporary value of the traditional book and its enduring efficiency asa communication tool, there is a tendency to become confused about the mean-ings of the words ‘text’ and ‘book’, and to consider the terms more or less syn-onymous. Without detracting in the least from the importance that the bookscan represent as study objects, or their aesthetic value and codicological charac-teristics, not to mention the rich source of information and hints that marginalannotations and ownership notes can provide for historians and philologists, thisblurring of definitions is indicative of a misapprehension that librarians aresomewhat susceptible to when interpreting the primary objective of their voca-tion: they fail to fully grasp that readers are chiefly concerned with texts, notbooks in themselves. To the vast majority of readers books are only importantinasmuch as they contain texts. The great advantage that libraries offer is thepossibility of assembling a large body of books and therefore making it possiblegain access to a vast number of texts – not only those of which the reader isalready aware of the existence of, but also many more that he or she will becomeaware of through quotes, reviews, and other more or less direct references. Untilabout twenty years ago, the immediate proximity of vast stores of heritage heldin national libraries and other extensive research collections offered an unques-tionable advantage to those who happened to live in capital cities and towns thathost major universities. Today, in the Digital Age, a text need only be hosted ona server connected to the Internet in order that anybody can gain access to it bymeans of a few keystrokes, thus eliminating the need to travel to the librarywhere the original text is kept. The fundamental importance of the text is there-fore unchanged, and in all likelihood this state of affairs will remain the caseuntil such time a method of transmitting information and knowledge that doesnot chiefly rely on the written word becomes firmly established.

Whether we like it or not, the technology that allows one to convert ourentire documentary patrimony into digital form already exists and wouldbecome widely employed were it not for the considerable costs involved, whichare simply not sustainable given current levels of funding. Even those who claimthat they do not want to give up in any way the possibility of printing a docu-ment prior to reading it are generally prepared to recognize the advantagesgained from immediate access to a text, and favourably view the allotment offunds for the immediate launching of digitisation campaigns. The concept of ahybrid library implies providing for the balanced coexistence of analogue and

94

PAUL GABRIELE WESTON


digital documents and making them available for readers to consult. As DavidPearson wisely observes, what we need now «is a clear strategy for digitisingimportant texts, the core of primary research material».1

The innumerable projects aimed at developing new and more efficient accesstechnologies to bibliographic resources have required a great deal of energy andlarge sums, as well as broad collaboration at various levels between interestedinstitutions. Among the various methods to be considered, web-based OPACsoccupy a particularly important place. However, these only represent a first steptowards accomplishing a far more onerous task: that of creating full-text digitalcontent. According to Kurt Gärtner,

if texts are at the centre of a text archive, then it is not enough to scan the completeholdings of special libraries, digitise the image and create a huge amount of metadatain order to access the images of the book pages. The traditional librarian surely willprefer this way of digitising a library collection, and as a result of this attitude, nation-al funding schemes have preferred projects which did not really leave the world of theprinted page. Image digitisation is still much more popular and receives more fundingthan projects aiming at full-text digital versions of primary research material.2

In the area concerned with realizing textual digital archives the panoramaappears far more fragmented. The bulk of projects completed up until the pres-ent which have involved libraries and academic consortia have tended to be ofsmall to medium size and focused on specific parts of collections, in addition toeditions containing texts chiefly of a literary nature. Even if the specific motiva-tion behind individual projects has been different, they have generally had incommon the fact that they have all depended on ad hoc funding, rather thanbeing guided by a fully developed long-term strategy, to the extent that in thereport titled Scoping the future of the University of Oxford’s digital library collections(1999), the observation is made that «most of the initiatives have been under-taken in isolation, coming up with different answers to the same questions, orsuffering from the familiar problem of reinventing the wheel».3

95

Digital Text Archives

1 D. PEARSON, Digitisation: do we have a strategy?, «Ariadne», 30, Dec. 2001,<http://www.ariadne.ac.uk/issue30/digilib/intro.html> (date of access: 20 Jan. 2009)

2 K. GÄRTNER, Comprehensive Digital Text Archives: A Digital Middle High German TextArchive and its Perspectives, First EU/NSF Digital Libraries All Projects Meeting, Rome, March25-26, 2002, <http://delos-noe.iei.pi.cnr.it/activities/internationalforum/All-Projects/RomeSlides/DTArchives.pdf> (date of access: 6 Feb. 2009).

3 S.D. LEE, Scoping the Future of the University of Oxford’s Digital Library Collections, FinalReport, Sept. 1999, <http://www.bodley.ox.ac.uk/scoping/report.html> (see section 7.1 TheNeed for Co-ordination: Projects into Services) (date of access: 20 Jan. 2009).


The consequence, then, is a chaotic mass of digital content, frequently char-acterised by poorly thought out basic principles and not a little redundancy. Thewebsites of libraries present digital reproductions of covers, title pages, fron-tispieces, indexes, portions of texts, etchings and other types of illustration creat-ed for exhibitions and other important events in the history of the institution.These snippets are offered as samples of things that are generally referred to as‘treasures of the library’ – an expression that tends to turn holdings into museumpieces, rather than treat them as documents to be read, viewed or studied as liv-ing components of a system intended to promote the circulation and dissemina-tion of knowledge. Sometimes the resources lack proper analytical cataloguingwithin catalogues, and occasionally they are to be found tucked away where itwould not generally occur to readers to look for them; in both cases there existsa risk of under-utilization of products that took a great deal of effort to create.

During the course of librarianship history the most successful initiatives havebeen those that succeeded in matching an application to the greatest number ofsubjects by means of relative functional simplicity, so as to achieve comprehen-siveness and inclusiveness of coverage. Short-title catalogues, which census allthe publications of the world’s leading countries within specific chronologicalperiods, together with current national bibliographies generated by insertingcollected data in the data grid known as International Standard BibliographicDescription (ISBD),4 represent the corner stones of interdisciplinary research,just as – albeit in a different way – the MARC format5 used for the creation ofelectronic bibliographic records also represents a fundamental building block.

Since the age of the Alexandrian Library there has existed a bold ambition toamass under one roof the greatest possible amount of knowledge and to devise ahandling system capable of rendering that ‘universe’ of knowledge readily acces-sible to readers. With the spread of printing and consequent massive increase inthe number of books , this ambition has proved to be ever less realizable, even ifthere has been no lack of attempts to generate records of the range of man’sknowledge by means of coordinates as part of a logical system (Gesner’s

96


4 IFLA, ISBD(G): General International Standard Bibliographic Description, Annotated Text,rev. ed. prepared by the ISBD Review Committee Working Group set up by the IFLACommittee on Cataloguing, München, K. G. Saur, 1992. See also IFLA, InternationalStandard Bibliographic Description (ISBD), Preliminary consolidated edition, München, K. G. Saur,2007, <http://www.ifla.org.sg/VII/s13/pubs/ISBD_consolidated_2007.pdf> (date of access:20 Jan. 2009).

5 Library of Congress, Network Development and MARC Standards Office, MARCStandards, <http://www.loc.gov/marc/> (date of access: 20 Jan. 2009).


Bibliotheca Universalis), or physical system (the organization of the BritishMuseum’s Library by Panizzi, which commenced with the compilation of thecatalogue of printed books). The same ambition informs IFLA’s vision when itpromotes projects, such as Universal Bibliographic Control or UniversalAvailability of Publications,6 which are both coordinated managing systemsdesigned to be applied to bibliographic patrimonies and aim at allowing anyoneto gain access to any document, wherever it is available. The digital environmentnow offers the possibility of creating omni-comprehensive collections of texts invirtual form. By now virtually everyone has been persuaded of the impossibilityof concentrating within the confines of a single institution – no matter howprestigious and resource rich – a collection of this kind and corresponding mag-nitude. Yet, libraries do not seem to fully grasp the importance of the phenom-enon and the opportunity that electronic supports offer, as, conversely, the pri-vate sector has; indeed some private firms have launched major projects andimportant campaigns which involve a considerable degree of commitment. Afull-text digital version of the publications produced in the English language upuntil the year 1700 has been made available by ProQuest (Early English BooksOnline),7 the same company that manages the Chadwyck-Healey LiteratureOnline 8 database. All told, one is effectively dealing with a patrimony composedof over 250,000 literary texts. A project of comparable relevance goes by thename JSTOR;9 this project is concerned with the digital reproduction of period-

97


6 The National Library of Australia, the Library of Congress, The British Library, theKoninklijke Bibliotheek, and the Deutsche Nationalbibliothek have agreed to participate ina joint alliance together with the Biblioteca Nacional de Portugal, IFLA and CDNL to assureongoing coordination, communication and support for key activities in the areas of biblio-graphic and resource control for all types of resources and related format and protocol stan-dards. This new alliance is known as ‘IFLA-CDNL Alliance for Bibliographic Standards(ICABS)’, <http://www.ifla.org/VI/7/icabs.htm> (date of access: 6 Feb. 2009).

7 From the first book printed in English by William Caxton, through the age of Spenserand Shakespeare and the tumult of the English Civil War, Early English Books Online(EEBO) will contain over 125,000 titles listed in Pollard and Redgrave’s Short-TitleCatalogue (1475-1640), Wing’s Short-Title Catalogue (1641-1700), the Thomason Tracts(1640-1661), and the Early English Tract Supplement – all in full digital facsimile from theEarly English Books microfilm collection, <http://eebo.chadwyck.com/home> (date ofaccess: 12 Dec. 2008).

8 Over a third of a million full-text works of poetry, prose and drama in English, togeth-er with online criticism and reference library, <http://lion.chadwyck.com/marketing/index.jsp> (date of access: 12 Dec. 2008).

9 JSTOR offers a high-quality, interdisciplinary archive to support scholarship and teach-ing. It includes archives of over one thousand leading academic journals across the humani-


icals, especially – but not exclusively – those addressing the humanities, therebyguaranteeing them maximized distribution. Even if it is not fundamentallyinspired by commercial ends, this particular initiative follows common manage-ment criteria, functioning like an independent not-for-profit organization.10

One could readily argue that the general attitude of libraries reflects their tra-ditional role as custodians, more than publishers, of bibliographic patrimony,the latter task being performed by commercial enterprises. Even if this view iswell-founded, there can be no doubt that the introduction of electronic technol-ogy has resulted in abrupt and important changes to the traditional workingmodel seen in the library sector (just as it has in others). The risk being run bylibraries is that of being cut out of the mechanism which serves to organize andcirculate knowledge as soon as publishers are able to directly provide users withdocuments that were traditionally always held in libraries. This developmentmight be inevitable in the long-term, given the general tendency to shorten thecommercial distribution chain and to devise economy of scale approaches byachieving critical masses of products. Nonetheless, libraries should reflect on thefact that their role as depositories of documentary heritage places them in theposition of holding the lion’s share of the materials that scholars and readersrequire, and this position should make them able to assume a more incisive roleas regards the creation of digital content. This is precisely what Karen Calhounrecommends in her much discussed report to the Library of Congress on theChanging nature of the catalogue and its integration with other discovery tools.11

There is a further argument in favour of greater involvement in the publica-tion of textual archives by national and public libraries, and those concernedwith training and education. All of them fulfil their institutional duty of makingavailable to their specific user group the documentary materials it requires, onthe one hand by establishing repositories in which materials are kept, makinguse of the techniques necessary to guarantee medium to long-term conservation,

98


ties, social sciences, and sciences, as well as select monographs and other materials valuablefor academic work. The entire corpus is full-text searchable, offers search term highlighting,includes high-quality images, and is interlinked by millions of citations and references.Journals are always included from volume 1, issue 1 and include previous and related titles.The most recently published issues (past 3-5 years) are not available, <http://www.jstor.org/>(date of access: 12 Dec. 2008).

10 On the topic of mass digitization projects see G. RONCAGLIA, I progetti internazionali didigitalizzazione bibliotecaria: un panorama in evoluzione, «DigItalia», 1, giugno 2006, pp. 11-30.

11 K. CALHOUN, The Changing Nature of the Catalog and its Integration with Other DiscoveryTools, prepared for the Library of Congress. Final Report: March 17, 2006,<http://www.loc.gov/catdir/calhoun-report-final.pdf > (date of access: 12 Dec. 2008).


and on the other by providing services – a catalogue, consultation and referenceservices – which serve to maximize the possibility that every user (employingvarious methods depending on his/her age, cultural level and practical aims) willbe able to find the information that he or she seeks, in line with Ranganathan’slaw «Every reader his book».12

We have believed up until the present day that it is in the public’s interest tomake available national collections such as the British Library or the nationallibraries located in Rome and Florence, the viability of which depends on publicsubsidy, inasmuch as these libraries make their books freely available to their par-ticular reference communities. If we view their mission from a digital perspective,we have to conclude that these libraries should make arrangements to providedigital versions of their books (and the texts they contain) pro communi doctorumvirorum commodo, in accordance with the commitment made by Pope Nicolas Vwhen the Vatican Library was established hundreds of years ago (in 1451)13 – amission that Leonard Boyle, who has served as the Prefect of the aforementionedlibrary in recent years, was proud to recall and, indeed, faithfully fulfil.

The massive demand for digital content is abundantly demonstrated by thegrowth of, and number visitors to, sites like Oxford Text Archive,14 theHumanities Text Initiative of the University of Michigan15 and the huge hold-ings of the University of Virginia’s Electronic Text Center.16 In Italy the BibItdatabase,17 which holds in excess of 1,700 texts (and which in recent years hassubstituted the earlier Progetto CiBit and extended its objectives), is rapidlygaining renown. No matter how ambitious a project launched by any singleinstitution or consortium is, it must always be seen as a contribution to anundertaking of much greater dimensions, like an individual tile in a vast anduniversal mosaic that we need to see completed. The long-term objective mightbe seen as setting out to realize a central archive (not necessarily physical, butrather virtual) in which metadata are amassed, namely descriptions and accesspoints, which include direct references to the urls of individual digital items.The Gallica18 digital library that aggregates, indexes and organizes according to

99


12 S.R. RANGANATHAN, The Five Laws of Library Science, Madras, London, The MadrasLibrary Association, Edward Goldston, 1931.

13 E. MÜNTZ – P. FABRE, La Bibliothèque du Vatican, Paris, Thorin, 1887, pp. 47-48.14 <http://ota.ahds.ac.uk/> (date of access: 20 Jan. 2009).15 <http://www.hti.umich.edu/> (date of access: 20 Jan. 2009).16 <http://etext.lib.virginia.edu/stats/> (date of access: 20 Jan. 2009).17 <http://www.bibliotecaitaliana.it/> (date of access: 20 Jan. 2009).18 <http://gallica.bnf.fr/> (date of access: 20 Jan. 2009).


thematic reading routes, and offers for consultation files created by theBibliothèque Nationale de France and a certain number of other institutions thatparticipate in the consortium, represents an important model of an architectureof this kind. The new Europeana19 digital library, that cooperates with partnersin various European countries and has significantly increased the range of class-es of documents represented, can be placed within the sphere of a very similardevelopment logic.

The notion of strategic coordination at a national level,20 enthusiasticallybacked by high calibre scholars such as Anthony Grafton21 and RobertDarnton,22 as much as it might initially seem excessively ambitious and ratherinsensitive towards autonomy in decision making on the part of single librariesand institutions, reveals a concern that the continuing climate of incertitudemight lead to a progressive erosion of libraries’ identities and an irrevocablereduction of the resources necessary for their proper functioning. Without ashadow of doubt it will be necessary for the library sector to place among its toppriorities in the immediate future an agreement that will make it possible to acti-vate appropriate mechanisms for the digital conversion of bibliographic patri-mony (in addition to strategies that will lead to the ultimate fulfilment of thisobjective).

For a long time catalogues have played a key role in the fulfilment of alibrary’s chief aim: namely, to save the reader’s time and to distribute within thelibrary’s reference community an awareness of the degree of progress made inacquiring knowledge in many different fields of study. Placing readers and infor-mation in contact with each other – another of Ranganathan’s laws: “Every bookits reader”23 – the catalogue, long before Google and other search engines exist-ed (and, we might be tempted to say, thanks to much its more efficient function-ing) has fulfilled its fundamental supporting role by providing unmediated

100


19 <http://www.europeana.eu/portal/> (date of access: 6 Feb. 2009).20 «to seek a national strategy for digitising existing collections of primary research mate-

rial» (B. FOLLETT, Just how are we going to satisfy our research customers, «LIBER Quarterly», 11,2001, pp. 218-223, p.222). «How much better it would be if there was a shared and central-ly managed programme [for digitising texts] across the nation» (R. LESTER, A few irreverentthoughts on ‘digitisation’…, «SCONUL Newsletter», 20, Autumn 2000, pp. 5-7).

21 A. GRAFTON, Future Reading: Digitization and its discontents, «New Yorker», November5, 2007, <http://www.newyorker.com/reporting/2007/11/05/071105fa_fact_grafton> (dateof access: 12 Dec. 2008).

22 R. DARNTON, The Library in the New Age, «New York Review of Books», 55/10, June12, 2008, <http://www.nybooks.com/articles/21514> (date of access: 12 Dec. 2008).

23 S.R. RANGANATHAN, The Five Laws…


information discovery and access. Today catalogues have to contend with amuch reduced use of bibliographical collections, extremely rapid and powerfulsearch tools, incessant changes in technologies, ever increasing expectations onthe part of knowing users, the hybridisation of documentary collections, theprogressive growth of electronic publishing, and new forms of knowledge trans-mission at the academic level. If we presuppose that the value of library collec-tions remains intact, we must concern ourselves with equipping research toolswith a functionality capable of fulfilling the tasks that we entrust to them. Wemust be adequately aware of the fact that because gaining access to the contentsof collections over time is only guaranteed by catalogue data, the lack of aneffective future strategy for catalogues will jeopardise the vast patrimony repre-sented by collections of books across the world.

Electronic catalogues, or so-called OPACs, still play a fundamental role bypermitting access to library collections, maintaining a consistent and authorita-tive form of bibliographic control, and providing a targeted information envi-ronment for specific user groups. Nonetheless, in recent years many articles haveappeared in which appeals have been made for changes to be made in order tofulfil the expectations of users accustomed to easy-to-use Web search engines,online bookstores, and seamless linking to full-text. In his 2005 paper producedfor the International Federation of Library Associations (IFLA), John Byrumfrom the Library of Congress wrote of the need for library catalogues to provideaccess to more content and to offer significantly enhanced functionality basedon the features of popular search engines. «This is a time when it is more impor-tant than ever to position the research library catalogue successfully within arapidly evolving information universe for scholarly research, teaching, andlearning, and to adapt to sea changes in information seeking behaviour».24 Agrowing number of students – and, alas – researchers have got into the habit ofbypassing library catalogues in favour of other types of search tools. Catalogues,as a consequence, carry less weight in the information-based world of research.The results obtained by conducting interviews and analysing literature suggestthat the cost-effectiveness of cataloguing tradition and practice is under fire, andthat a typical research library’s catalogue’s most valuable feature is its support forinventory control and as «last mile» technology to achieve delivery of the

101


24 J.D. BYRUM JR., Recommendations for urgently needed improvement of OPAC and the role ofthe National Bibliographic Agency in achieving it. World Library and Information Congress. 71thIFLA General Conference and Council ‘Libraries - A voyage of discovery’ (Oslo, Norway,August 14th - 18th 2005), <http://www.ifla.org/IV/ifla71/papers/124e-Byrum.pdf> (date ofaccess: 12 Dec. 2008).


library’s assets into the hands of local users.25 The main problem with this is thatresearch library online catalogues reflect only a small fraction of the vast uni-verse of scholarly information. Information seeking studies in libraries havetended to focus on information sources and systems rather than the people whouse them. Worthwhile research has been carried out in order to identify areaswhere catalogues can be improved, but as yet these improvements have scarcelybeen implemented.26

The development of web OPACs around the mid-1990s made it possible forusers to carry out research from a distance by means of an easy-to-use interface.A few years after this an extension of this facility also permitted, using the samesign-on, advanced searches of web resources, in tandem with those carried outon local catalogues, electronic periodicals and other digital resources created bylibraries. The tools currently in use allow readers – either on the spot or remote-ly – to enter a reference query through electronic reference services (the ask alibrarian service function) and submit a loan request with the same login. Theconcept of an online catalogue has developed little by little as web OPACs havetransformed into portals in the true sense. The new generation library catalogueperforms at least three roles. Firstly, it functions as a bibliographic archive, theelectronic version of the traditional paper-supported card index that it replaced,thereby furnishing an index to holdings for users who are searching for particu-lar items. As a logical extension of this role, OPACs with increasing frequencyprovide links to electronic texts, thus liberating the user of the need to locatepaper-supported examples of works on a library’s shelves.

Secondly, acting as a portal – in a way not so different from that of a library’shomepage – it manages links to non-bibliographic data relating to the readersthemselves (e.g. information about overdue books, fines, etc.) or the functioningof the library itself (e.g. opening hours, special events, job openings, etc.) Thisfunction could be limitlessly expanded by connecting data considered to be ofinterest to readers.

The third role, known by the term ‘bibliographic enrichment of the cata-logue‘, consists in the enhancement of the capacity of a record to provide infor-mation on the characteristics of a single item, by means of a contextual presen-

102


25 OCLC, Perceptions of libraries and information resources. A report to the OCLC Membership,Dublin (OH), OCLC, 2005, <http://www.oclc.org/reports/2005perceptions.htm> (date ofaccess: 12 Dec. 2008).

26 S. VAIDHYANATHAN, A Risky Gamble With Google, «The Chronicle Review», 52/15, Dec.2, 2005, p. B7 <http://chronicle.com/weekly/v52/i15/15b00701.htm>.


tation of elements such as the table of contents, a first chapter, book jacket illus-trations, images, reviews – that is, information that readers or prospective usersof a book could find of interest. In order to provide this service a cataloguedraws directly on information generated by the book publishing industry, whichin relation to the so-called enhanced OPAC functions as a promotional artefact,advertising the existence of the item and the services the library can provide,and at the same time making a statement of authority about the communicativelinks that are supported and facilitated. The question of authority, it should besaid, represents one of the key elements in the role played by the library duringthe process of information creation and transmission.27

The OPAC potentially has a fourth function as the management of full-textdata and bibliographic data converge, and the bibliographic function of theOPAC itself becomes enabled for full-text searching rather than remaining pri-marily an index. This technology is not widely implemented in current OPACinstallations, but it is altogether likely to develop, and will therefore add a fur-ther layer of complexity to its architecture.

As we all know, the flow of information, from source, through encodingtransmitter, signal and decoding receiver to destination, all linked by a sharedcode, which is a shared system of values or meanings, is by no means straightfor-ward and is, in fact, impeded by indeterminacies at each step. Without this code,no communication would be possible, since it would not be possible to passmeaning from one step to the next. The efficiency of an OPAC search, which isin reality a double act of communication – a question followed by an answer –depends to a very great extent on the degree to which the code is consistentlyunderstood and applied by the agents adopting the role of source and destina-tion. On the one hand, the code comprises the body of cataloguing rules andstandards according to which the database has been constructed (AACR,MARC, LCSH, DDC in the English-speaking countries; RICA, UNIMARC,Florence National Library Subject Headings, and again DDC in Italy), togetherwith a further set of rules which govern the design of the OPAC software itselfand the ways in which it presents data back to the user. The above-mentionedbibliographic control standards do not constitute an integrated set of instruc-tions. There is, for example, no explicit mapping of either AACR or RICA toMARC, and there is no one-to-one correspondence between the subject head-ings and the classification code. Neither are any of these standards altogether

103


27 P.G. WESTON, La gestione elettronica delle biblioteche, in Biblioteconomia: princìpi e questio-ni, a cura di G. Solimine e P.G. Weston, Roma, Carocci, 2007, pp. 221-256.


internally consistent, none of them has remained constant over time, and veryfew (and even then, probably very small) libraries have attempted to keep theircatalogue records retrospectively up-to-date with changes to all of them. On theother hand, the code consists of the user’s understanding of these rules: the lev-el of his or her information literacy in respect to the particular OPAC he or sheis using. Now, if an OPAC search for a known bibliographic entity, using index-es based on the traditional tracings provided by catalogue records (author, title,ISBN) will produce a relatively unambiguous result, with a keyword search, andeven more so with a full-text search, a first area of indeterminacy arises from thefact that any question on the part of the user must be mediated by language inorder for it to be posed. A second area of indeterminacy is the understanding ofthe context in which the search string was retrieved. It is within this area thatnavigational tools such as topic maps could play an important role.

Topic maps are defined as «an ISO standard for the representation and inter-change of knowledge, with an emphasis on the findability of information».28

Additionally, this same specification defines «an XML-based interchange syntaxfor topic maps»29 known as XTM, which dates from 2001. Even when it firstemerged, the possibility of leveraging existing bibliographic information so as tocreate a topic map of library holdings seemed promising, since MARC data arealready indexed through a very sophisticated and standardized system that relieson topic analogues (subject, title, author and keyword indexes), as well as a veryextensive subject heading hierarchy provided by controlled vocabularies and theDewey system. Why then would it not be possible to render the already seman-tically tagged data as a topic map? At first glance, the use of topic maps in a cat-alogue, to the extent that it is theoretically possible, turns out to be problemati-cal for more reasons than one.30 Above all, the subject headings do not necessar-

104


28 ISO/IEC 13250-2:2006: Topic Maps — Data Model, Geneva (Switzerland), Inter -national Organization for Standardization, 2006 (minor changes in 2008),<http://www.isotopicmaps.org/sam/sam-model/>. As an easy introduction to the subject,see S. PEPPER, The TAO of topic maps: finding the way in the age of infoglut, 2001,<http://www.ontopia.net/topicmaps/materials/tao.html> (date of access: 12 Dec. 2008);M. BIEZUNSKY, Introduction to the topic maps paradigm, in XML Topic Maps: creating and usingtopic maps for the web, ed. by J. Park and S. Hunting, Boston, Addison-Wesley, 2003, pp. 17-30; F. MESCHINI, Le mappe topiche: come imparai a non preoccuparmi e ad amare i metadati,«Bollettino AIB», XLV/1, 2005, pp. 59-72.

29 ISO/IEC FDIS 13250-3:2007: Topic Maps — XML Syntax, Geneva (Switzerland),International Organization for Standardization, 2007, <http://www.isotopicmaps.org/sam/samxtm/> (date of access: 12 Dec. 2008).

30 E. IGLESIAS – S.S. HYE, Topic maps and the ILS: an undelivered promise, «Library Hi


ily convey relationship information. Moreover, indexes are lists of occurrencesand lack related data, which would be very useful to have at hand. For this rea-son producers of new generation catalogues have developed technologies such asfaceted searching, which provide users with the possibility of associating namesand concepts. The problem, in this case, is that checks on the uniformity of theterms that appear in these lists are entrusted entirely to the underlying manage-ment application, which is not always able to anticipate the criteria that will beapplied by the higher platform to create the facets. A second and more challeng-ing problem consists in the difficulty of preserving for each term present in afacet the historical and geographical context that gives meaning to the term. Inthis way a historical period can only have a meaning when associated with a par-ticular country. Once the period has become a term itself on a list it ceases tohave much meaning, and can even be misleading if used in combination withother terms that are not among those with which it was originally associated.The same thing can happen when one is dealing with an event related to thebiography of a specific person.

Most problematic of all, internal inconsistencies in cataloguing as well as theMARC standard have made the problem of exporting this metadata into anykind of topic map barely feasible thus far. The benefits of using topic mappingtechnology in the increasingly complex library systems environment may out-weigh the obstacles, but the library community will certainly have to effectsome changes. Other standards, while less rich than MARC, may become moreuseful as the network of resources expands beyond the library communitywhere MARC is not the preferred metadata standard.31 This is rather unfortu-

105


Tech», 26/1, 2008, pp. 12-18.31 In the sphere of publishing and bookselling Editeur, the International Association that

brings together the most important players in the sector has actively promoted ONIX, a meta-data schema chiefly aimed at the representation of information useful for the management ofcommercial activity, which employs the Dublin Core metadata schema for its bibliographicdescriptive component. Dublin Core – originally promoted by the OCLC network – can beused in a range of applications for the creation of digital libraries and, because of the commondesire for a simple and generic structure, also frequently serves as an interoperability tool forthe aggregation of data generated in various cultural settings, and is therefore endowed withdifferent syntactic and semantic structures. An evolution of MARC towards the use of XMLand an extension of its use in contexts other than strictly catalographical ones is exemplified byMODS. Finally, mention should be made of the RDA (Resource Discovery and Access), a newstandard for resource description and access designed for the digital world. RDA will providea flexible framework for describing all resources (analogue and digital), data that is readilyadaptable to new and emerging database structures, as well as data that is compatible withexisting records in online library catalogues, although its creators have gradually backed away


nate, since much of the energy expended on cataloguing efforts goes into assign-ing subject headings to intellectual objects and then creating authority controlson these assignments, so that the user knows to some extent that when search-ing for Chaucer this refers to ‘Chaucer, Geoffrey, d. 1400’. Various attemptshave been made and continue to be made to ensure that the rich associative def-initions contained in the subject headings can be revealed to the user.Technologies such as the faceted searching strategy mentioned earlier help thesearcher to narrow or broaden his/her search by setting limiters based on cate-gories such as format, date or topic. Information already contained in theMARC record is used to expose identical items or trim away those which are lessrelevant to the searcher’s interests.

The real strength of topic maps is that they rely on specific associationsapplied to topics. Topics can consist in practically anything, hence if you havethe topic ‘Chaucer’ and a topic described as ‘the Wife of Bath‘ you could form anassociation. However, you would first require a topic described along the lines of‘works by’. This would result in associations such as:

(1) Chaucer is an author.(2) Chaucer wrote The Canterbury Tales.(3) ‘The Wife of Bath‘ is in The Canterbury Tales.32

106


from their original stance that RDA could be expressed easily in MARC. The body responsi-ble for its development is the Joint Steering Committee for the Revision of AACR (JSC),which consists of representatives from six major Anglo-American cataloguing communities,including the American Library Association (ALA), the Australian Committee on Cataloguing(ACOC), the British Library (BL), the Canadian Committee on Cataloguing (CCC), theChartered Institute of Library and Information Professionals (CILIP), and the Library ofCongress (LC). Built on the foundations established by the Anglo-American CataloguingRules (AACR), RDA will provide a comprehensive set of guidelines and instructions onresource description and access, covering all kinds of content and media. The new standard isbeing developed primarily for use in libraries, but consultations have been held with othercommunities (archives, museums, publishers, educators, book dealers, library systems vendors,etc.) in an effort to reach effective levels of alignment between RDA and the metadata stan-dards used in those communities. Through the adoption of RDF (Resource DescriptionFramework) logic, RDA has been developed explicitly to take advantage of the Semantic Web.

32 E. IGLESIAS – S.S. HYE, Topic maps and the ILS…, p. 14. The following sequence repre-sents the code that expresses one of the associations indicated:

<association><instanceOf><topicRef xlink:href = “#written-by/></instanceOf><member>

<roleSpec><topicRef xlink:href= “#author/></roleSpec><topicRef xlink:href = “#Chaucer” >


There are various advantages to structuring data in this manner. Each item ortopic carries with it information about its context. We know from the abovesnippet of code that the ‘Wife of Bath‘ has this specified role ‘roleSpec‘ as a char-acter in a work written by Chaucer. Thus, when a user clicks on a link to the‘Wife of Bath’ all of this contextual information is apparent. Some of this infor-mation could be gleaned from subject headings, but not all of it. Faceted search-ing might tell you that the ‘Wife of Bath’ is associated with Chaucer if the sub-ject heading ‘Wife Of Bath Chaucer Geoffrey D 1400 Characters‘ is defined.Beyond this point association building becomes more difficult. A keywordsearch for ‘Wife of Bath’ might turn up a book of Chaucerian criticism. A userwith an assignment to write a paper on the Wife of Bath might have no indica-tion as to why this item was relevant. Only with prior knowledge of the associa-tion, or careful scrutiny of the record, would the connection become apparent.A topic map would immediately reveal the association. Once one understandsthe potential of topic maps, it becomes easier to see how encoding data so that itincludes associations can transform it into valuable asset.

This leads us to propose the enhancement of the traditional OPAC by meansof text-driven navigational tools: that is by using TEI33 in conjunction with top-ic maps in order to produce a large website which can be navigated easily inmany directions. As shown by the examples of the New Zealand Electronic TextCentre (NZETC)34 and the Diary of Samuel Pepys,35 there are several good rea-sons for doing so.36 On the one hand, topic maps can be used as a tool for pre-senting TEI-encoded texts in HTML form. Many electronic text archives trans-form their TEI texts into HTML prior to their publication on the Web.Typically, each chapter or page is transformed from TEI into a separate webpage. Such a method produces websites that have the same structure as a physi-cal book. However, TEI is far more expressive than HTML and can encodemany other features of interest beyond simply chapters, pages, and paragraphs.

107


</member><member>

<roleSpec> <topicRef xlink:href = “#character”/> </roleSpec><topicRef xlink:href = “#Wife of Bath”/>

</member></association>33 Text Encoded Initiative, <http://www.tei-c.org/> (date of access: 12 Dec. 2008).34 <http://www.nzetc.org> (date of access: 12 Dec. 2008).35 <http://www.pepysdiary.com/> (date of access: 12 Dec. 2008).36 C. TUOHY, Topic Maps for Cultural Heritage Collections, Topic Maps 2008,

<http://www.topicmaps.com/tm2008/tuohy.pdf> (date of access: 10 Jan. 2009).


For example, TEI is also used to encode information about people and placesand events, as well as literary criticism and linguistic analysis. In other words,the code can be extended to suit all kinds of scholarly needs. These more com-plex aspects of text encoding can only be converted into HTML with difficulty.Because it is designed to be convenient for scholars to encode complex informa-tion, rather than for ordinary readers to understand it, it is necessary to converta TEI mark-up into another form suitable for display. For instance, where a TEIcorpus includes references to people, these references might be collated so as togenerate an index. For practical purposes, it is often necessary to extract infor-mation from TEI and place it in a database, so that it can be queried convenient-ly and transformed into a website. The topic map standard is identified as a suit-able means for solving this problem. A topic map can be seen as a kind of webdatabase with a highly flexible structure.

On the other hand, a topic maps-based website has a more subject-centricarchitecture, as opposed to the resource-centric organisation used by the major-ity of digital library websites, in which individual documents are the primary orexclusive objects of interest. A topic map structure will accommodate not onlydigitised texts and images, but also their subjects and themes and their authorsand publishers, as well as the people and places mentioned or depicted in thosetexts and images. Because of the generality of the topic maps paradigm, the con-ceptual structure can be extended as required, e.g. to include extra classificationschemes, such as Linnaean classification for biological texts, or a genealogicaltree, to provide more specific types of relationships between texts, such as two ormore variant forms of the same term, so as to establish links to bibliographiccitations or to external resources of any kind.

Texts have to be transcribed into XML by encoding their logical structure,such as their division into paragraphs, sections, and chapters, in addition to bib-liographic metadata such as the author, subject classification, and publicationinformation, as well as any identifiable subject and relationship implicit in thetexts. All this information is then used to generate a navigational frameworkcapable of delivering enhanced resource discovery and navigation both withinand between collections. A key component of the work is authority control:namely, unambiguously identifying people, places, and other entities mentionedin the digitised texts.

It should be mentioned that the NZETC’s topic map is constructed automat-ically by transforming each of the TEI documents, the authority database, andother sources, into individual XML Topic Map documents, and merging thesetopic maps using open source software (TM4J Topic Map engine). The NZETCsystem is based on international standards for the representation and inter-

108



change of knowledge. Digitised texts are encoded using Text Encoding forInterchange (TEI). Authority lists are maintained in a purpose-built databaseand exchanged using Metadata Authority Description Schema (MADS). TEIand MADS documents are transformed by using Extensible Stylesheets (XSLT)into XML Topic Maps (XTM). The topic maps themselves use the CIDOCConceptual Reference Model CRM.37 The collection includes more than 2,000texts and tens of thousands of records of people and places, totalling about110,000 topics.

The final issue to be addressed in this paper is long-term preservation, whichagain has to do with metadata and the adoption of standardized procedures andstrategies. If one disregards the technology preservation option, there are cur-rently two main proposed strategies for long-term digital preservation: the firstis the emulation of original hardware, operating systems and software; the sec-ond consists in the periodic migration of digital information from one genera-tion of computer technology to its successor.

Emulation strategies are based on the premise that the best way to preservethe functionality and the ‘look-and-feel’ of digital information objects is to pre-serve them together with their original software so that they can be run on emu-lators that can mimic the behaviour of obsolete hardware and operating systems.Emulation strategies would involve encapsulating a data object together withthe application software used to create or interpret it, in addition to a descriptionof the required hardware environment - i.e., a specification for an emulator. Itseems that these emulator specification formalisms will require human readableannotations and explanations (metadata). Jeff Rothenberg states, for example,that the emulation approach requires «the development of an annotation schemethat can save […] explanations [of how to open an encapsulation] in a form thatwill remain human-readable, along with metadata which provide the historical,evidential and administrative context for preserving digital documents».38

The periodic migration of digital information from one generation of com-puter technology to the next is currently the most tried-and-tested preservationstrategy. However, as Seamus Ross points out, «data migration inevitably leads

109


37 The CIDOC Conceptual Reference Model (CRM) provides definitions and a formalstructure for describing the implicit and explicit concepts and relationships used in culturalheritage documentation, <http://cidoc.ics.forth.gr/> (date of access: 10 Jan. 2009).

38 J. ROTHENBERG, Avoiding technological quicksand: finding a viable technical foundation fordigital preservation. Washington (D.C.), Council on Library and Information Resources, 1999,p. 27, <http://www.clir.org/pubs/reports/rothenberg/contents.html> (date of access: 10 Jan.2009). See also J. ROTHENBERG, Authenticity in a Digital Environment, Washington (D.C.),Council on Library and Information Resources, 2000.


to some losses in functionality, accuracy, integrity and usability»39. In someinstances, this is likely to be rather significant. David Bearman, for example, hasremarked that if electronic records are migrated to new software environments,«content, structure and context information must be linked to software func-tionality that preserves their executable connections».40 However, if this cannotbe done, he suggests that «representations of their relations must enable humansto reconstruct the relations that pertained in the original software environ-ment».41 Successful migration strategies will, therefore, depend upon metadatabeing created to record the migration history of a digital object and to recordcontextual information, so that future users can either reconstruct or - at thevery least - begin to understand the technological environment in which a par-ticular digital object was originally created.

There is currently a certain amount of speculation about the relative meritsof the two strategies. Rothenberg, for example, claims that migration has little torecommend it and calls it «an approach based on wishful thinking». He criticis-es the approach because he feels that it is impossible to predict exactly what willhappen in the future and because the approach is costly and labour-intensive. Inthe absence of any alternative, a migration strategy may be better than no strat-egy at all; however, to the extent that it provides only the illusion of a solution,this approach may be worse than doing nothing at all in some cases. In the longrun, migration promises to be expensive, error-prone, at most only partially suc-cessful, and ultimately unfeasible. Those who are in favour of emulation arguethat what needs to be preserved is the information system itself, rather than thatwhich the system produces. This approach implies capturing «all transactionsentering and leaving the system when they are created, ensuring that the origi-nal context of their creation and content is documented, and that the require-ments of evidence are preserved over time».43

110


39 S. ROSS, Consensus, communication and collaboration: fostering multidisciplinary co-operationin electronic records, in Proceedings of the DLM-Forum on Electronic Records, Brussels, 18-20December 1996, INSAR: European Archives News, Supplement II, Luxembourg, Office forOfficial Publications of the European Communities, 1997, p. 331.

40 D. BEARMAN, Electronic evidence: strategies for managing records in contemporary organiza-tions, Pittsburgh (Pa.), Archives and Museum Informatics, 1994, p. 302.

41 Ibidem.42 D. BEARMAN, Reality and Chimeras in the Preservation of Electronic Records, «D-Lib

Magazine», 5 (4), April 1999, <http://www.dlib.org/dlib/april99/bearman/04bearman.html>(date of access: 8 July 2009).

43 C. LYNCH, Canonicalization: a fundamental tool to facilitate preservation and management ofdigital information, «D-Lib Magazine», 5 (9), September 1999, <http://www.dlib.org/dlib/


Regardless of whether emulation-based or migration-based preservationstrategies are adopted – and it is likely that both will play some role – the long-term preservation of digital information involves the creation and maintenanceof metadata:

Within an archive, metadata accompanies and makes reference to each digital objectand provides associated descriptive, structural, administrative, rights management,and other kinds of information. This metadata will also be maintained and will bemigrated from format to format and standard to standard, independently of the baseobject it describes.43

As a result, preservation metadata has become a popular area for research anddevelopment in the archive and library communities. Archivists and recordsmanagers have concentrated on the development of record keeping metadata,while other groups have dealt with defining metadata specifications for particu-lar needs. An achievement of great importance has been the development of amodel known as the Reference Model for an Open Archival Information System(OAIS). The latter is a high-level reference model, which is defined as an organ-isation of people and systems that have «accepted the responsibility to preserveinformation and make it available for a designated community».44 Needless tosay, the OAIS model does not just deal with metadata; it also defines and pro-vides a framework for a range of functions that are applicable to any archive –both digital and otherwise. These functions include those described within theOAIS documentation as ingest, archival storage, data management, administra-tion and access. Amongst other things, the OAIS model aims to provide a com-mon framework that can be used to help understand archival challenges andespecially those relating to digital information.

As part of this framework, the model identifies and distinguishes betweenthe different types of information (or metadata) that will have to be exchangedand managed within an OAIS repository. In order to develop a standard set ofmetadata for supporting digital preservation activities, in June 2003, OCLCand RLG jointly sponsored the formation of the PREMIS (PreservationMetadata: Implementation Strategies) working group,45 a team of international

111


september99/09lynch.html> (date of access: 10 Jan. 2009).44 Consultative Committee for Space Data Systems, Reference model for an Open Archival

Information System (OAIS), Red Book, Issue 1, CCSDS 650.0-R-1, Washington (D.C.),National Aeronautics and Space Administration, pp. 1-11, <http://ssdoo.gsfc.nasa.gov/nost/isoas/ref_model.html> (date of access: 10 Jan. 2009).

45 <http://www.oclc.org/research/projects/pmwg/> (date of access: 10 Jan. 2009).


experts in the use of metadata, representing five different countries and a vari-ety of domains, including libraries, museums, archives, government agencies,and the private sector.

Part of the working group’s charge was to develop a core set of imple-mentable preservation metadata, broadly applicable across a wide range of digi-tal preservation contexts and supported by guidelines and recommendations forcreation, management, and use. This portion of the working group’s charge wasfulfilled in May 2005 with the release of Data Dictionary for PreservationMetadata: Final Report of the PREMIS Working Group. Version 2.0 was publishedin March 2008.46 The Report provides a store of resources on preservation meta-data. First and foremost is the Data Dictionary itself, a comprehensive, practicalresource for implementing preservation metadata in digital archiving systems.The Data Dictionary defines preservation metadata that:

– support the viability, renderability, comprehensibility, authenticity, and identi-ty of digital objects in a preservation context;

– represent the information most preservation repositories need to know in orderto preserve digital materials over the long-term;

– emphasize ‘implementable metadata‘: rigorously defined, supported by guide-lines for creation, management, and use, and oriented toward automatedworkflows; and

– embody technical neutrality: no assumptions are made about preservationtechnologies, strategies, metadata storage and management, etc.47

In addition to the Data Dictionary, the working group also published a set ofXML schema to support implementation of the Data Dictionary in digitalarchiving systems. Following the release of the Data Dictionary in 2005, thePREMIS working group retired and the PREMIS Maintenance Activity, spon-sored by the Library of Congress, was initiated to maintain the Data Dictionaryand coordinate other work to advance understanding of preservation metadataand related topics.

Needless to say, any international project in the field of digital text archivesshould prioritise long-term preservation strategies.

112


46 PREMIS Editorial Committee, PREMIS Data Dictionary for Preservation Metadata, ver-sion 2.0, March 2008, <http://www.loc.gov/standards/premis/v2/premis-2-0.pdf> (date ofaccess: 6 Feb. 2009).

47 Ivi, p. 8.


MARIA GRAZIA SAIBENEUniversità di Pavia

THE WANDERER. TEXT, INTRATEXT, INTERTEXT:EDITING OLD ENGLISH ELEGIES

Abstract. A critical edition (Text) of The Wanderer must be based on a close scruti-ny of the manuscript text (Intratext) which will clarify and evaluate the manuscriptevidence and all the aspects which together characterize the work in question. TheWanderer, like the other Old English elegies, is made up of a variety of ‘movements’which differ from each other thematically and stylistically, influenced as they are bythe oral-formulaic tradition. As against a printed edition which offers us a static text,an electronic edition is capable of presenting this text in a variety of forms (in theoriginal, in transcription, in an interpretative edition, and in the ways it has been pre-sented in earlier editions). In short, an electronic edition can render immediatelyaccessible in the hypertext all the evidence which will enable the reader to graspintertextual connections (Intertext) and to allow him to arrive at an informed andpersonal interpretation.

1. Text

THE WANDERER

1 «Oft him anhaga are gebideđ,metudes miltse, þeah þe he modceariggeond lagulade longe sceoldehreran mid hondum hrimcealde sæ,

5 wadan wræclastas: wyrd biđ ful aræd.»1

Traditional modern editions of Anglo-Saxon poetic texts present poemsaccording to criteria which aim to critically reconstruct the text. These editions

113

1 A.L. KLINCK, The Old English Elegies: A Critical Edition and Genre Study, Montreal &Kingston-London-Ithaca, McGill – Queen’s Univ. Press, 1992, p. 75.


take into account the author, the style and the genre while also seeking out theorigin and the date of the composition. These aspects and criteria condition edi-torial choices, above all for texts from distant cultures and which were poemsstill linked to orality or destined for oral delivery.

I will go on to take a look at the main characteristics of the Old English ele-gies, but I begin by stating that ‘modern’ editions provide a distorted idea of themanuscript texts, even if they are highly readable and understandable for themodern reader.2 In the Introduction to New Approaches to Editing Old English Verse,O’Brien O’Keeffe has tackled these and other issues:

As Doane argues, editors of Old English verse follow an approach that privilegesthe aesthetic and the authoritative in order to produce a clear reading text. […]Such concerns, however, occlude numerous features of speech that Doane arguesare encoded in the manuscripts by such ‘erratic’ features as variable spacing, freemorphemic word division, sporadic punctuation, capitalization, and so forth.Such ‘speaker-based textuality’, […] markedly different from the textuality ofLatin manuscripts, was a product of a traditional understanding of the past con-ditioned by present circumstances.3

Taking into account what Doane argues,4 O’Brien O’Keeffe states that manyclues present in the manuscript traditions are not fully appreciated and con-veyed in modern editions. I took precisely this as my starting point and Iattempted, in my research, to examine what the best form for the edition of theOld English elegies may be. In fact, I intend to publish an electronic edition ofThe Wanderer and The Seafarer that will be a critical and interpretative editionwhich makes full use of hypertext technology.

Over time we have witnessed a shift from the Lachmannian position, aimedat the reconstruction of a text that is as close as possible to the original, to agreater consideration of the witnesses of the manuscript tradition, the variantsand the work of the copyists. The mouvance of medieval texts has been rightly

114

MARIA GRAZIA SAIBENE

2 K. O’BRIEN O’ KEEFFE, Introduction, in S. L. Keefer – K. O’Brien O’ Keeffe (eds.), NewApproaches to Editing Old English Verse, Cambridge, D.S. Brewer 1998, pp. 1-9. «[…] we are allaware that Old English poetry was never set out by lines, no less spaced at the caesura, not reg-ularly capitalized. […] and that our notions of authorship are inevitably modern.» (Ivi, p. 2).

3 S.L. KEEFER – K. O’BRIEN O’KEEFFE (eds.), New Approaches…, p. 6. 4 A.N. DOANE, Oral Texts, Intertexts, and Intratexts: Editing Old English, in J. Clayton –

E. Rothstein (eds.), Influence and Intertextuality in Literary History, Madison, WI, Univ. ofWisconsin Press, 1991, pp. 75-113.

A.N. DOANE – C.B. PASTERNACK (eds.), Vox intexta. Orality and Textuality in the MiddleAges, Madison, WI, Univ. of Wisconsin Press, 1991.


highlighted, as has the fact that they were often composed as pieces to be per-formed orally in front of a specific audience. More attention has been paid to thevariants but also as to how the poem has been transcribed in the manuscript asregards the material aspects. Adhering to this new line of philological research, Ibelieve that one of the aims in the edition of medieval texts, in part related toorality, is that of understanding and, as far as possible, making clear in the mod-ern edition the techniques used in composing the text.

For editions of medieval poems it is important to consider the copy or thecopies of the manuscript text in their context, in relation to other texts, andexamine how the manuscript has been composed. I quote Irving Jr. who statesthat: «Recent editors pay far more attention to the information furnished by themanuscript itself […] What was the manuscript for? What kind of people put ittogether for what kind of people to read or listen to?».5

In my opinion, the most suitable approach in studying these poetic texts isone that takes into account the structure and the characteristics of poetic com-position as well as its style and vocabulary. In fact, other approaches that consid-er primarily content in arriving at an interpretation should not be the sole guideof an analysis of the text in creating the edition as this may, at times, lead to over-interpretations.

For medieval poetic texts that still have links to orality, it might be worth-while formulating new editions that serve as a guide to the reader. Such a layoutcan be seen in the proposal of edition of the following lines of Beowulf:

Such a format might even suggest, faintly, the ‘oral’ or performative nature ofthe verse […]. Thus, a well-known passage from Beowulf might look like this:

«Swa biđ geomorlic gomelum ceorleto gebidanne

þæt his byre ridegiong on galgan

þonne he gyd wrece sarigne sangþonne his sunu hangađ hrefne to hrođre

ond he him helpe ne mægeald ond infrod

ænige gefremman.» (2444-9)6

115

The Wanderer. Text, Intratext, Intertext

5 E.B. IRVING JR., Editing Old English Verse: The Ideal, in S.L. Keefer – K. O’BrienO’Keeffe, New Approaches…., pp. 11-20 (p. 11).

6 Ivi, p. 14.


The half-line division has been maintained in this new editorial format butthe white spaces and a modified metrics layout has been used to reproduce theintonation and to highlight certain elements. Lines 2446-7 display a more cus-tomary layout that emphasizes the parallel structure of the two lines.

Even the punctuation in the manuscript, usually dots, ought to be taken intoconsideration when deciding where editions should divide into parts or sectionsmedieval poetic texts, as Keefer states: «Clues remain in the manuscript toanother configuration that has been obscured by editorial decision, and withthose clues, another interpretation, or, at the very least another layer has beenoverlooked».7

Consequently, my option is for an electronic edition of the two Anglo-Saxonelegies that makes it possible to compare between multiple forms and stages ofreception. Electronic editions also take into account, at the intertextual level,repeated formulae and expressions.

Only hypertext links between various texts and between various parts allowthe text to be exploited in an open, multimedia form for today’s readers andscholars too. This type of edition allows readers the opportunity to assess thework of the critic who created the particular edition but also to participate per-sonally in its interpretation in this latest, novel stage in the text’s reception. Onthe advantages of a hypertextual representation, Keefer declares:

[…] the progress achieved in the development of computer hypertext editing hasmade the presentation of facsimile, diplomatic, semidiplomatic, and critical ver-sions of Old English verse increasingly accessible to those with equipment andexpertise available to them. Yet neither of these media is likely to replace theprinted book, or supersede the hard-copy edition. The ongoing theories of edit-ing Old English verse must enter a new balance between responsible presenta-tion of texts that may require emendation or tidying, and responsible preserva-tion of physical information that may ultimately affect the received scholarshipon the text itself.8

This approach is a well balanced compromise of different concerns: on theone hand there is the need to emend the text in the light of the manuscript tra-dition and, on the other, the necessity to consider as useful each clue present inthe manuscript to create an edition that conserves the spirit with which the textwas written and also the composition’s structure.

116


7 S.L. KEEFER, Respect for the Book: A Reconsideration of ‘Form’, ‘Content and ‘Context’ in TwoVernacular Poems, in S.L. Keefer – K. O’Brien O’Keeffe (eds.), New Approaches…, pp. 21-44(p. 41).

8 Ivi, p. 43.


2. Small capitals

Of the more interesting codicological aspects in reconstructing the composi-tion and eventual sections in which The Wanderer is structured, we can look at theuse of capital letters and small capitals. Here are the lines that display these clues:

l. 1 OFT him anhaga are gebideđl. 6 Swa cwæđ eardstapa earfeþa gemyndigl. 8 Oft ic sceolde ana uhtna gehwylcel. 15 Ne mæg werig mod wyrde wiđstondanl. 73 Ongietan sceal gleaw hæle hu gæstlic biđl. 88 Se þonne þisne wealsteal wise geþohtel. 97 Stondeđ nu on laste leofre duguþel. 111 Swa cwæđ snottor on mode gesæt him sundor æt rune

The scribe has capitalized the first word, ‘OFT’, and ornamented the ‘O’making it two lines high. This marks the beginning of the elegy and also theintroductory section, which occupies the first five lines. At l. 6 we find the Swa-clause, repeated at l. 111, «Swa cwæđ eardstapa»; «Swa cwæđ snottor on mode».Through the use of this formula and the small capitals, the intention is clearly todistinguish the sections and introduce the two characters, the wanderer and thewise man. These also represent two different perspectives from which the condi-tion of man on earth can be explored. A certain symmetry can also be noted inthe collocation of this word as the first Swa is found after the opening five linesand, after another appearance at l. 111, there are four other lines until the end.

So, just as the opening lines introduce the subject and a possible salvation byGod’s grace, followed in a later speech by the wanderer when he advises valiantmen to keep their suffering to themselves, in the final section we see a corre-spondent of the two themes but in reverse. At first there is the invitation not tomanifest one’s sorrow too hurriedly and then comes the conclusion that Godabove is the only source of consolation and a fortress for us. This framed struc-ture, or ‘envelope pattern’, characterizes, as will become clear, even specific sec-tions of the poem.

At l. 15 Ne introduces a gnomic conclusion which states that a troubled soulcannot fight its destiny and the subsequent ne introduces a parallelism and avariation: l. 16: «ne se hreo hyge helpe gefremman». This is a conclusion arrivedat after the wanderer’s reflections while its position at ll. 15-16 is also relevant aswe will see that the poem is structured in a series of verses that correspond tosections of 5 or multiples of 5, as in this case. Also, half-line 5b ends the intro-duction with the sentence: «wyrd biđ ful aræd» «destiny is inescapable».

117



From the next part up to l. 73, there are no other small capitals. I believe thatthe absence of these clues, which at the start and at the end introduce sectionsthat have to stand out, visually too, is because the central part has characteristicsthat are more in tune with the oral performance and so there are dots wheremajor pauses should be taken. Other clues are also employed, for example theadverb forþon at the start of the sentence, as well as a variety of formulaic and sty-listic uses.

Ongietan, at l. 73, is a reference to the wise man who has to comprehend howeverything will go to rack and ruin. This is the main theme of the elegy The Ruinand is one that has enjoyed a long tradition in Anglo-Saxon literature, as well asin medieval Latin. This type of usage is more linked to a written form ratherthan that present in the first half of The Wanderer that appears closer to oral tra-dition. At l. 88, introduced by Se þonne, there is a variation of the precedingtheme as the wise man reflects on devastation, slaughter and life on earth while,at l. 91, the Ubi sunt theme is introduced with a formula. Here too we find par-allel and traditional structures, introduced by Hwær and then by Eala, in sectionsthat are homiletic and evoke more a written tradition.

The last small capital at l. 97 (Stondeþ) introduces, as it does in The Ruin, themotif of the wall: l. 98 «weal wundrum heah / wyrmlicum fah». The wall risesamong the ruins, is amazingly high and ‘decorated with snakes’. At a symboliclevel, it may recall Christ, the ‘corner stone’.9 The significance of the verb is aclear counter to the ruins that are then described and that characterize the worldand life on earth, so that the poet concludes by referring to salvation owing tothe grace of God. At the end of the poem we find the same verb, stondeþ, whichreaffirms that, for man, God is an unmoveable fortress: l. 115 «þær us eal seofæstnung stondeþ».

In conclusion, I think that, as well as the initial capital letter, the otherforms that have small capitals introduce sections that need to be highlighted,above all to guide the reading. The absence of these clues in the middle sectionof the elegy would seem to suggest that it is a transcription of a text that haslinks to the oral tradition. This can be gauged from the stylistic and formulaicuses employed; traits that suggest that the poem was designed for an oral per-formance.

118


9 M.G. CAMMAROTA, The Ruin. Proposta di rilettura, «Linguistica e Filologia», 4, 1997, pp.25-48.


3. Dots in the manuscript

The following table displays where the dots appear, with reference to the lineor the half-line and with the word that follows also stated:

l. 5 . Swa; l. 7 . Oft; l. 14 . Ne; l. 16 . forđon; l. 18 . swa; l. 19 . oft ; l. . 24. sohte; l. 33 . gemon; l. 36 . forđon; l. 38 . đonne; l. 40 . þinceđ; l. 44 .đonne; l. 48 . þonne; l. 57 . forþon; l. 63 . forþon; l. 65 . ne; l. 66 . ne;l. 67 . ne; l. 68a . ne; l. 68b . ne; l. 69 . beorn; l. 72 . Ongietan; l. 73 . þonne;l. 74 . swa; l. 81 . in; l. 87 . Se; l. 91 . hwær; l. 92a . hwær; l. 92a . hwær;l. 92b . hwær; l. 93a . hwær; l. 93b . eala; l. 94a . eala; l. 94b . eala; l. 96 .Stondeþ; l. 98 . eorlas; l. 105 . eall; l. 107 . her; l. 108a . her; l. 108b . her;l. 109a . her; l. 110 . Swa

We can begin by noting that before each small capital there is a dot. Further,these dots generally signify the end of a part and so are followed by adverbs orconjunctions (above all time conjunctions) or even relative pronouns. The fre-quent recurrence of þonne and forþon only confirms that the composition of thesetexts was fragmented and had a mosaic-like structure and also that these connec-tors signal the passage from one section to another. These sections present dif-ferent themes and even have different styles. In three cases, a verb that express-es ‘think, remember, understand’ follows the dot. These are expressions thatrelate to the wanderer who recalls his dead companions and dreams of beingable to embrace his lord once again; while it is the wise man thinking andreflecting in the second half of the composition. The wanderer was part of aband of armed men and the two nouns in the above list have strict ties to thewarriors referred to. The Stondeþ form is a reference to the wall that rises fromthe ruins and has a capital letter as it introduces the motif that gains significanceonce again towards the end of the elegy.

The only problem in terms of the interpretation springs from the dot foundat l. 81 before in forđwege «on the pathway to the other side» (passing away).Here, the dot does not seem to signify a pause or break as the indirect object islinked to and taken by the verb ferede «brought»; thus this may have been anoversight or been added later and not due to the copyist’s intervention.However, if we consider the position of the dot, it may be that it highlights theindirect object which introduces and summarizes all the single elements listedin the parallel structure introduced by sume…sume and which refer to the end ofmankind.

119



To conclude this analysis and in order to reconstruct and interpret the struc-ture and sections of the poem, it may be worth noting the frequency with whichterms appear after the dots:

ne = 6; hwær = 5; swa = 4; forđon = 4; đonne = 4; her = 4; eala = 3;oft = 2; sohte = 1; gemon = 1; þinceđ = 1; beorn = 1; Ongietan = 1;in = 1; Se = 1; stondeþ = 1; eorlas = 1; eall = 1.

Time connectors are found more often, as are elements that are repeated inparallel structures, above all in the second part of the poem. Terms that appearonly once are verbs, nouns and adjectives that signal certain elements or themesto be given prominence (on the one hand men and warriors; on the other God,the immovable ‘fortress’).

4. Compounds

In the transcription of compounds, it is apparent that the copyist writes thetwo parts of the compound separately. Compared to modern editions where thetwo parts are combined, this different practice could, no doubt, be significant fordiction and alliteration. In fact, there may be links between forms that are partof the same alliterative sequence which are obscured in modern editions. Thisposes the problem of how best to render compound elements. Another concernis whether it is right to signal instances where the second element of a com-pound is involved in an alliterative sequence (which in the manuscript appearsseparately).

Another interesting aspect for the oral delivery of the poem concerns thetranscription of weaker elements that, in the manuscript, are generally connect-ed to the terms that follow them: þeahþe, nemæg, neto, nenæfre are just a few exam-ples. The most frequently found case is the proclitic negation that also appearsrepeatedly in ll. 66-69 in a parallel sequence. In the case of the prefix ge-, thereis a tendency to separate it from the rest of the form. This is perhaps to show thatthe accent lies on the root and so connect the initial sound of the root in thealliterative structure. Examples in the poem are: ge bind, ge seldan; ge þyldig, gecrong (wlonc), ge scæft. We can gather from these examples that an edition whichremains close to the manuscript in rendering compounds and in transcribingweak elements in the verse could better evidence the characteristics of the dic-tion that the copyist wanted to highlight in his own transcription.

120



5. Criteria of textuality in The Wanderer and in the ‘oral written texts’

As well as the clues linked to orality that can be found by analysing the man-uscript, attention should also be paid to the textual aspects that characterize TheWanderer and the other compositions that Doane defines oral written texts.10

These texts maintain their links with the oral tradition but they have beentranscribed in the manuscripts and so we need to consider the contribution ofthe copyist. Further, the mobility of these texts is not mirrored by the ‘fixed’ waythey are rendered in printed critical editions. Doane observes: «Old English tex-tual studies have so far developed no systematic way of dealing with the tradi-tional production and motility of texts (mouvance) influenced by orality».11

Pasternack defines these poetic texts as ‘inscribed verse’ as they are an inter-face between the oral and the written tradition: «They do not function in thesame ways as printed texts, and they share enough with oral practices […]. I shalltherefore put orality in my circle of reasoning as an influence on the poetry’s tex-tuality».12 For such texts, that are mainly handed down in codex unicus, we alsoneed to consider what we mean by performance as they could have been textualrepresentations destined to a specific audience. Doane states:

[…] many or most of the preserved manuscripts of Old English poems representrare and unique performances presented to specific and unique audiences; more-over they are likely produced by aural transmission or at least in an atmospherewhere aural habits predominate over visual ones.13

I think that it is important to emphasize that the analysis of The Wanderer doesnot lead me into saying that all sections in the composition have the same rela-tionship between the oral and the written tradition, as will be seen in section 6.

Doane defines the manuscript text as an ‘intratext’ and underlines that itderives from a dialectic relationship between tradition and poet, between thepast where tradition is created and the present in which the text is produced:«The dialectic between productive past and producing present, between tradi-tion and poet, creates an intratext that must be understood in terms of its own

121


10 A.N. DOANE, Oral Texts…, pp. 82-83: «Moreover, because medieval ‘oral written texts’are always chirographically produced, the distinction between oral and written is not so sharpas it appears at first to members of a print culture.».

11 Ivi, p. 75.12 C.B. PASTERNACK, The Textuality of Old English Poetry, Cambridge, Cambridge University

Press, 1995, p. 4.13 A.N. DOANE, Oral Texts…, p. 92.


wholeness and uniqueness».14 Having said this, it is the poet himself who is thebearer of the past and of the tradition that was present and shared even by theaudience: «[…] the poet is the bearer of a past that is a common possession of allthe persons active as poets or audience within a given tradition».15

Modern critics and scholars encounter huge problems in trying to rediscoverthese aspects of orality. This is due to the fact that there is not enough knowl-edge of the matter and that they have only the manuscript text to work on intheir reconstructions. Another factor that has to be considered is the continuedmanipulation the texts have undergone by poets, copyists, those who have triedto re-elaborate the text or even those for whom the poems were destined interms of its reception and interpretation. Pasternack argues that the interpreta-tion of these texts is not given or fixed by the author but that it depends largelyon the links that the recipient makes with other texts he is familiar with: «Thesense the audience makes from the composition depends in large part on therelationships individuals perceive with other texts already heard and read».16

Given this, we ought to investigate intertextual links when interpreting acomposition and render them available in the electronic edition in hypertextform. Doing so would mean that today’s recipient or scholar could give his owninterpretation or evaluate critically the interpretation and the critical edition putforward by the editor.

Unlike authorial texts or texts where the author is well known, those definedby Doane as ‘oral written texts’ form part of a tradition (see the use of formulaicexpressions). These texts have links with orality, display inter-textual referencesand are anonymous. These aspects, according to Pasternack, should be takeninto consideration:

Old English manuscript poetry, including the text that we now call The Wanderer,remains close to its oral roots in its reliance on audible structures and traditionalexpressions, in its fluid relationship to other compositions and in its anonymity.17

It is because of this that I prefer the term ‘composer’ to poet as what has comedown to us is the product, as has been stated, of a tradition to which both thecomposer and the audience belong.

122


14 Ivi, p. 103.15 Ivi, p. 79.16 C.B. PASTERNACK, Anonymous Polyphony and The Wanderer’s Textuality, «Anglo Saxon

England», 20, 1991, pp. 99-122 (107).17 Ivi, p. 99.


In my opinion, it is not productive to try to speculate on the identity of theauthor of a given composition and, above all, to consider the poem as unitary,logically consequential and regulated by the author’s intentions. The criteria oftextuality that have characterized the structure of these texts lie elsewhere.

In my analysis of The Wanderer, I adopted Pasternack’s theory of a composi-tion as made up of blocks, characterized by ‘movements’: «These building blocksstand out as formally distinct, semi-independent units, contrasting in style orcontent with the text that precedes and follows».18

Any differences between parts of the composition should not be judged, ascritics have been wont to do in the past, as stemming from a corrupt tradition orin which additions have been made. Instead, Pasternack’s ‘mosaic composition’should be considered: «Their mosaic-like structure breaks the continuity thatrepresents chronology and causality and so demands that the readers invent theconnections themselves».19 In fact, the unity of the composition is obtainedthrough repetitions and internal correspondences whereas the sections are dis-tinct but thematically linked.

The fluidity and the specific type of composition meant that the originalrecipient, the modern reader and critic have all to recognize the parts and alsothe voice that is present in the ‘aural’ indicators: each interpretation, like everycomposition, will mostly remain and will thus be unique and personal.

Pasternack outlines the main traits of these texts:

Instead of implying an author, Old English verse implies tradition. Formulaicechoes and patterns that are frequently used to express an idea function as a codethat readers can interpret as ‘tradition’. In doing so, they recognize the presenttext’s place in a network of expressions and thought: the movement-structure freesthe reader from thinking of the text as an entity that is indissolubly whole, and thelanguage points the reader away from the present text toward similar expressionsof the same idea that may be spoken or inscribed in any number of other texts.20

6. Analysis of the structure of movement I

To illustrate the textual criteria mentioned above, I will provide a personalanalysis of the structure of movement I:21

123


18 Ivi, p. 100.19 Ivi, p. 104.20 C.B. PASTERNACK, The Textuality…, p. 19.21 An analysis of the structure of The Wanderer was put forward by C.B. PASTERNACK,

Anonymous polyphony…


l. 6 Swa cwæđ eardstapa earfeþa gemyndigwraþra wælsleahta winemæga hryre.Oft ic sceolde ana uhtna gehwylcemine ceare cwiþan nis nu cwicra nan

l.10 þe ic him modsefan minne durresweotule asecgan ic to soþe watþæt biþ in eorle indryhten þeawþæt he his ferđlocan fæste bindehealde his hordcofan hycge swa he wille.

l.15 Ne mæg werig mod wyrde wiđstondanne se hreo hyge helpe gefremman. forđon domgeorne dreorigne oftin hyra breostcofan bindađ fæste. swa ic modsefan minne sceolde.

l.20 oft earmcearig eđle bidæledfreomægum feor feterum sælan

I have presented the text divided into lines and half-lines but have used onlythe capitals and dots that appear in the manuscript. However, I have not insert-ed other capitals and marks to render a structural analysis more visible andunderstandable, unlike in modern critical editions.

At lines 8a and 9b we note the phonic echo of ana «alone», an adjective thatdescribes the condition of the wanderer (see l. 1 anhaga) with nan «no-one».

Similarly interesting is the structure, ll. 10-21, that I would define as a‘frame’. In the middle of the text there are two lines, ll. 15-16, that present witha variation the motif of the grief-stricken man that can neither counter destinynor be of any help. This means that the valiant warrior has to keep his sufferingto himself, literally ‘in his chest’. The two half-lines, 15b wyrde wiđstondan and16b helpe gefremman, present a parallel structure. Another structure that takes intwo lines is found at l. 13b and l. 18b. Here again the motif corresponds and thetwo half-lines have the same formula but employ chiasmus: 13b fæste binde / 18bbindađ fæste. This is the motif of the closing and retaining of suffering within andthere is a variation of the term: l. 13a ferđlocan, l. 18a breostcofan. To conclude,another external ‘frame’ structure can be found at l. 10 and at l. 20. Here thewanderer’s state of mind is presented, with a variation, in the two a-lines at: 10amodsefan / 20a earmcearig. From l. 20 on, a series of formulaic expressions linkedto the theme of exile is introduced. 22

124


22 S.B. GREENFIELD, The Formulaic Expression of the Theme of ‘Exile’ in Anglo-Saxon Poetry,«Speculum», 30, 1955, pp. 200-206.


From the analysis of this block of verses that are measured, as mentioned ear-lier, on 5s, and multiples of 5s, the links to orality are clear. Not only are thereformulae and variations but, above all, a structure based on the repetition andvariations in expressions that begin from a central core and that go on to createechoes in envelope patterns. Even if the composer of The Wanderer cannot belikened to an oral poet, the first part of the text displays all the characteristicsthat are still linked to orality.

Correspondences between lines and parallel constructions can often befound in the composition. In the more homiletic second part, however, thestructure is closer to the models of the written tradition, as can be seen from ll.65-72:

l.65 wintra dæl in woruldrice wita sceal geþyldig .Ne sceal no to hatheort ne to hrædwyrde .ne to wac wiga ne to wanhydig .ne to forht . ne to fægen . ne to feohgifre .ne næfre gielpes to georn ær he geare cunne .

l.70 beorn sceal gebidan þonne he beot spriceđoþþæt collenferđ cunne gearwehwider hreþra gehygd hweorfan wille . 23

These lines are found in the third movement and are presented in a parallelstructure that sees the use at the start of each half-line of ne followed by to, whiletwo repetitions appear in half-line 68a. This anomaly may be due to the separa-tion of the half-lines in the modern edition while the presence of dots might seemto suggest pauses to distinguish the three parallel members. This structure clearlyderives from models and poems of the written tradition, in particular Latin tradi-tion. However, as mentioned previously, the composer also includes a chiasmus,at l. 69b geare cunne and l. 71b cunne gearwe, that creates a frame structure.

This passage lists in the negative form the type of behaviour that the wiseman should not display before concluding that he ought to know, above allbefore making any commitments. I believe that anyone who edits a text of thistype should aim to gather as much information as possible from the manuscript,both in terms of its structure and in the form of the transcription. These are use-ful in reconstructing the links with orality and, above all, the manner of the oraldelivery.

125


23 For this series of lines too, what was included was faithful to the manuscript in termsof capitals and dots while the lines and half-lines were separated.


7. Intertextual correspondences in The Wanderer

Defining ‘intertext’ and therefore the intertextual relationships that link thiselegy with other texts, we need to consider the use of formulaic language and therepetition of certain expressions and themes. This issue is examined in detail inmy Riscrittura e riuso delle immagini poetiche nel Wanderer.24

The use of certain expressions does not necessarily imply that the text is‘fixed’ as there are also variations included. These sometimes bear traces of orig-inality and often result in a different structuring of the elements. Even ‘tradi-tional’ themes and motives such as the theme of exile, can be handled in variousways. This variety can be witnessed in a comparison of The Wanderer and TheSeafarer. Both these elegies present a range of themes: sea voyage, the loss offriends and loved ones and the motif of solitude. While The Seafarer relates anallegorical voyage which the mariner wants to continue joyfully in order to reachhis longed-for destination and be saved in the after-life (having first described asea voyage), The Wanderer, though it offers a Christian perspective, deals morewith the notion of suffering and with the wanderer’s condition. This is achievedthrough comparisons with the past that contrasts starkly with the present. In TheWanderer we are more aware of the elegiac tone throughout and man’s sufferingon earth is underlined more firmly.

Below is a brief list of the corresponding expressions in the two elegies andthose that appear in other Anglo-Saxon poems.

The Wanderer The Seafarer Other poems

l. 1a anhaga l. 62b anfloga Beow. l. 2368a earm anhagal. 6a eardstapa Beow. l. 103a mearcstapal. 111a snottor on mode Ex. l. 439 snottor in sefanl. 20a earmcearig l. 14a earmcearigl. 2b modcearigl. 24a wintercearigl. 4b hrimcealde sæ l. 14b iscealdne sæl. 5a wadan wræclastas l. 15b wræccan lastuml. 21a freomægum feor l. 16a winemægum bidrorenl. 20b eđle bidæled Beow. l. 1275a dreame bedæledl. 45b wineleas guma Gen. l. 1051 wineleas wræccal. 13a ferđlocan l. 55a breosthordl. 14a hordcofan l. 58b hreþerlocanl. 18a breostcofan

126


24 M.G. SAIBENE, Riscrittura e riuso delle immagini poetiche nel Wanderer, in M.G.


The most frequent correspondences, at the level of single expressions or for-mulae, relate to the figure of the lonely wanderer, his state of mind and to thetheme of exile. In both texts there is also a series of correspondent images thatrefer to the sea, to nature, winter and seabirds. However, these images were alsodiscussed in a previous essay and will not be examined here.25

These expressions and formulae are reused and varied and, apart from thetwo elegies examined, can be found in other elegies and in texts from other gen-res such as Beowulf, or religious poems like the Anglo-Saxon Genesis and Exodus.This is why an intertextual analysis of the compositions paves the way for a mul-ti-level interpretation and also allows us to discover different meanings and con-notations.

The loner, the man in exile, unaccompanied in his earthly voyages, is foundnot only in The Wanderer l. 1a anhaga «one who is alone», but also in Beowulf.Here the main character crosses wide expanses of sea l. 2368a earm anhaga«wretched loner» and the solitude is, in both cases, due to a loss (loss of compan-ions and his lord). The same noun is the model for the apax legomenon in TheSeafarer at l. 62b anfloga «he who flies alone». Critics have generally interpretedthis as a reference to a sole marine bird26 but, analysed in context, it is the spiritof the wanderer who soars over the sea. Based on the correspondence in TheWanderer, I would opt for a more metaphorical form of expression related to theseafarer.

There is an interesting link between the figure of the eardstapa (Wand. l. 6a)«the wanderer» and the term that in Beowulf (l. 103a) defines Grendel as mearc-stapa «the vagabond of the border lands». In both cases, the reference is to anexile who inhabits an isolated, marginal place.

The wise man is presented at l. 111 as snottor on mode «wise in his mind» andthe correspondences in this case are with religious poems like Exodus l. 439 snot-tor in sefan «wise in his soul» and Andreas l. 470 wis on gewitte «wise in his mind».

In both elegies, we find correspondent expressions that refer to the state ofsuffering: Wand. l. 20a earmcearig, l. 2b modcearig, l. 24a wintercearig and Seaf.l. 14a earmcearig; we also find expressions that describe the icy sea: Wand. l. 4bhrimcealde sæ and Seaf. l. 14b iscealdne sæ; and, to conclude, a formula that refers

127


Cammarota (ed.), Riscritture del testo medievale: dialogo tra culture e tradizioni, Bergamo,Bergamo University Press, 2005, pp. 125-157.

25 Ibidem.26 C. CUCINA, Il Seafarer. La navigatio cristiana di un poeta anglosassone, Roma, Edizioni

Kappa, 2008.


to the paths of exile: Wand. l. 5a wadan wræclastas «take the path of exile» andSeaf. l. 15b wræccan lastum «on the path of exile».

There are also formulae, as mentioned earlier, related to the theme of exilethat Greenfield has catalogued in an essay.27 These are formulae that conveydeprivation both in terms of the loss of loved ones and a loss of joy. There arecorrespondences between Wand. l. 21a freomægum feor «far from relatives» andSeaf. l. 16a winemægum bidroren «deprived of dear relatives», but the same type offormula occurs outside the elegies, such as Wand. l. 20b eđle bidæled «deprived ofa home» and Beow. l. 1275a dreame bedæled «deprived of joy» or again Wand.l. 45b wineleas guma «man without friends» and Gen. l. 1051 wineleas wræcca«exile without friends». The formula in the Anglo-Saxon Genesis refers to Cain,the Christian symbol of exile, but Anglo-Saxon poets often employ it (the poetwho composed Beowulf defines the monsters as belonging to the ‘stock of Cain’).

In the last examples we can see how one of the two elements in the formulais varied and so the poet, though basing himself on fixed models, manages toinclude variation depending on the context and his particular intention. So, it isapparent that adhering to tradition does not preclude innovation. Another pointto be made is that if we read The Wanderer in the light of intertextual references,single expressions or formulae are enriched by religious or Christian meanings:the theme of exile explored in the figure of the wanderer or the seafarer can beextended, above all in the gnomic sections, to the condition of man on earthwho has to seek his true destination beyond this worldly life.

These last expressions find correspondences in the two elegies and refer tothe motif of the ‘chest’ within which suffering is enclosed. The expressions arevaried and repeated and emphasize how the warrior or valiant man has to strivenot to reveal his true state of mind; he has to maintain a stoic attitude. Examplescan be found in Wand. l. 13a ferđlocan «place of the soul», l. 14a hordcofan «placeof the heart», l. 18a breostcofan «room of the chest» and in Seaf. l. 55a breosthord«treasure of the chest», and l. 58b hreþerlocan «place of the soul». In both compo-sitions, these references are found in lines that are physically close to one anoth-er and refer to the same theme. This precept is expressed also in gnomic textsthat appear in the Exeter Book, like Precepts, Maxims and Homiletic Fragment.

In this study, I wanted to illustrate how intertextual correspondences can beused to demonstrate how these texts are composed. They are informed by apoetic tradition but this does not prevent the composer from introducing varia-

128


27 S. B. GREENFIELD, The Formulaic Expression of the Theme of ‘Exile’ in Anglo-Saxon Poetry,«Speculum», 30, 1955, pp. 200-206.


tions and linking and adapting, in an original way, traditional material for a spe-cific performance and a specific audience.

Intratextual cross-references such as, in The Wanderer at l. 1a, anhaga, echoedat l. 40a earmne anhogan, should be considered; these references unite the com-position and connect the movements. There are also phonic and rhythmic repe-titions that are aimed at oral delivery and serve to emphasize particular parts.Cornell analyses the two elegies and considers especially the variety in the repe-tition, demonstrating that the vowels and consonants most often found are jus-tified by the frequency with which certain themes appear and the use of partic-ular words and their synonyms.28 Cornell’s essay is of interest for its study ofrepetition but it is limited in that it fails to analyse the textual and composition-al criteria of these elegies.

In my opinion, one aspect that deserves special attention is the alliteration,specifically how the initial sound of a word at the end of the b-line can create thehook for the alliterative series in the subsequent line, something that is true ofboth elegies.

It is interesting how in The Wanderer even the second element of a compoundcan have the same function, as, for example, in ll. 35b-36a: «[…] his goldwine /wenede to wiste […]» where the initial sound of wine is echoed for alliteration inthe b-line. Another example is at ll. 68b-69a: «[…] ne to feohgifre / ne næfregielpes to georn […]». This confirms my hypothesis that we should pay atten-tion to the transcription of compounds in the manuscript that have, in general,two distinct elements (unlike in modern editions) and highlight at a phonic lev-el and in alliteration the autonomy of the compound’s second element. This factshould be included in modern editions for what have been described as ‘oralwritten texts’ as the phonic and rhythmic repetitions are crucial in oral compo-sition and diction.

8. Conclusions

In conclusion, it is important to underline the advantages of an electronicedition over a critical edition in printed form. One benefit is that a recipientwho accesses an electronic edition can check in context the editor’s choice ofvariants. He can also follow a personal path and come to his own conclusions.

129


28 M. CORNELL, Varieties of Repetition in Old English Poetry, Especially in The Wanderer andThe Seafarer, «Neophilologus», 65, 1981, pp. 292-307.


An electronic edition of a text is not fixed and definitive but can instead beviewed as the starting point for other interpretations. The aim would be theestablishment of partnerships between editors and users. An electronic editioncan make use of a database or hypertextual presentation through which it wouldbe possible to connect the text to data from other compositions, to dictionaries,to commentaries, or to facsimiles. The text thus becomes an element in a widercultural context. The critical editorial process must, however, remain central butthe text can be reconstructed in terms of its composition and, perhaps, of itsoriginal use via materials that put it into context and that allow comparisonswith other texts.

As for the two Anglo-Saxon elegies examined in this study and for othertexts that preserve links to the oral tradition, that are characterized by mouvance,and which rely, even today, on the reader to interpret them, the choice of anelectronic edition and a hypertext presentation appears an improvement on pastprinted versions. Though printed texts have evident advantages, they render thetext static and authorial. Following a marking up of the text and a hypertextorganisation of the data, the aim in producing an electronic edition would be toopen up the possibility for wider comparison. Another benefit would be thatscholars could highlight in their interpretations the intertextual references thatwere fundamental for the recipient medieval audiences. We have to consider allof this and so create editions that favour rather than obscure crucial aspects andelements in the composition and tradition of medieval Germanic texts.

130



ROBERTO ROSSELLI DEL TURCOUniversità di Torino

IL PROGETTO VERCELLI BOOK DIGITALE:CODIFICA E VISUALIZZAZIONE DI UN’EDIZIONE

DIPLOMATICA GRAZIE ALLE NORME TEI P5

Abstract. The P5 version of the TEI Guidelines for Electronic Text Encoding andInterchange introduced many changes to the TEI encoding schemas. A very interest-ing new feature is the possibility to build a digital facsimile of a manuscript: this is thefirst step to produce an image-based edition of a manuscript (or even of epigraphicinscriptions). It is now possible to use the TEI markup in combination with othertools to produce an electronic edition showing a digitized image and the correspond -ing transcription, linking text and image. This is the method that will be used to cre-ate the Digital Vercelli Book electronic edition.

1. Il progetto Vercelli Book Digitale

Il progetto Vercelli Book Digitale nasce nel 2000 sull’esempio delle edizionidigitali realizzate nel decennio precedente. Proprio sul finire degli anni ‘90,infatti, alcuni progetti di studiosi anglosassoni, quali l’Electronic Beowulf di KevinKiernan e The Wife of Bath’s Prologue on CD-ROM di P. Robinson e N.F. Blake,per la prima volta permettevano di apprezzare le potenzialità di un’edizionedigitale di testi conservati su manoscritti medievali. Grazie a un’edizione digita-le è finalmente possibile studiare un codice senza essere obbligati a spostarsipresso l’istituzione che lo custodisce e che spesso pone restrizioni proprio riguar-do alla consultazione. Per certi aspetti, inoltre, e pur non dimenticando che alcu-ni dubbi possono essere sciolti soltanto consultando l’originale, un’edizionedigitale di un manoscritto fornisce strumenti di indagine superiori rispetto allatradizionale consultazione diretta: visualizzazione di dettagli con illuminazioniparticolari, ingrandimenti a piacere, filtri grafici per analizzare le immagini e la

131


possibilità di effettuare ricerche testuali complesse costituiscono un patrimonioil cui valore risulta evidente allo studioso. Da non sottovalutare, infine, i benefi-ci che la diffusione digitale di un manoscritto consente di raggiungere per quan-to riguarda la preservazione dello stesso; infatti sottrarre un codice al frequentewear and tear da parte degli studiosi interessati a consultarlo significa limitare ipossibili danni e micro-danni cui potrebbe essere soggetto al momento in cuiviene manipolato fisicamente.

1.1. Il manoscritto Codex Vercellensis (Vercelli Book)

Il Codex Vercellensis, o Vercelli Book com’è conosciuto nel mondo anglosassonee ormai anche in Italia, è un manoscritto redatto verso la fine del X secolo, con-tenente una miscellanea di opere a carattere religioso in versi e in prosa.Conservato a Vercelli presso l’Archivio e Biblioteca Capitolare sotto la segnaturaCodex CVII, consta di 136 fogli di sottile pergamena di circa 31x20 cm di dimen-sione, molto ben conservati, ciascuno dei quali contiene da 23 a 32 righe ditesto. Secondo il parere pressoché unanime degli studiosi il manoscritto è operadi un unico copista, particolarmente attento e minuzioso nell’uso della grafia delperiodo, la minuscola quadrata anglosassone. Il Vercelli Book ospita 23 omelie inprosa e 6 componimenti poetici secondo il metro allitterativo anglosassone.Molto probabilmente il committente intendeva riunire testi religiosi in prosa epoesia in modo da comporre un prezioso florilegio spirituale, utile per la medi-tazione e la preghiera. La presenza del manoscritto a Vercelli è provata sin dal-l’inizio del XII sec.,1 ma il percorso che, in qualche momento dell’XI secolo, haportato questo codice redatto nell’Inghilterra meridionale a trovare come pro-pria sede definitiva la città piemontese non è affatto chiaro: secondo una delleipotesi più plausibili si tratterebbe del dono di un pellegrino, grato per l’ospita-lità ricevuta a Vercelli durante il lungo cammino verso Roma; secondo un’altrateoria il manoscritto avrebbe fatto parte di una più vasta collezione di oggettipreziosi depositata a Vercelli da un prelato anglosassone punito per i suoi com-portamenti dalle gerarchie ecclesiastiche.

Il Vercelli Book riveste un’importanza particolare per gli studi di lingua e lette-ratura anglosassone: è uno dei quattro manoscritti risalenti alla fine del X secolo(gli altri sono l’Exeter Book, il Cotton Vitellius A XV e lo Junius 11) che contengo-no circa il 90% di tutta la produzione poetica anglosassone; di questi, inoltre, è

132

ROBERTO ROSSELLI DEL TURCO

1 In calce al f. 24v, infatti, si legge un verso del Salmo XXVI in scrittura carolina minusco-la tipica dell’Italia del Nord e databile intorno al 1100.


l’unico a essere conservato al di fuori dei confini del Regno Unito. I testi lettera-ri contenuti in questi codici sono essenziali per la comprensione della lingua edella cultura dell’Inghilterra anglosassone, e sono oggetto di studio da parte diricercatori di tutto il mondo.

Alcuni dei componimenti poetici in esso contenuti spiccano per qualità arti-stica: così il Sogno della Croce e i due poemetti firmati da Cynewulf, Elena e i Fatidegli Apostoli; si noti, inoltre, che non esistono altre copie di quasi tutti i testi poe-tici del Vercelli Book. Per quanto riguarda le omelie, queste rivestono particolareinteresse nella storia della letteratura religiosa del periodo, nonché per gli studio-si della prosa anglosassone; e anche in questo caso alcuni di questi testi non han-no corrispondenti nella documentazione letteraria anglosassone.

Il lascito del manoscritto ai religiosi vercellesi, infine, testimonia della parti-colare importanza che Vercelli ebbe per lungo tempo nel Medioevo come puntodi scambio e di sosta sulla via dei pellegrinaggi per Roma.

1.2. Obiettivo del progetto

Lo scopo primario del progetto è quello di produrre una versione digitale delVercelli Book, in maniera tale da proporre allo studioso un’alternativa praticabileed efficace rispetto alla consultazione diretta del manoscritto. Il Vercelli BookDigitale raccoglierà tutti i testi conservati nel manoscritto, proponendoli in piùformati, accompagnandoli con strumenti di consultazione e analisi testuale. Perraggiungere questi obiettivi verrà creato un programma di navigazione delleimmagini che permetterà di sfogliare virtualmente il manoscritto in modi diver-si: esaminando i singoli fogli che lo compongono, utilizzando se così si desiderastrumenti quali ingrandimenti di specifici dettagli, etc.; esaminando la singolaimmagine con accanto l’edizione diplomatica del manoscritto, con la possibilitàdi collegare il testo dell’edizione al punto corrispondente dell’immagine e vice-versa; confrontando il testo dell’edizione diplomatica e quello dell’edizione cri-tica, senza rinunciare alla possibilità di richiamare rapidamente l’immagine cor-rispondente; effettuando ricerche testuali complesse sul testo del manoscritto,sia a livello paleografico, sia a livello filologico-letterario. In breve, ci si proponedi fornire allo studioso uno strumento completo e facilmente utilizzabile pereffettuare ricerche di vario tipo sul Vercelli Book, disponibile su supporto ottico(CD-ROM/DVD). Grazie a tale edizione sarà possibile diffondere la conoscen-za del manoscritto e al tempo stesso preservarlo dalle sollecitazioni che la con-sultazione diretta inevitabilmente comporta.

133

Il progetto Vercelli Book digitale


1.3. Piano operativo

Per raggiungere i suoi obiettivi, il Vercelli Book Digitale prevede una strutturamodulare del progetto tale da permettere di eseguire in parallelo, ove possibile,operazioni diverse. Tra le più importanti possiamo citare la digitalizzazione delmanoscritto effettuata da personale tecnico specializzato usando una fotocameradigitale ad altissima risoluzione; una trascrizione completa di tutti i testi, effet-tuata utilizzando una marcatura approfondita del testo conforme alle normeTEI, che servirà da base per la successiva edizione diplomatica, nonché per lericerche testuali; un’edizione diplomatica dei testi, strumento essenziale per lostudioso, e anche per lo studente, di letteratura anglosassone, base inoltre peruna successiva edizione critica; l’inclusione di materiale supplementare, qualifonti e versioni alternative dei testi, la prima trascrizione del manoscritto ad ope-ra di C. Maier (conservata a Londra), le prime edizioni e traduzioni in varie lin-gue dei testi, una bibliografia esaustiva riguardo il Vercelli Book e le opere in essocontenute, etc; la progettazione e implementazione di un software con interfac-cia grafica che permetta la navigazione delle immagini e dei testi, nonché di tut-ti i documenti correlati, inclusivo di un motore di ricerca XML, che permettaricerche complesse sulla base della codifica TEI.

1.4. Stato di avanzamento del progetto

La digitalizzazione del manoscritto è stata portata a termine con successo:grazie ai fondi erogati dalla Regione Piemonte possiamo disporre di un set diimmagini di ottima qualità sulla base delle quali costruire l’edizione. Per quan-to riguarda i testi la trascrizione è completa al 60-70% circa, e verrà ultimatanel corso del 2009; seguirà un periodo di revisione incrociata da parte dei col-laboratori impegnati nel progetto, in modo da ridurre al minimo gli errori dimarcatura e, al tempo stesso, verificare la necessità di eventuali modifiche alloschema di codifica adottato (basato sugli schemi TEI, versione P5).2 Per quan-to riguarda la visualizzazione sono stati creati dei fogli di stile XSL-T che per-mettono la conversione nel formato HTML e il caricamento dei testi in unqualsiasi navigatore Web; per la versione finale è previsto l’uso di un software

134


2 In attesa della versione P5 delle norme TEI è stato utilizzato uno schema di codificatemporaneo, quello del progetto Digital Scriptorium (http://www.scriptorium.columbia.edu/),basato sulla versione precedente, la TEI P4, arricchita con caratteristiche mirate a migliorarela qualità della trascrizione.


specifico, che permetta l’integrazione del motore di ricerca e di altre funziona-lità, da sviluppare come progetto separato e riutilizzabile per altre edizionidigitali.3

2. Collegamento testo immagine nelle edizioni digitali

La Filologia Digitale è ancora, potremmo dire, nella sua infanzia, ciò malgra-do si è già affermata una tipologia delle edizioni digitali piuttosto ampia. Quellache in inglese viene chiamata image-based digital edition (o full digital edition) nonconsiste semplicemente in un’edizione critica o diplomatica in forma ipertestua-le, ma include le immagini del manoscritto da cui è tratto il testo ed è arricchitada una serie di funzionalità che, pur tenendo a mente lo status pionieristico del-la disciplina, al momento sono considerate «standard» in quanto caratterizzantiquesto specifico tipo di edizione: la visualizzazione delle immagini in formatie/o risoluzioni diverse, una «lente d’ingrandimento» virtuale, l’evidenziazione dispecifici dettagli del manoscritto, spesso basata sui risultati di un’operazione direstauro fisico o virtuale, e un motore di ricerca testuale.

Molte di queste funzionalità, come si può evincere dalla lista propostapoc’anzi, sono strettamente correlate alla parte iconografica dell’edizione, mauna in particolare assume un’importanza centrale: il rapporto fra testo e imma-gine. Le edizioni digitali degli ultimi anni, ad esempio l’Electronic Junius el’Electronic Exeter Book di B.J. Muir, propongono allo studioso un collegamentofra il testo dell’edizione e le immagini del manoscritto. Questo collegamentopuò assumere due forme:

· collegamento mirato (hot-spot): una specifica area dell’immagine viene evi-denziata in maniera tale che, interagendo con la stessa, vengono visualizzatedelle informazioni quali note editoriali, la versione migliorata di un dettaglio,un commento al testo, etc.; non tutti gli hot-spot hanno ovviamente una fun-zione di collegamento con il testo, perché non tutti i dettagli che richiedonoil ricorso a questo metodo sono di tipo testuale: si pensi ad esempio a caratte-ristiche come la presenza di decorazioni o figure, oppure di fenomeni (proba-tio pennae, sigle) che non appartengono alla definizione del testo critico insenso stretto;

135


3 Il navigatore sarà messo a disposizione della comunità accademica come software opensource una volta ultimato lo sviluppo.


· collegamento generalizzato: tutto il testo dell’edizione viene messo in rela-zione diretta con le immagini, o parti di immagine, corrispondenti, in mododa poter accedere facilmente all’area del manoscritto partendo dal testo, eviceversa; questo collegamento può avvenire su livelli diversi: foglio o partedello stesso, riga, persino a livello di singola parola.

Alcune edizioni digitali offrono uno dei due tipi di collegamento, o entram-bi,4 ma non sembra esservi stata una seria riflessione sul piano teorico e metodo-logico riguardo questa specifica caratteristica; piuttosto essa sembra essere deriva-ta direttamente da altre forme di comunicazione multimediale e dal semplice fat-to che la tecnologia lo consente. Questo è per certi versi sorprendente, in quantoqualsiasi curatore di edizione critica si preoccuperebbe, ad esempio, di spiegare egiustificare eventuali innovazioni nel layout tradizionale dell’apparato critico.

Con questa osservazione non si vuole respingere l’introduzione di tale meto-do nelle edizioni digitali, al contrario, sono convinto che si tratti di una caratte-ristica assai utile. Tuttavia farne uso semplicemente «perché possiamo» non sem-bra una base teorica sufficiente; ritengo che debba esserci una discussione gene-rale sul piano metodologico (è una caratteristica desiderabile? è facilmente inte-grabile con altre convenzioni filologiche?), su quello tecnico (come può essererealizzata?) e, ultimo ma non meno importante, su quello dell’implementazione(quale interfaccia utente può essere considerata ottimale?).

2.1. Aspetti metodologici e implementazione

Una seria riflessione teorica sull’integrazione di testo e immagine nelle edi-zioni digitali richiederebbe un articolo separato, soprattutto in considerazionedel fatto che tale argomento rientra nel ben più ampio tema del superamentodell’interfaccia bidimensionale dell’edizione tradizionale, ed è quindi stretta-mente correlato a innovazioni quali un generale ripensamento di forma e fun-zione dell’apparato critico. Pertanto mi limito a proporre qualche considerazio-ne generale, utile per un proseguimento della discussione in altra sede.

In generale, direi che non ci sono dubbi sull’opportunità di ricorrere a colle-gamenti mirati di tipo hot-spot: si tratta di un metodo assolutamente non invasi-vo, in quanto attivabile a discrezione dell’utente, che permette di fornire infor-mazioni ricche e ben articolate riguardo a specifici dettagli del manoscritto.L’implementazione proposta dalle edizioni digitali attualmente disponibili,

136


4 È il caso delle edizioni elettroniche create da B.J. Muir, si veda il sito <http://www.evel-lum.com/>.


basata su finestre temporanee di tipo pop up o comunque distinte rispetto allafinestra principale di visualizzazione delle immagini, è del tutto congrua ed effi-cace in rapporto alle informazioni trasmesse (transitorie e supplementari rispet-to al resto dell’edizione). Si tratta in effetti di una caratteristica delle più recentiedizioni elettroniche molto diffusa e apprezzata.

Una valutazione relativa a costi e benefici, dunque, riguarda soprattutto ilcollegamento diretto ed esaustivo fra il testo dell’edizione e le immagini delmanoscritto, soprattutto nel caso si tratti di un’associazione a livello di riga o sin-gola parola. Un’obiezione possibile è che si tratterebbe di un abbellimento nonnecessario, quasi un «giocattolo», perché uno studioso competente non avrebbedifficoltà a collegare manualmente un passo del testo nell’edizione critica con ilpunto del manoscritto corrispondente. L’osservazione è senza dubbio corretta,ma limita in maniera inaccettabile il pubblico di riferimento per le edizioni digi-tali; volendo spingere agli estremi la logica di questa obiezione, inoltre, sipotrebbe affermare che lo studioso competente non abbisogna d’altro se non diuna buona riproduzione fotografica del manoscritto: perché includere una partetestuale come un’edizione diplomatica dello stesso, allora? Lo stesso concetto diimage based digital edition sarebbe discutibile e dovrebbe cedere il passo a due tipidistinti di edizione elettronica: il facsimile digitale e l’edizione critica o diploma-tica in formato ipertestuale.

Non è tuttavia indispensabile ricorrere a un’eccessiva esasperazione dialetti-ca per respingere questo tipo di obiezione. Il collegamento generalizzato fratesto e immagine può contare numerosi argomenti a proprio favore, cito i piùimportanti:

· si tratta di un innegabile ausilio nella navigazione, soprattutto nel caso in cuil’edizione riguardi un manoscritto di grandi dimensioni, o includa più di unmanoscritto: anche se limitato al collegamento fra i fogli del manoscritto e iltesto corrispondente eviterebbe di dover cercare e visualizzare manualmentel’immagine o il brano desiderato in seguito agli spostamenti in avanti o indie-tro che si verificano durante la consultazione;

· se il collegamento è fisso, risolve un problema relativo all’interfaccia utente: lavisualizzazione in maniera indipendente di testo o immagini implica che lostatus dell’interfaccia non è mai certo, in quanto l’utente non può essere sicuroche il testo corrisponda effettivamente all’immagine visualizzata, e viceversa;

· si tratta di un ausilio per ricerche da parte di studiosi non esperti di una parti-colare tradizione testuale o linguistica;

· per lo stesso motivo costituisce un prezioso ausilio didattico in quanto permet-

137



te allo studente di seguire e confrontare il testo riga per riga, se non addirittu-ra parola per parola;

· a prescindere dalle modalità di visualizzazione, costituisce un importante fra-mework per altre funzionalità, ad esempio la ricerca nel testo di parole o detta-gli paleografici, sempre in base alla codifica effettuata.5

Un collegamento «fisso», dunque, tale per cui ad un avanzamento nella con-sultazione del testo critico corrisponda un aggiornamento dell’immagine visua-lizzata in modo da avere una sincronizzazione perfetta fra testo e immagine,offre numerosi vantaggi rispetto a un’edizione in cui testo e immagini sianosemplicemente giustapposti. Questi vantaggi, tuttavia, non devono essere conse-guiti al prezzo di un’eccessiva complessità tecnica, tale da rendere difficoltosa lacreazione di un’edizione digitale di questo tipo. Risulta dunque fondamentalenon solo il tipo di soluzione tecnica per realizzare tale collegamento, ma anchel’esistenza o meno di strumenti che rendano conveniente per lo studioso appli-care tale soluzione nell’esecuzione del proprio progetto di edizione digitale.

2.2. Realizzazione tecnica

La precedente versione degli schemi di codifica e delle norme TEI (TEIP4)6 offriva un supporto generico per quanto riguarda le immagini, ma nessunmeccanismo specifico per collegare immagini e testo. Tale collegamento, per-tanto, non poteva essere realizzato al livello della codifica TEI, ma solo in unmomento successivo, quale la visualizzazione in un navigatore Web. L’uso diimage maps all’interno di pagine HTML si presta molto bene, in effetti, alla crea-zione di hot-spot e alla visualizzazione dei contenuti per mezzo di finestreHTML separate.

Un collegamento diretto e costante, ad esempio fra riga dell’edizione e rigadel manoscritto, sembra invece impresa più complicata, ma usando una combi-nazione di tabelle HTML, semplici istruzioni Javascript e un’adeguata prepara-zione delle immagini anche questo è possibile, come mostra il sito sperimentaleDe Casu CiƷaris Dutis Regis Iabin <http://web.uvic.ca/hrd/lydgate/>. Per primacosa è necessario tagliare le immagini in «fette» (slices) sottili, ciascuna delle qua-li corrisponde a una riga:

138


5 Si pensi alla possibilità di marcare una per una le particolarità grafiche del manoscrittoe di poterle poi rintracciare automaticamente visualizzandole nel contesto grafico e testuale.

6 Ancora disponibile sul sito del Consorzio TEI: <http://www.tei-c.org/Guidelines/P4/>.


Successivamente le immagini sono ricomposte in un frame HTML usandouna tabella per unire le singole parti; a ciascuna immagine viene associataun’istruzione in Javascript per associarla al testo corrispondente:

<table border=”0” cellpadding=”0” cellspacing=”0” width=”604”>[…]

<tr valign=”top”><!— row 1 —><td><img id=”ln0” onclick=”parent.ShowTrans(12,1,0)”

style=”cursor: pointer; cursor: hand;” src=”12_1/0.jpg”width=”604” height=”15” border=”0”></td>

<td><img src=”12_1/shim.gif” width=”1” height=”15”border=”0”></td>

<td><img id=”img0” src=”12_1/shim.gif” width=”16” height=”7”border=”0”></td>

</tr><tr valign=”top”><!— row 2 —>

<td><img id=”ln1” onclick=”parent.ShowTrans(12,1,1)”style=”cursor: pointer; cursor: hand;” src=”12_1/1.jpg”width=”604” height=”19” border=”0”></td>

<td><img src=”12_1/shim.gif” width=”1” height=”19”border=”0”></td>

<td><img id=”img1” src=”12_1/shim.gif” width=”16”height=”7” border=”0”></td>

</tr>[…]</table>

mentre in un frame aperto a lato del primo viene inserito il testo in un’altratabella; anche in questo caso il collegamento viene effettuato per mezzo di istru-zioni Javascript:

<tr><td id=”12_1_18”><span class=”linenum”>12/1/18 </span></td><td><img src=”../images/shim.gif” width=”16” height=”7”

border=”0” id=”img12_1_18”></td><td class=”linecell” onclick=”parent.ShowImage(12,1,18)”><span class=”red”>De caʃu Cıʒarıs dutıs Regıs Iabın</span></td>

139



</tr><tr>

<td><img src=”../images/shim.gif” width=”16” height=”7”border=”0”></td> <td></td><td></td></tr>

Nell’immagine proposta qui sotto è visibile il risultato finale:

140


Collegamento testo-immagine con HTML e Javascript

Questo metodo è piuttosto ingegnoso e, limitatamente a una presentazionesu navigatore Web, efficiente, ma presenta numerosi svantaggi:

· è piuttosto complicato nella messa in opera;

· richiede molto tempo per la suddivisione in «fette» delle immagini;


· questa operazione richiede molta attenzione perché le dimensioni delle «fette»sono fisse, quindi è molto probabile che la corrispondenza di alcune di esse conuna riga del manoscritto sia imperfetta;

· non è compresa la realizzazione di hot-spot, che deve essere eseguita separata-mente (ad esempio con il metodo delle image maps) aggiungendo ulteriorecomplessità;

· l’allineamento testo-immagine viene effettuato al livello della presentazione avideo: la codifica del testo, se effettuata in XML, ne resta del tutto all’oscuro.

Quest’ultimo punto è particolarmente pregiudizievole per la creazione diedizioni digitali basate su una codifica in XML, ed è il motivo per cui alcuni stu-diosi hanno formulato proposte o creato strumenti allo scopo di inserire infor-mazioni di tipo spaziale (coordinate cartesiane delle aree delle immagini) diret-tamente nella marcatura, in modo da avere un collegamento diretto fra testo eimmagini.

È questo il caso del software EPPT (Edition Production and PresentationTechnology),7 un software sviluppato sotto la guida di Kevin S. Kiernan che offreal curatore tutti gli strumenti necessari per la creazione di edizioni digitali e infuturo permetterà di visualizzarle in un navigatore Web grazie all’uso di fogli distile. Due dei moduli di EPPT, ImagText e xTagger, permettono di integrareimmagini e testo inserendo le coordinate come attributi degli elementi testualiin maniera semplice ed efficace; infatti è sufficiente delimitare con il cursoredel mouse le aree delle immagini per ricavarne le coordinate. Sfortunatamentetale software è stato concepito prima della pubblicazione delle norme TEI P5,per cui non è direttamente compatibile8 né con la versione P4, né con la piùrecente P5.

141


7 Si veda il sito <http://beowulf.engl.uky.edu/~eft/eppt/>.8 Si noti, tuttavia, che EPPT è in grado di caricare e modificare documenti XML creati

con qualsiasi DTD, quindi un eventuale problema di compatibilità può verificarsi solo nelcaso si vogliano riportare sotto TEI documenti creati o modificati con EPPT. Una personaliz-zazione degli schemi TEI, o un foglio di stile XSL-T possono risolvere il problema.


Collegamento testo-immagine con EPPT

Una proposta relativa alla implementazione di Digital Facsimiles in modocompatibile con le norme TEI è stata avanzata da R. Gartner e L. Burnard nel2001. Il documento originale, Draft Recommendations for TEI Digital Facsimiles, èancora oggi disponibile sul Web9 e mostra chiaramente come, nelle intenzionidegli autori, il collegamento testo-immagine dovesse essere realizzato per mezzodi puntatori ed elementi standard degli schemi di codifica TEI:

The TEI Guidelines propose a number of methods for aligning parts of anSGML or XML document (see www.tei-c.org/Guidelines/SA). These may bebriefly summarised as follows

1. use the corresp attribute to assert that a <pb> (or other reference point) inthe transcription corresponds with a <pb> in the facsimile (or the reverse);

142


9 All’URL <http://users.ox.ac.uk/~lou/wip/digfax.html>.


2. alternatively, use a stand-off <link> element to assert the associationbetween the two <pb> elements;

3. for cross-document linking, an intermediate <xptr> may be used as thetarget.10

Tuttavia la proposta non ebbe seguito e gli sviluppatori della TEI (compresoLou Burnard) hanno scelto di seguire un’altra strada: fra le novità della versioneP5 spicca la nuova sezione «Digital facsimiles» nel capitolo 11 - Representation ofPrimary Sources,11 che fornisce tutte le informazioni necessarie riguardo la crea-zione di un facsimile digitale e l’associazione del testo della trascrizione con leimmagini di un manoscritto. Come primo risultato dell’inclusione del modulotranscr nel proprio schema di codifica un nuovo attributo, facs, è disponibilecome attributo globale:

@facs (facsimile) points to all or part of an image which corresponds with thecontent of the element12

Questo attributo può essere usato in qualsiasi elemento TEI per associare ilcontenuto dello stesso a un’immagine:

<p n=”1” facs=”#para1.jpg”> <head facs=”#head.jpg”> <pb facs=”#page1.jpg”/>

Si tratta di un meccanismo estremamente potente e flessibile che permette dicollegare il testo desiderato a un’immagine. Non sarebbe sufficiente da solo, tut-tavia, a permettere l’associazione di testo e immagini a livello di riga e parola (ameno di non voler ricorrere a una «scomposizione» dell’immagine in manierasimile a quanto visto per il metodo basato su HTML e Javascript); per raggiun-gere questo obiettivo sono stati introdotti alcuni nuovi elementi:

<facsimile> contains a representation of some written source in the form of a setof images rather than as transcribed or encoded text. <surface> defines a written surface in terms of a rectangular coordinate space,optionally grouping one or more graphic representations of that space, and rec-tangular zones of interest within it.

143


10 Ivi, sez. 4 «Aligning transcription and facsimile».11 Si veda <http://www.tei-c.org/release/doc/tei-p5-doc/en/html/PH.html>.12 Ibidem.


@start points to an element which encodes the starting position of the text cor-responding to the inscribed part of the surface. <zone> defines a rectangular area contained within a surface element.13

L’elemento <facsimile> è di tipo strutturale e si pone allo stesso livello di<text> o addirittura in alternativa a quest’ultimo: può infatti contenere gli ele-menti <front> e <back> esattamente come <text>. Quando il modulo transcrviene aggiunto allo schema di codifica è possibile scegliere fra

– un <teiHeader> e un <text>– un <teiHeader> e un <facsimile>– un <teiHeader>, un <facsimile> e un <text>

Queste possibilità strutturali permettono una grande flessibilità, perché èpossibile creare un facsimile contenente soltanto le immagini del manoscritto,oppure un documento TEI che comprenda sia il facsimile, sia la trascrizione delmanoscritto; in quest’ultimo caso è anche possibile scegliere di effettuare il col-legamento testo-immagine, oppure no.

Un facsimile contenente soltanto le immagini avrebbe dunque questa strut-tura:

<TEI> <teiHeader> <!—…—> </teiHeader> <facsimile>

<graphic url=”page1.png”/> <graphic url=”page2.png”/> <graphic url=”page3.png”/> <graphic url=”page4.png”/>

</facsimile> </TEI>

Mentre una trascrizione che comprenda i riferimenti alle immagini (grazie afacs) avrebbe questo aspetto:

<TEI><teiHeader>

144


13 Ibidem.


<!—…—></teiHeader><text>

<body>[…]<pb facs=”page1.png”/>

<!— inserire qui il testo di pagina 1 --><pb facs=”page2.png”/>

<!— inserire qui il testo di pagina 2 -->[…]

</body></text>

</TEI>

Nel secondo caso l’unico collegamento possibile fra testo e immagine è a livel-lo della singola pagina del manoscritto: con un appropriato foglio di stile XSL-Tè possibile far comparire le immagini e i testi in due frame separati di una paginaHTML, e fare in modo che allo scorrimento del testo venga visualizzata l’imma-gine corrispondente.14 Non vi sono informazioni sufficienti, tuttavia, per effet-tuare un collegamento a livello di riga o singola parola, e non è nemmeno possi-bile creare degli hot-spot. Per ottenere questi risultati è indispensabile creare undocumento TEI del terzo tipo, ovvero uno che contenga sia l’elemento <facsimi-le>, sia <text>, e sfruttare gli altri elementi menzionati per inserire le coordinatedelle aree desiderate e stabilire i collegamenti con gli elementi contenenti testo.

La soluzione più efficace e completa, pertanto, richiede l’uso degli elementi<surface> e <zone> all’interno di <facsimile>

– per definire precisamente le aree delle immagini;– per associare tali aree e anche immagini secondarie all’immagine principale;– per collegare gli elementi di testo a tali aree e alle immagini secondarie.

Ogni area viene individuata per mezzo di un sistema di coordinate cartesia-ne, che vengono registrate come valori dei seguenti attributi di <surface> e<zone>:

ulx, uly coordinate x e y dell’angolo superiore sinistro (upper left)

145


14 In alternativa all’uso del solo elemento <text> sarebbe teoricamente possibile affiancar-gli un <facsimile>, ma in assenza di collegamenti fra testo e immagini (in particolare a speci-fiche parti delle immagini) il vantaggio consisterebbe soltanto nel poter raccogliere all’internodi <facsimile> tutti i puntatori alle immagini; in tal caso, tuttavia, sarebbe indispensabileaggiungere dei riferimenti all’interno di <text>, con aumento della complessità della codifica.


lrx, lry coordinate x e y dell’angolo inferiore destro (lower right)

Il sistema di riferimento è comune per tutti gli elementi che registrano lecoordinate, dunque non vi è possibilità di errore (a condizione, ovviamente, chei valori spaziali inseriti negli attributi siano corretti); l’unica limitazione, conse-guenza del sistema adottato, è che le aree definite possono essere soltanto di tiporettangolare.15

L’elemento <surface> individua l’area occupata da scrittura in un’immagine:

<surface ulx=”50” uly=”20” lrx=”400” lry=”280”><graphic url=”page1.png”/>

</surface>

e può contenere più di un elemento <graphic>, ad esempio per specificare versio-ni diverse della stessa immagine (bassa e alta risoluzione nell’esempio che segue):

<surface><graphic url=”page1-highRes.png”/><graphic url=”page1-lowRes.png”/>

</surface>

In alternativa, invece di <graphic> può contenere uno o più elementi<zone>: quest’ultimo individua una specifica area di un’immagine all’internodello stesso sistema di coordinate usato da <surface>; può trattarsi di tutta l’areadell’immagine:

<zone ulx=”0” uly=”0” lrx=”500” lry=”321”> <graphic url=”page1.png”/>

</zone>

146


15 Nel caso in cui le aree in questione siano orientate in diagonale invece che in orizzon-tale o verticale, ad esempio in manoscritti di autori contemporanei, è comunque possibileeffettuare una trascrizione usando gli schemi TEI: le coordinate delle aree in questione devo-no essere inserite nel documento TEI usando il formato SVG; sarà poi compito del softwaredi visualizzazione evidenziare tali aree sullo schermo ed effettuare il collegamento con iltesto. Il linguaggio SVG (Scalable Vector Graphics) è uno schema di codifica XML che permet-te di descrivere oggetti grafici bidimensionali sotto forma di coordinate; l’inclusione in undocumento TEI risulta quindi assai semplice: è sufficiente specificare il namespace SVG perpoter impiegare direttamente gli elementi disponibili.


oppure di una porzione ristretta, utile per creare un hot-spot in un momento suc-cessivo:

<zone ulx=”120” uly=”48” lrx=”143” lry=”56”> <graphic url=”gloss.png”/><desc>Scribe gloss</desc>

</zone>

Nell’esempio che segue viene definita una superficie occupata da scritturache contiene due <zone> corrispondenti ai casi descritti sopra:

<facsimile><surface ulx=”50” uly=”20” lrx=”400” lry=”280”>

<zone ulx=”0” uly=”0” lrx=”500” lry=”321”> <graphic url=”page1.png”/>

</zone><desc>first page</desc><zone ulx=”120” uly=”48” lrx=”143” lry=”56”>

<graphic url=”gloss.png”/><desc>Scribe gloss</desc>

</zone> </surface>

</facsimile>

Per associare il testo dell’edizione agli elementi <surface> e <zone> è neces-sario assegnare un identificatore univoco a ciascun elemento del facsimile usan-do l’attributo xml:id; in seguito grazie all’attributo facs qualsiasi elemento concontenuto testuale del documento può fare riferimento alla rispettiva immagineo area. Viceversa, per allineare le immagini al testo è necessario assegnare unidentificatore univoco a ciascun elemento che contiene testo usando l’attributoxml:id; in seguito sarà sufficiente usare l’attributo start negli elementi <surface>e <zone> per specificare l’elemento contenente il testo corrispondente. Un sem-plice esempio di collegamento bidirezionale:

<text><body>

<div><pb facs=”#page1” n=”1” xml:id=”text_page_1”/><p>Lorem ipsum … <gloss facs=”#det1”>semper</gloss></p>

</div></body>

</text><facsimile>

<surface start=”#text_page_1” ulx=”50” uly=”20”lrx=”400

147



lry=”280”><zone xml:id=”page1” ulx=”0” uly=”0” lrx=”500” lry=”321”>

<graphic url=”page1.png”/></zone><desc>First page.</desc><zone xml:id=”det1” ulx=”120” uly=”48” lrx=”143” lry=”56”>

<graphic url=”gloss.png”/><desc>Scribe gloss.</desc></zone>

</surface></facsimile>

3. Strumenti software per la registrazione delle coordinate

Per quanto il meccanismo predisposto dalla TEI sia estremamente logico edefficace, due delle operazioni necessarie, individuare i valori delle coordinate einserirli correttamente nel proprio documento TEI XML, sono a prima vistalaboriose e impegnative. Fortunatamente esistono numerosi strumenti softwareche aiutano nell’individuazione delle coordinate: alcuni programmi di disegnoin formato bitmap, strumenti specifici per programmatori di siti web, etc.

Lo strumento ideale sarebbe un software che, una volta determinate le coor-dinate, si prenda cura anche dell’inserimento delle stesse come valori degli attri-buti in un documento XML. Come abbiamo visto il progetto EPPT, creato conlo scopo di aiutare il filologo nella preparazione di un’edizione digitale, includestrumenti che permettono di effettuare sia il rilevamento delle coordinate, sia laloro inclusione all’interno del documento XML alla base della propria edizione.Sfortunatamente non è, almeno per il momento, conforme allo standard TEIP5, per quanto non sarebbe difficile effettuare una conversione in tale formato.

Un software molto interessante, concepito specificamente per questo scopo,è l’Image Markup Tool sviluppato da M. Holmes presso l’Università di Victoria(Canada).16 Si tratta di uno strumento open source che permette all’utente di defi-nire aree rettangolari in un’immagine e di associarle ad annotazioni di qualsiasitipo. Non richiede nessuna conoscenza preliminare della TEI, e nemmeno diXML, perché salva i dati direttamente nel formato TEI P5. Grazie all’interfacciautente piuttosto amichevole, infine, può essere impiegato con profitto dopo

148


16 La pagina del progetto si trova all’indirizzo <http://www.tapor.uvic.ca/~mholmes/image_markup/index.php>. Il programma è disponibile soltanto per Windows, ma può esse-re utilizzato anche sotto Linux usando Wine, e sotto MacOS X con DarWine.


poco tempo. Nella schermata proposta qui sotto IMT viene usato per marcare lerighe del f. 112r del Vercelli Book, come pure altre aree dell’immagine che verran-no utilizzate per la preparazione di hot-spot.

149


Annotazione di un’immagine del Vercelli Book usando IMT

Il file XML in standard TEI P5 prodotto da IMT può essere utilizzato in qual-siasi momento per produrre una versione in formato HTML, adatta quindi peressere visualizzata con un navigatore Web. Il software si prenderà automaticamen-te cura di generare non soltanto la pagina HTML, ma anche tutto il codiceJavascript necessario per creare un collegamento testo-immagine di tipo bidirezio-nale. In aggiunta, le immagini che superano le dimensioni stabilite dall’utentesono ridimensionate per la visualizzazione nel navigatore, tuttavia verrà proposto


l’ingrandimento delle aree oggetto di annotazione mostrando porzioni dell’imma-gine originale. La versione HTML può essere salvata per essere modificata in unmomento successivo. Un sito creato usando IMT, utile per valutare l’utilità di que-sto software, è quello del progetto Le mariage sous l‘Ancien Régime (http://mariage.uvic.ca/). Nell’immagine che segue è possibile vedere il risultato dell’annotazionedel f. 112r del Vercelli Book usando IMT descritta sopra: sono visibili sia gli hot-spot,sia le righe di testo, distinti in base al colore; per ogni riga di testo è stata annota-ta la trascrizione, visualizzata cliccando sull’area rettangolare evidenziata in bluper mezzo di una finestra a scomparsa; tutte le aree annotate, infine, possono esse-re ingrandite con un doppio click del mouse e possono essere corredate di un bre-ve testo informativo visualizzato quando il mouse passa sopra di esse.

150


La pagina web creata automaticamente da IMT


IMT è uno strumento molto efficace, ma presenta alcune limitazioni: la ver-sione attuale supporta l’annotazione di una sola immagine per volta, quindi nonè possibile produrre un documento che raccolga la marcatura relativa a un inte-ro manoscritto; non è previsto alcun supporto per la modifica del documentoprodotto; i file TEI XML prodotti, infine, devono essere integrati successiva-mente con quelli relativi alla trascrizione. Di conseguenza IMT deve essere inte-grato in un ambiente e metodo di lavoro che includa altri strumenti software, inparticolare un editor XML avanzato che supporti lo standard TEI e un mezzo divisualizzazione di tutto il testo della trascrizione. L’alternativa è usare unambiente integrato come EPPT ed effettuare la conversione in TEI P5 in unsecondo tempo.

Un approccio completamente diverso al problema è quello prospettato daH.A. Cayless:17 le immagini del manoscritto vengono convertite nel formatovettoriale SVG, quindi vengono elaborate in modo da identificare le righe ditesto vergate dallo scriba; in un passo successivo ogni riga viene evidenziata permezzo di un rettangolo, registrando automaticamente le coordinate dell’areacorrispondente; tali coordinate, infine, vengono utilizzate per definire le areedell’immagine originale da evidenziare in corrispondenza con la trascrizione informato TEI P5. La procedura qui riassunta è in realtà piuttosto complessa ecomprende l’uso di più strumenti software combinati in modo da produrre incascata i dati necessari per il passo successivo. Si tratta di un metodo ancora spe-rimentale, solo parzialmente automatizzato e probabilmente inadatto a mano-scritti con una mise en page complessa; tuttavia la prospettiva di sollevare il filolo-go dall’obbligo di tracciare manualmente le aree del manoscritto da collegare allatrascrizione risulta assai promettente e degna di essere seguita dalla comunitàaccademica con interesse.

4. Conclusioni

I nuovi elementi introdotti negli schemi di codifica TEI P5 offrono una soli-da base per realizzare un collegamento fra testo e immagine sia di tipo puntuale(hot-spot), sia generalizzato e continuativo. La relativa complessità, e soprattuttola notevole quantità di dati che tale operazione richiede, rendono tuttavia

151


17 H.A. CAYLESS, Linking Page Images to Transcriptions with SVG, <http://www.balisage.net/Proceedings/html/2008/Cayless01/Balisage2008-Cayless01.html>, intervento in occasionedel convegno Balisage: The Markup Conference, Montréal, 12 - 15 August 2008.


alquanto oneroso, in termini di tempo richiesto, sfruttare tale caratteristica. Percreare edizioni digitali basate su immagini, pertanto, è indispensabile ricorrere astrumenti software appropriati, come l’Image Markup Tool. Quest’ultimo è unottimo strumento per l’annotazione delle immagini, ma ancora inadatto a essereimpiegato per manoscritti di dimensioni medio-grandi, o per edizioni che inclu-dano immagini di più manoscritti, perché non permette di associare le immagi-ni annotate al testo separato della trascrizione; inoltre ogni immagine deve esse-re annotata separatamente. Potendo contare sulle fondamenta create dal consor-zio TEI, tuttavia, è soltanto questione di tempo prima che la comunità di ricer-catori che gravita intorno a tale standard provveda a creare e offrire alla più vastacomunità di utenti TEI gli strumenti necessari a questo scopo.

152



Medieval texts - Contemporary media. The art and science of editing in the digital age

Documents

Transcript of Medieval texts - Contemporary media. The art and science of editing in the digital age