Trans Internet-Zeitschrift für Kulturwissenschaften 10. Nr. Jänner 2002  

Encoding Wittgenstein

Some remarks on Wittgenstein's Nachlass, the Bergen Electronic Edition, and future electronic publishing and networking

Alois Pichler (Bergen)


  Isn't it interesting that the technological development which has now made it possible to publish complex handwritten material both on high scholarly standards and at the same time in easily accessible form, is also likely to prevent such complex corpora from coming into existence? For nowadays' writing desks' deletions, cancellations, rearrangements etc. disappear into cyberspace and only rarely survive on paper for the next generation's editors.


I. Wittgenstein's Nachlass, its structures and the question of how to edit it
II. The Wittgenstein Archives, its machine-readable version of the Nachlass and the Bergen Electronic Edition
III. Possible implications for text theory and practice from working with Wittgenstein's Nachlass in electronic form

I. Wittgenstein's Nachlass, its structures and the question of how to edit it

On his death, Ludwig Wittgenstein (1889-1951) left behind a philosophical Nachlass of some 20,000 pages. Apart from the Tractatus typescripts, these papers were at that point unpublished and largely unknown. The extent of the material was a surprise, even to his friends. G.E.M. Anscombe (1919-2001), Rush Rhees (1905-1989), and Georg Henrik von Wright (b. 1916), to whom Wittgenstein had entrusted the management of his literary estate and to "publish as many of my unpublished writings as they think fit" (Wittgenstein's testament, January 29th 1951), repeatedly had to redefine their task as more and more material came to light. The three literary executors had to collect the original manuscripts and typescripts from several places; the Nachlass is now - with a few exceptions - preserved at the Trinity College Library at Cambridge, the Austrian National Library in Vienna and the Bodleian Library at Oxford. Two years after Wittgenstein's death G.E.M. Anscombe's and Rush Rhees' publication of Philosophical Investigations/Philosophische Untersuchungen (1953) became the first book edition to be produced from the Nachlass. This edition has become the most important publication that has grown out from the Nachlass. During the 50 years following Wittgenstein's death more than twenty substantial books and articles have been published from the Nachlass, mainly edited by the literary executors (see Pichler 2000). By making the Nachlass progressively more available for the scholarly community, they played an essential role in it becoming apparent that Wittgenstein was to be considered as one of the most significant philosophers of the 20th century.

When referring to Wittgenstein's Nachlass we mean the texts listed in Georg Henrik von Wright's Nachlass catalogue (von Wright 1969, latest edition, in von Wright 1993). There, von Wright divides the Nachlass manuscripts and typescripts into three numbered groups, groups 101-182, 201-245 and 301-311. The 100-numbers refer to manuscripts, written in Wittgenstein's hand and consisting primarily of notebooks and bound volumes; the 200-numbers refer to typescripts, dictated by Wittgenstein directly from his manuscripts or prepared at an office for typewriting from the manuscripts or from other typescripts; the 300-numbers refer to dictations written either by hand or typed, that had been dictated to friends, colleagues and students in connection with e.g. lectures and seminars. Thus, the typescripts which are the Nachlass sources for the posthumous publication of Philosophical Investigations/Philosophische Untersuchungen, for example, have been assigned the numbers 227 (corresponding to Philosophical Investigations/Philosophische Untersuchungen, Part I) and 234 (corresponding to Philosophical Investigations/Philosophische Untersuchungen, Part II). For each Nachlass item von Wright's catalogue gives not only a number identifier, but also a date, a brief description, and the number of pages which the item spans. Several of the items are discussed in greater detail in a supplement to the main catalogue. Being such a thoroughly prepared basis for further Nachlass research, von Wright's catalogue has not only established itself as the standard system of reference but also as a primary source for information on the Nachlass in general. Some items, listed in von Wright's catalogue or remembered by others, are considered lost. On the other hand, the public recently has learned about the existence of items which previously had been unknown (see e.g. Koder 1993).

In assigning numbers to the Nachlass items von Wright took into account a range of different criteria, such as their physical characteristics, their chronological order, or Wittgenstein's own grouping of certain items into various series ("Bände"). One has to remember, above all, that von Wright's numbers refer to the separate physical parts of the Nachlass, not to thematically ordered units of text. In the case of the manuscripts this generally means that the hand-written material is classified according to the separate manuscript books (notebooks, ledgers, etc.), irrespective of how the text is arranged. Therefore, the relationship between the catalogue numbers and the Nachlass parts can sometimes be surprising, as in the case of the first two "Bände", MSS 105 and 106. Roughly speaking, here the text runs first through the recto pages of MS 105, continuing through the recto pages of MS 106, followed by the verso pages of MS 106, and finally on to the verso pages of MS 105. Thus, although there is a physical unit which Wittgenstein called "Band I", and which is referred to by a single catalogue number ("105"), the Nachlass item 105 consists, in fact, of two independent text sequences, the one running through the recto, the other through the verso pages. (Such text discontinuities are mostly a consequence of Wittgenstein working on different manuscripts when in different places, e.g. Vienna and Cambridge.) Another aspect which strikes one when working with the catalogue numbers, is that - not surprisingly for Nachlass material - handwritten and typewritten materials are quite generally mixed, both structurally and physically. Some manuscripts contain typescript cuttings, while the majority of typescripts contain handwritten additions.

Wittgenstein's writings, not unexpectedly, contain deletions, over-writings, insertions, marginal remarks and annotations, substitutions, re-orderings, as well as orthographic errors and slips of the pen, and the like. But while these features are quite typical of handwritten material in general there are certain features of Wittgenstein's working method which make his Nachlass rather special. One such feature is the habit of constantly working with variants of syntactically and semantically alternative expressions. Even when revising his work in such cases Wittgenstein often did not clearly decide in favour of a specific variant, leaving many possible readings open. Naturally, this creates a problem for its editor. Another special feature of Wittgenstein's working method is the habit of continuously taking over passages from one text corpus to another and developing and modifying them further and further. As a result the Nachlass is highly "intertextual" and, in a certain sense, repetitious. One could even say that the Wittgenstein Nachlass consists of nothing but a great number of different versions and modifications of a relatively small set of primary units of text or of thought (cf. Huitfeldt 1994:p.38, Pichler 1994:p.18). But appropriately to acknowledge this fact leads one to recognize that the strength of Wittgenstein's philosophy is intimately connected with a certain redundancy of thought and work. For it was only such a density of themes which could have allowed him to do the kind of concrete, precise and thorough, and at the same time deep thinking, which he demanded from himself in philosophy. Wittgenstein used different languages and different functions of language, linear, non-linear and "hyper-textual", argumentative and non-argumentative ways of developing a thought, thereby applying both linguistic and non-linguistic means, such as mathematical and logical notation, diagrams and graphics. He devised, developed and amended these aids throughout the entire period of his 40 years of adult life, over and over again trying them out on the same units of content. Through his methodological repetitiveness Wittgenstein could create a huge repertoire of research methods. To this his influence not only on 20th century philosophy, but on disciplines quite generally, is perhaps largely due.

Against the background of these few remarks on the complexity and richness of Wittgenstein's Nachlass and his thinking and working in general, it is understandable that the Nachlass - in being both mirror and carrier of them - presents several problems and challenges to one who approaches it with editorial intention. How shall an editor deal with a text which demands several different readings? How in different book publications shall one edit similar Nachlass texts without making the publication products too repetitive? We see that the features of textual repetitiveness and textual openness create a challenge. It is not astonishing, then, that the posthumous book publications show considerably different editorial approaches. Some of them involve much editorial intervention, others less. Only a few show what we could call the characteristics of a critical edition, or a firm, comprehensive policy of dealing with the Nachlass. Wittgenstein's editors have been criticised for putting together selections from a range of different manuscripts. But what gives possible ground for criticism is not so much that this is what they have done, but that they sometimes have done it without indicating the sources in detail.

One must acknowledge, however, that most of the "mistakes" of Wittgenstein's editors seem natural and understandable when viewed with due regard for the specific conditions under which Wittgenstein's Nachlass was to be made available. Among these conditions were the expectations of early Wittgenstein scholarship. Scholars would, in the first ten to twenty years after Wittgenstein's death have had a clear preference for "reader"-versions, that would make the Nachlass easily and quickly accessible, over and above more long-term projects that would yield more elaborate editions. A quick solution to the difficult question of balancing publication interests and scholarly standards appears to be that of publishing Wittgenstein's Nachlass in facsimile. This in fact was done, in 1967, when on the initiative of the literary executors, Cornell University made a microfilm of the entire Nachlass, such as it was known at the time, and then published it as The Wittgenstein Papers (Cornell University Libraries, Ithaca, N.Y., 1968). Many research institutes bought copies of pages produced from this microfilm or copies of the microfilm itself. But the Cornell edition is incomplete, and its technical quality and its dependability are uneven. Even without these limitations, though, it is clear that a facsimile as such is bound to be inadequate for certain purposes. For someone unfamiliar with Wittgenstein's handwriting, a facsimile is of limited use, and, of course, a facsimile is not machine-searchable.

By the early '80s, then, neither the existing book editions which contained only selections from the Nachlass, important though they are, nor the Cornell facsimile seemed to satisfy the requirements of Wittgenstein research. A Gesamtausgabe was not available, nor had sufficient agreement been reached over what it should be like. Without either distorting or hiding essential aspects of the Nachlass, or demanding from the publisher and from the reader too much in the way of finances, patience and willingness to adapt to some very special editorial policies, it seemed an intractable thing indeed, to put the entire Wittgenstein Nachlass into book form.


II. The Wittgenstein Archives, its machine-readable version of the Nachlass and the Bergen Electronic Edition

The issue of a Gesamtausgabe in book form is still a live issue and by no means settled. But now that an electronic edition of Wittgenstein's Nachlass, which fulfils most of the expectations that were to be fulfilled by a traditional Gesamtausgabe, is available, this situation is less troubling. The electronic edition to which I point here has been produced in co-operation with Oxford University Press at the Wittgenstein Archives at the University of Bergen (WAB), and has been published under the title Wittgenstein's Nachlass: The Bergen Electronic Edition (Oxford University Press 2000).

As far back as in the early '80s the Norwegian philosophical community considered it wise to opt for "marked-up" or "encoded" machine-readable transcriptions first, and rather to postpone working on a book Gesamtausgabe. Indeed, it was seen that by doing things in this sequence the postponed work could be better done. A machine-readable version of the entire Nachlass would not only grant full access to it and make the whole corpus highly manageable, it would actually also form an optimal basis for a later Gesamtausgabe in book form. For, firstly, it would provide the text basis such a Gesamtausgabe requires; secondly, it would encourage and facilitate scholarly discussion about what such a Gesamtausgabe could be like and should be like. Without the issue of the Gesamtausgabe being settled, Wittgenstein research could still very fruitfully be carried out with the help of the machine-readable version, and at the same time such research would yield impulses for any future desired settling of the issue. If the issue remained to be settled - but I have been getting ahead of myself.

At this point, two comforting facts must be held firmly in view. They are: 1. Through its work between 1990 and 1999, WAB has produced a machine-readable version of the Wittgenstein Nachlass. 2. Between 1998 and 2000, from this machine-readable version a publicly available electronic edition has been produced - the Bergen Electronic Edition. The machine-readable version consists of a set of high resolution Photo CD facsimiles, together with a set of source transcriptions of the Nachlass, and the software for processing both source transcriptions and facsimiles (examples of source transcriptions can be accessed from <> ). The system of markup of the source transcriptions is equally satisfactory for both philological and technical use. Further, this system is highly amendable and adaptable. From its very inception in 1990 WAB could welcome researchers from all over the world and offer them access to the machine-readable version still under development. These visiting researchers could pursue their studies of Wittgenstein or text encoding or applied information technology with the resource still growing. As these researchers were benefiting from the unique research tool which the machine-readable version was, WAB and what it was offering, too, were benefiting from such use. From the machine-readable version an electronic edition was created - the Bergen Electronic Edition (BEE), of which the first volume was published in 1998. Since BEE is available for purchase world-wide, the Wittgenstein research community is now for the first time offered complete and efficient access to the Wittgenstein Nachlass (save two minor items, dictations 307 and 308, which are not yet included in BEE). Whereas the machine-readable version can hardly be called a Gesamtausgabe, one might want to call BEE an "electronic" Gesamtausgabe. However, such a labelling is misleading, since it invites the neglect of important differences between a traditional Gesamtausgabe and modern electronic editions such as BEE. But what is it that makes BEE and its "mother", the machine-readable version, special, and special not only in comparison with a Gesamtausgabe or book editions in general, but in comparison with other electronic editions, too? Let me expand a little.

The markup of the machine-readable version's source transcriptions deserves special attention, both with regard to its format and the level of detail. Marked-up ("encoded") machine-readable texts have a flexibility which paper editions cannot match. Thus, thanks to the markup, a suitably marked-up electronic text can be filtered for presentation and further use in many different ways. The Standard Generalized Markup Language (SGML), a markup language, which had become an international standard for markup by 1986, imposed certain restrictions which, for the transcription work of the machine-readable version of the Wittgenstein Nachlass, were undesirable. One of the limitations of SGML at that time was the lack of software which could support the kind of editorial work which was required in WAB's project. Another limitation was SGML's strictly hierarchic structure which did not allow a satisfactory handling of overlapping features such as those that occur frequently in handwritten manuscripts and extensively in Wittgenstein's Nachlass. Therefore Claus Huitfeldt, the initiator of WAB, developed a new code system that was more suitable for the transcription of the Wittgenstein Nachlass. This code system was called MECS (Multi Element Code System). MECS was then combined with a registration standard which was developed at WAB specifically for the purpose of the meticulous transcription of the Nachlass. This registration standard was called MECS-WIT. MECS contains some of the features of SGML but it also contains additional, simpler procedures for representing certain structures not easily manageable in SGML. All SGML marked-up documents are formally MECS conforming, but not vice versa. Conversion of MECS marked-up documents to SGML-formed documents is to a large extent unproblematic, except in dealing with those features which MECS can handle and SGML cannot. MECS was designed to ensure that MECS marked-up documents can be easily formatted for output to standard applications. Also, the system of MECS was set up as a system which allowed for innovation and improvements as users require. Therefore, filter sets were made to allow for the presentation of the MECS marked-up source transcriptions in standard formats, such as plain ASCII format, Word Perfect and HTML. 1997-1998 a major effort was put into defining profiles for exporting the MECS source transcriptions to the Folio Flat File format since Folio VIEWS (version 3.1) had been chosen as the interface for the Bergen Electronic Edition.

Almost all the publications from the Nachlass have been available in electronic form for almost a decade now, published by InteLex Corporation (Charlotteswille) in the Past Masters series (Past Masters. The Collected Works of Ludwig Wittgenstein. Clayton, G.A.: InteLex Corporation 1993. See <>). However, in being planned from the beginning as an electronic product, BEE is different from this edition. In the first place it is different with respect to comprehensiveness, and then also with regard to structure and functionality. This should by no means be taken to be a slight on the value of the InteLex edition. "The Collected Works of Ludwig Wittgenstein" definitely deserves its place alongside BEE, but it needs to be said that BEE marks a new departure. In BEE the Nachlass texts are presented in two versions: the "diplomatic" version and the "normalized" version. The diplomatic version functions as a detailed letter-by-letter representation of the original, which, thanks to apt source transcription markup, is capable of representing deletions, substitutions, insertions or any other writer-action by Wittgenstein. The normalized version on the other hand provides a "reading" version of the texts. Spelling is, again with the help of source transcription markup, normalised, and certain features, such as deleted text, are suppressed. Folio VIEWS being a fully index-based retrieval system which organizes the data as infobases, offers powerful search facilities from which the user of BEE benefits greatly. The entire Nachlass corpus can be searched within seconds, and search results can be displayed in keyword-in-context lists as well as in full context. Since BEE includes an infobase for the diplomatic and the normalized version respectively, each of the infobases is fully searchable in its own right. In addition, special fields have been established, e.g. for searches within specific volumes and Nachlass groups, for variant text readings, for searches within date ranges, for specific languages, Wittgenstein's Geheimschrift passages, graphic material, or for searches within mathematical notation. In addition to presenting the Nachlass in the diplomatic and normalized versions BEE includes digital facsimile images of all the Nachlass pages. The normalized and the diplomatic versions are inter-linked page by page, but they are also inter-linked with the facsimile images page by page. This provides for the possibility of quick and efficient moving between the diplomatic and normalized levels and the facsimile level. InfoVIEW (version Lite 2.2) is used to display the electronic facsimiles and includes the usual functions of image browsers, such as zoom, print and sequential presentation. InfoVIEW also offers a number of image-enhancement functions.

Encoding Wittgenstein's Nachlass and preparing it for electronic publication is such an immense challenge that prudent division of labour and careful and systematic international co-operation are called for. The machine-readable version and, naturally also BEE, would have been impossible without close co-operation with Wittgenstein's literary executors and the libraries and archives where the items of Wittgenstein's Nachlass are held. The Norwegian Computing Centre for the Humanities developed BEE's internal linking mechanisms and rendered the electronic facsimiles - delivered by Oxford University Press - into appropriate format for BEE. External expertise was sought on the level of detailed text work: Michael R. Biggs (University of Hertfordshire) advised on and helped with editing the graphical parts of the Nachlass; Laurence Goldstein (University of Glasgow) and William Boos (University of Iowa) were indispensable co-editors of the logical and mathematical parts of the Nachlass.

WAB's future perspectives include the possible linking of the machine-readable version and BEE with "The Collected Works of Ludwig Wittgenstein" in the InteLex' Past Masters series, and perhaps also with the electronic edition of Wittgenstein's correspondence, currently being prepared at the Brenner Archives in Innsbruck. One might even consider a larger project, aiming at a combined Works, Nachlass and Correspondence edition, amended with commentary and rich bibliographical and biographical material. With alertness to the technical possibilities of XML (Extended Markup Language), work is being done at WAB and the Humanities Information Technologies Research Programme (HIT Centre) on developing software which provides for easy, secure and "intelligent" conversion from MECS to XML, yielding also the required DTDs (Document Type Definitions) and XSLs (Stylesheets). Straightforward conversion from MECS to XML is possible already now, but it works "unintelligently" in the handling of features which are peculiar to WAB's transcriptions, in particular the overlapping scope of codes and the use of codes with multiple elements. Equipping the user with XML transcriptions and predefined DTDs and Stylesheets, will provide him/her with unprecedented opportunities in the use of the text. But the user will by no means be restricted to the use of the DTDs and Stylesheets offered by WAB, since (s)he will be able to use them as templates which can be customised according to specific needs.

The impact which electronic publishing and networking will have on Wittgenstein research cannot be underestimated. Electronic linking and commentary work will contribute to improving the accessibility and understanding of Wittgenstein's texts in ways which are unprecedented. Moreover, it is to be expected that increasing spread of Wittgenstein research through electronic channels will yield not only changes in the use and understanding of Wittgenstein's work, but probably will also be part of that development which seems to go hand in hand with the "electronic turn" quite generally. The development which I am concerned with here is a certain growing tendency, of constructivism, in the way texts are responded to. This development is best understood and studied with specific examples. I turn for my examples again to WAB's work. For the scholar not involved in text encoding and electronic editorial philology, it is easy to overlook the fact that the relationship of BEE to WAB's machine-readable version is anything but simple. This relationship is not that of an emanatio, but rather that of a creatio libera - but of course not ex nihilo - and therefore in an important sense contingent. While BEE is produced from WAB's machine-readable transcriptions with the application of filters specially designed for BEE, these filters can in principal be defined in an endless number of ways. Thus, it would not be difficult to produce a series of quite different texts from the same source transcriptions. It is in the nature of traditional editions in general, as well as in the nature of BEE - although here to a much smaller degree - to be restricted in both format and content, whereas it is in the nature of modern marked up machine-readable versions that they allow for a series of formally equally valid, but at the same time quite different conversions to text. Naturally, BEE is not as comprehensive, flexible and open-structured as the basis from which it was produced - the "mother" machine-readable version at WAB. However, there are scholars who wish that BEE was so. Such scholars would like to be able to take a step back and to work ad fontes. In practice, what they want is a BEE so close to WAB's machine-readable version that one hardly could call it an edition any longer, but rather a device to create one's own edition, in accord with one's own research interests. There is no reason for not supporting such desires. Quite the contrary. Such approaches are not only fair, but they are in the process of making a new promising kind of Humanities, as will become clearer from Section III. It seems only advisable then to welcome them with an open mind, as long as this does not entail condoning certain extremes of openness which some critics associate with the postmodern world. To be aware of such general freedoms and to care for them only enhances the interest of our work, and the more this work is done, the more the awareness and care will grow.


III. Possible implications for text theory and practice from working with Wittgenstein's Nachlass in electronic form

In this section I am not going to discuss any possible impact that BEE might have on our understanding of Wittgenstein's philosophy. Rather, here I will turn to some general questions regarding the way texts work and how we regard texts, always of course with WAB material as my touchstone.

A transcriber at WAB once stated: "Machine-readable texts make it more clear to us what texts are and what text editing means: Texts are not objectively existing entities which just need to be discovered and presented, but entities which have to be constructed. They are products of both the author and the reader. All that exists in the case of Wittgenstein's Nachlass are scripts which first of all need to be identified, interpreted and organized. Having a machine-readable version of Wittgenstein's Nachlass provides a multiplicity of ways to organize and construct texts, it makes this easy - and it makes it obvious that there is an element of construction." (Pichler 1995:p.774f) It is the case that in many respects theories about texts bear marks of the tension between epistemologically realist and epistemologically idealist or constructivist approaches. It is no accident that such tensions are also visible in current reflections about text encoding. Consider for example what the American philosopher and text encoding specialist Allan Renear has to say about the "back stage ideologies" of text encoding. According to him, in text encoding there has been a development from realism to antirealism:

"Phase one of the textual ontology was a form of Platonism... The straightforward ontological question posed by DeRose et al. (1990) "What is Text, Really?'' was given a straightforward ontological answer: "text is a hierarchy of content objects,'' or, in a slogan and an acronym, text is an "OHCO.'' ... Phase two was pluralistic. When researchers from the literary and linguistic communities began using SGML in their work, the implicit assumption that every document could be represented as a single logical hierarchical structure quickly created practical problems for text encoding (Barnard et al. 1988). ... researchers from TEI found that there seemed to be many "logical'' hierarchies that had equal claim to be constitutive of the text. Thus where the original OHCO Platonists and the designers of SGML took the editorial hierarchy of genre to be primary, the literary scholars of the TEI took the structures elicited by specialized disciplines and methodological practices to be equally important. ... Phase three was Antirealistic. Modifying OHCO Platonism to Pluralism introduces the role that disciplinary methodologies and analytic practices play in text encoding. Some text encoding theorists see text encoding not as merely identifying the features of a text but as playing a more constitutive role. ... Pluralistic Realism allowed that there are many perspectives on a text, but assumes that texts have the structures they have independently of our interests, our theories, and our beliefs about them. The Antirealist trend in text encoding rejects this view seeing texts as the product of the theories and analytical tools we deploy when we transcribe, edit, analyze, or encode them." (Renear in Biggs/Huitfeldt 1997:pp.350ff, edited version of an electronic discussion)

The "development" that Renear talks about might be regarded as no more than an effect of the higher profile of constructivist scholars who just happen to be involved in important projects. Hence, it is conceivable that for a long time these positions have existed side by side in different camps, and that the "development" is nothing but a result of a change in the "balance of power" among the different camps. But while it is my view that electronic publishing and electronic networking have strengthened and will continue to strengthen the constructivist (antirealist) approach, it is also my view that Renears' analysis points not to a necessity, but yet still to a truth, namely that right from the beginning the work would have benefited from a deliberate constructivist attitude.

In dealing with texts we are tempted towards realism. This is because our text training predominantly involves printed books, which in their uniformity offer the practised reader very little resistance and seem to allow for a smooth and relatively unproblematic "reading off". Wittgenstein rightly talks of the "alles gleichmachende Gewalt der Sprache, die sich am krassesten im Wörterbuch zeigt, und die es möglich macht, dass die Zeit personifiziert werden konnte; was nicht weniger merkwürdig ist, als es wäre, wenn wir Gottheiten der logischen Konstanten hätten" (TS 213:p.522). Having approached printed texts in the realist spirit, we are easily seduced into looking at all texts and also complex Nachlass texts in the same way, and to take their content and structures as made of pre-given and flatly relevant entities which need simply to be registered. If this is true for reading, then it is more conspicuously true for transcribing and editing. But the habit underlying the spirit which is being described here leads into aporias. Indeed, from my own experience, the realist approach does make aporias in transcription and editing work much more likely than the constructivist approach. For me, a striking example of this has always been a "realist" interpretation of WAB's transcription guideline, "to represent the original as accurately as possible". Following this rule in the realist spirit, when e.g. transcribing text where one and the same letter occurs in different character-shapes, as with an occurrence of "r"s in both convex and concave shapes, one faces the fact, that it is possible to record the difference between these "r"s very accurately. But is it also desirable or necessary so to do? Clearly, with the tasks that we usually have, the project of recording such graphical differences is not on the list. On the other hand, if the tasks included strong graphological interests, then it would definitely have to be on such a list. It is this if-clause which gives a clue to our general problem of dealing with the temptation of realism. For, it is indeed the answer to the question about the interests which we have in the use of a text (both those which we are aware of and those which we are not) that decides on what is to be on our list. A discussion about the rights and wrongs of an edition is bound to remain superficial as long as it fails to address the rights and wrongs of the interests which that edition reflects. When trying to secure him/herself in a "realist" "true and accurate representation" alone, the transcriber or editor has indeed no good reason why (s)he should pay less attention to "r"s than to words as a whole. Thus, it is often the case that an editor's realist attempt to give an exact representation of textual elements and relations prompts an extremely complicated methodology thereby encumbering the edition with an unnecessarily complex critical apparatus and a painstakingly but nevertheless pointlessly designed system of diacritical signs (cf. Wittgenstein's idea that a tiny misunderstanding of language can prompt an entire philosophical system).

To a reader who might think that the constructivist position espoused here opens for text work which is uncontrolled and for that reason may easily become incoherent and inconsistent, I would like to reply in the following way. There is no such danger; indeed, the contrary is the case. For when carefully encoding and editing with well considered interests kept in mind, one will try to serve these interests in articulate and discerning ways. It is only by doing this, that individual research interests can be followed up in any subsequent extracting of specific encoding, without thereby interfering with other research interests. Hence, if one has an interest both in the text genesis (and therefore cares about text insertions) and in syntactic and semantic relations (and therefore cares about text variants within one and the same remark), one will have to encode variants which happen to arise from insertions, both as text variants and as text insertions. Subsequently, one is then able e.g. to extract syntactic and semantic variants, whether or not these variants are also of genetic interest and have been accordingly so encoded. In contrast to this, if insertions which also happen to constitute text variants, are encoded as text variants only, a thorough and efficient study of the text genesis is made impossible or at least very difficult. We see now that the distinction of diplomatic and normalized versions in BEE answers exactly to just such an explicitly articulated framework of interests. In this way two important "families" of interests - the one pertaining to the graphical and the other to the syntactical and semantic - can be handled separately. If one attempted to handle the members of both families simultaneously it would make for nothing but confusion. The facility of being able to discipline the work of encoding and editing by obeying the needs of a carefully considered framework of interests not only enhances manageability of the text to be encoded and edited, it also offers the user transparency of its operations. All this goes to show that being aware of, and acknowledging, the constructive element inherent in editorial practice as such, helps to make this practice explicit, transparent and controllable for others, and makes one the more willing and also the more able to discuss openly one's research interests and their consequences.

Although BEE is a big step ahead towards transparency, giving, as it does, the user the possibility of checking the editor's operations, BEE still does not seem transparent enough for those scholars who have experienced the greater transparency of WAB's machine-readable version. Although BEE is no print edition, it clearly obeys some of the rules of classical print editions with their associated restraints of access. It is this which the Austrian philosopher Herbert Hrachovec sees as a serious limitation:

"The user is allowed to read and manipulate texts via FolioViews (...) She can copy selected texts, paste and print them - but neither can she touch the indexing mechanism nor modify any of the underlying data. There is a strict separation between interface and non-transparent, computational deep structure (...) But, it might be objected, where is the problem? Wittgenstein's writings are at one's disposal, all of them, and in an extremely comfortable fashion. True enough, judged by the standards of the printed book. Yet, as was discussed in section two, Wittgenstein's Nachlass transcends the limits of such standards and an electronic edition might be better suited to capture those peculiarities. It might be organized so as to mirror Wittgenstein's editorial techniques, starting with single remarks as elementary building blocks and putting them together in a variety of ways, following Wittgenstein's lead. His working process, not its result, could be taken as the guiding principle. (...) Yet, the Bergen edition does not offer any tools to actually rearrange its content or redesign its appearance on the primary level. For all its flexibility and ease of use the Bergen edition is still in the conceptual grip of classical printed editions. Does it have to mimic the necessities of print culture?" (Hrachovec 2000:, accessed 27.10.2001).

Does BEE have to imitate print culture? The answer, of course, is: No. Or: perhaps it does in 2000, but will not necessarily in 2020. Or, in one sense, it will always have to. Wouldn't it be anachronistic if BEE should seek to mimic printed editions? After all, publishing techniques have changed, and with this change editing has been and is being remodelled by the electronic medium. The medium which for more than a hundred years has been without rivals, the book edition, now has received a rival which makes us reconsider all the branches of book editing. Historical-critical book editing, which is particularly important to us, has had to face certain obvious restrictions attaching to the paper medium. Essentially it had to come up with one text to be the right text, with the text's relatives subsumed as pendants, and all the time cultivating a methodology which made all this work. But now, it would be nothing less than anachronistic to attach oneself to a culture of editing methods the basis for which is changed. And it would be even more of a mistake to adopt it without due reflection for historical-critical editing in the electronic medium. In Wittgensteinian terms, such a mistake would amount to a confusion of "language games".

The attitude towards historical-critical editing, expressed above, must not be misunderstood. What I want to say here is that we have to check very carefully where we stick to tradition and adhere to standards which help us follow up our interests in the use of a text, and where we adhere to standards which helped us once to do so, but now may be by-passed as potential hindrances. It is not out of the question that editing Wittgenstein's Nachlass will be crowned by a historical-critical book Gesamtausgabe. But it is equally important to acknowledge how proper it was that such a prospect should not have set the standards for the Norwegian project of producing a machine-readable version. If one wants canonical text as it has been traditionally provided by historical-critical editions, then such text may be prepared from the machine-readable version. After all, such text can be prepared more economically, more flexibly and with better control from source transcriptions in a primary data format than from a file destined for direct paper output, the format of which may be out of date when the file is finally ready for printout. I have already remarked that highly marked up transcriptions allow not only for a range of "translation" possibilities of the basis into different text outputs, but access to such transcriptions with accompanying software also allows the editorial processes of text construction to be questioned, checked and changed. In the case of historical-critical book editing such processes have been hidden from the user's view. In the case of machine-readable versions such processes are visible and revisable. The editorial decisions once to be taken only by the editor, and its consequences handed to the user, are no longer necessarily to be taken only by him. Such decisions now can be taken by the user, with the freedom to construct "diplomatic", "normalized", or any other versions according to desire, and to choose text portions as required, without being forced to set the needs of today as the standards for tomorrow. So, it appears that Hrachovec is right: To publish sophisticatedly encoded transcriptions, accompanied by software to process them (XML is a promising possibility here) and alongside facsimiles of the original for checking their accuracy, is the right way to go.

But there seems to be another aspect to notice, which prompts a positive answer to Hrachovec' key question. Yes, BEE does have to imitate book culture. Users of the Nachlass, of the machine-readable version, and even of BEE ask: "How may I properly quote from this?" They ask: "Among the several different ways of reading the text, how am I to choose between them?" These two questions alone show that users are likely to desire some help and need some. Paradigmatic examples of how to use the text will give them that help. One of the great values of BEE is that it is a most versatile paradigmatic example for showing e.g. how to make use of the machine-readable version and how to edit Wittgenstein's Nachlass. But BEE, undeniably, is more a device to access the Nachlass than a prescription of how to quote from it. That said, it does not leave its users unsupported. Both WAB's machine-readable version and BEE encourage taking advantage of an exact reference system accommodating Nachlass item, Nachlass page and Nachlass section. To be able to cite with exact reference to Nachlass item and Nachlass page is a minimum requirement which will permit users of the Nachlass to check the quotation and to investigate it further in the original context. Even more than in the normal practice of using quoted material, it is important here to document one's citation practice properly. We see then, that as things are the situation of working with the
Nachlass is still very fluid, with paradigmatic exemplars in the need and also in the process of being established. Electronic editions will therefore have to imitate book culture, to the extent that we connect paradigmatic examples in text work with book culture.

With the disappearing of a traditional field new fields are in the making. The historical-critical editor will no longer have to seize decision-making from the user. But (s)he will lay out the material for the user as comprehensively and consistently as possible. (S)he will also offer paradigmatic examples of how to work with a given material, and construct "the right texts" out of it. Thus, even if it becomes the case that in a decade or so the entire Wittgenstein corpus should be available on the World Wide Web, encoded in sophisticated ways and yet easily processable to fulfil the needs of different users, the expert and paradigmatic sampler will be needed not less, but more. We might even admit that "realist" invocations of "the real text" or "the correct text" will have a central function with this need of the future for establishing text paradigms in the face of the electronic masses approaching us. All this will have to take place in the framework, and on the premises, of the electronic medium. As new standards of quality and judgement develop with the developing electronic medium, scholars in the Humanities are the ones most likely to become the authors of works providing the required paradigmatic exemplars.



Section I of this article has its history. From 1990 to 1999 I took part in the project of producing a machine-readable version of Wittgenstein's Nachlass at the Wittgenstein Archives at the University of Bergen (WAB). In this connection I wrote an introduction to Wittgenstein's Nachlass (in Pichler 1994). This introduction depended, naturally, on work carried out by others and in particular on Georg Henrik von Wright's Wittgenstein (von Wright 1986), Claus Huitfeldt's and Viggo Rossvær's The Norwegian Wittgenstein Project Report (Huitfeldt/Rossvær 1989) and Claus Huitfeldt's "Toward a Machine-Readable Version of Wittgenstein's Nachlaß" (Huitfeldt 1994). In a joint endeavour the text was later used by Peter Cripps (also at WAB) and myself to write an introduction to the Nachlass for the Bergen Electronic Edition (BEE). Subsequently this introduction of ours was once more "shaped", this time by the Archives' Director, Claus Huitfeldt, and then published in 1998 as part of the Introduction to BEE (Wittgenstein's Nachlass: The Bergen Electronic Edition. Oxford: Oxford University Press 1998-2000. See <>). I would like to thank Oxford University Press for permission to make further use of this text in my article. I would like to thank Claus Huitfeldt and Peter Cripps for allowing me to publish here a renewed version of a text that has had a history of joint efforts and successful co-operation. - Many of the technical parts of section II make sustained use of terminology, concepts and methodology developed by Claus Huitfeldt in the late '80s and early '90s, and used by him in papers such as "MECS - A Multi-Element Code System" (Huitfeldt 1993a) and "Manuscript Encoding: Alphatexts and Betatexts" (Huitfeldt 1993b). Again to Claus Huitfeldt I express my gratitude for allowing me the freedom to use concepts and terminology which are the fruit of his work. - In regard to section III I am indebted to lively theoretical discussions with text encoding experts and text philosophers and to the utterers of certain seemingly harmless but nevertheless provocative statements, such as, "The aim of transcription is to represent the originals as accurately as possible". They made me think carefully about questions of text and understanding, and led me to strengthen and develop further my views in these questions.

To Claus Huitfeldt, Kevin Cahill and Michael Biggs I am grateful for discussions on drafts of this paper. I am deeply indebted to Ralph Jewell for help with fine-tuning it.


© Alois Pichler (Bergen)

TRANSINST       table of contents: No.10


Barnard, D. T., et al. (1988). "SGML-Based Markup for Literary Texts: Two Problems and Some Solutions". Computers and the Humanities 22 pp.265-276.

Biggs, Michael/Huitfeldt, Claus (ed.) (1997). "Philosophy and Electronic Publishing". In: The Monist. Interactive Issue 80 S.348-367 (La Salle, Illinois 1997). See an electronic version of the article at <> . Last access to this and all further mentioned web sources: 2002-01-03.

DeRose, St. J./Durand, D./Mylonas, E./Renear, A. (1990). "What is Text, Really?". Journal of Computing in Higher Education 1.2 pp.3-26.

Hrachovec, Herbert (2000). "Wittgenstein on line / on the line". Wittgenstein Archives Discussion Forum, <> .

Huitfeldt, Claus (1993a). MECS - A Multi-Element Code System. Downloadable (version 1998) from <> .

Huitfeldt, Claus (1993b). "Manuscript Encoding: Alphatexts and Betatexts". Downloadable from <> .

Huitfeldt, Claus (1994). "Toward a Machine-Readable Version of Wittgenstein's Nachlaß". In: Philosophische Editionen. Erwartungen an sie - Wirkungen an sie p.37-43. Beihefte zu editio 6. Ed. by Hans Gerhard Senger. Tübingen: Max Niemeyer Verlag.

Huitfeldt, Claus/Rossvær, Viggo (1989). The Norwegian Wittgenstein Project Report 1988. Bergen: The Norwegian Computing Centre for the Humanities.

Pichler, Alois (1994). Untersuchungen zu Wittgensteins Nachlaß. Bergen: Working Papers from the Wittgenstein Archives at the University of Bergen 8. See - No 8 <> .

Pichler, Alois (1995). "Advantages of a Machine-Readable Version of Wittgenstein's Nachlaß" In: Culture and Value. Beiträge des 18. Internationalen Wittgenstein Symposiums 13.-20. August 1995 Kirchberg am Wechsel S.770-776. Ed. by Kjell S. Johannessen and Tore Nordenstam. Die Österreichische Ludwig Wittgenstein Gesellschaft: Kirchberg am Wechsel.

Pichler, Alois (2000). Bibliographie der deutsch- und englischsprachigen Wittgensteinausgaben. Downloadable from <> .

Von Wright, Georg Henrik (1986). Wittgenstein. Translated by Joachim Schulte. Frankfurt am Main: Suhrkamp.

Von Wright, Georg Henrik (1986). "Wittgensteins Nachlaß". In: Wittgenstein pp.45-76. Translated by Joachim Schulte. Frankfurt am Main: Suhrkamp.

For quotation purposes:
Alois Pichler: Encoding Wittgenstein. Some remarks on Wittgenstein's Nachlass, the Bergen Electronic Edition, and future electronic publishing and networking. In: TRANS. Internet-Zeitschrift für Kulturwissenschaften. No. 10/2001ff. WWW:

TRANS     Webmeister: Peter R. Horn     last change: 24.01.2002     INST