a. The topographic transcriptions decipher the texts of the manuscripts and reconstruct their page layout. The result is merely an attempt to recreate the impression of the original document (e.g. by reconstructing the topography, the font, and the type of paper). The type font used for the transcription of typescripts is 'Courier'; holograph manuscripts or additions in Beckett's hand are in 'Arial Narrow'. The colours correspond with the colours of Beckett's writing tools.
In the modules of the BDMP where topographic transcriptions are not feasible due to the amount of manuscript pages (for instance in the case of long prose texts such as the novels of the trilogy), the purpose of bridging the gap between image (facsimile) and text (XML transcription) is served by a segmented pop-up transcription ('text/image'), drawing directly on the linear transcription in XML, but segmented (i.e. only presented in small segments) and linked directly to the relevant zone on the facsimile.
The TEI Special Interest Group on genetic editions suggests indicating the coordinates of the delineated zones in the <ge:document> (the document-oriented approach), i.e. separate from the text-oriented transcription. The BDMP does not have a <ge:document> portion in each transcription, but stores these coordinates in a separate file (coordinates.xml).
b. The linear transcriptions of the documents preserved at the holding libraries translate the signs on the manuscript into a textual format, with as little diacritical signs as possible.
Document Level: Writing Tools and Revision Campaigns
To mark revision campaigns, the TEI Special Interest Group working on genetic editions suggests the declaration of 'stage' in the XML header and the use of a stage attribute. However, these different stages cannot always be discerned unequivocally, especially if the author made corrections and additions in the same writing tool as the main text. In many cases, one of the few indications of a new writing campaign is a change of writing tools. For instance, when a manuscript in black ink features some additions in red crayon, the new writing tool very often indicates a new writing campaign. The BDMP therefore offers an alternative by highlighting the different writing tools to facilitate the study of revision campaigns.
A good example for such a study of revision campaigns is UoR MS 2933-5
. This version was written with a typewriter and contains corrections in both black and red ink. These writing tools are encoded in the XML markup and visualized in different fonts/colours.
Similarly, immediate alterations (currente calamo) have been encoded with the 'type' attribute 'instant correction' in the deletion tag (for details, see the brief technical documentation in the appendix).
In the default visualizationall deletions and additions are indicated. For
instance, the first line of the first document (UoR MS 2933/1/1
Tout tout le temps Toujours à la même distance...
If users do not wish to see these deletions and additions, a 'top layer' visualisation may facilitate the reading. In this 'top layer' visualization, deleted passages are not displayed and additions are not distinguished as additions, resulting in a reading text of the final version of the draft:
Tout Toujours à la même distance ...
As the example indicates, the transcription does not standardize what is deciphered. Beckett may have intended the addition 'Toujours' to be incorporated in the sentence starting with 'Tout' (implying that 'toujours' should start with a lower-case 't'), but if - as in this case - he wrote it with a capital 'T', the transcription endeavours to render the letters and punctuation as they are found in the manuscript.
In some cases the 'top layer' option may give a misleading
impression. For instance in UoR MS 2935/3/8
the protagonist's hands are said to be 'At
rest after all what
they did.' Beckett first wrote
'all', then deleted it, added 'what', deleted this addition again, and did not
substitute it with any alternative. Because of this lack of any substitution,
none will be visualized in the 'top layer' option.
A comparison with the next version indicates that Beckett (by means of the deleted addition without substitution) silently reverted to the original word 'all'. But in and of itself, the manuscript of UoR MS 2935/3/8
shows more hesitancy than resolution. In this case, the 'top layer' option will respect the unresolved nature of this moment of hesitation and the resulting lacuna in the sentence's syntax.
Dossier Level: versions and paralipomena
A document may contain several versions of a work (or of more than one work). Peter Shillingsburg has defined a version as 'one specific form of the work - the one the author intended at some particular moment in time' (Shillingsburg 1996
, 44), which may serve as a suitable working definition. According to Siegfried Scheibe's definition, textual versions are 'achieved or unachieved elaborations of the text that diverge from one another. They are related through textual identity and distinct through variation' (Scheibe 1995
, 207). Theoretically this implies that one single authorial revision is enough to create a new version. For pragmatic reasons, however, the definition of version as employed in the Beckett Digital Manuscript Project is broader than a writing stage, i.e. one version may contain several layers of revision or 'campagnes de révision
The main danger of a teleological perspective is the neglect of passages (especially in the early manuscripts) that did not make it into the base text (see also chapter 'Base texts
'). Nonetheless, there may be several versions of these passages, so that they should be comparable too. For instance, in the case of Stirrings Still, the first paragraph of the first few abandoned sections starts with the phrase 'Tout Toujours à la même distance', translated as 'All always at the same remove.' After a few versions, Beckett abandoned this path in the writing process, so that it remained a dead end.
In textual criticism 'paralipomena' are usually defined as holograph material that does not belong to a version, but is somehow linked to it, thematically or otherwise (Mathijsen 1997
, 44-47). Since its Greek etymology means 'what is left out', paralipomena have often been left out of scholarly editions. The BDMP includes them, since these loose jottings, ideas, false starts and potential alternatives are of special importance to the study of the composition process.
Collation and Relative Calibration
Scholarly editions in a printed format usually present textual variants between versions in an apparatus variorum. To save paper space the invariant passages are left out and the editor has to make use of a complex set of diacritical signs, which may be somewhat intimidating. Such an apparatus also requires a considerable effort from its readers to reconstruct the textual context of the variants. Without the restrictions of the codex format, the electronic apparatus can offer the possibility to indicate the variants in their context: the structure of the whole sentence always remains intact and variants between two versions are highlighted.
Traditionally the notion of variants applies to either variation between copies of ancient or medieval documents by scribes, or between different editions of a work. When dealing with modern texts, it may be useful to make a distinction between 'genetic' variants ('rewritings') and 'transmissional' variants. Moreover, the edition of bilingual works requires an extra category of 'translation variants.' On the Estate of Samuel Beckett's request the genetic edition of the BDMP focuses on the prepublication history and is limited to genetic and translation variants. A critical apparatus of transmissional variants can be presented as a separate electronic tool accompanying critically edited texts of Samuel Beckett's complete works as envisaged by Faber and Faber.
Genetic variants (rewritings)
Pierre-Marc de Biasi argues that in the case of modern manuscripts it is difficult to speak of 'variants'; instead he uses alternative expressions such as 'rewritings' or 'genetic operations', arguing that it is impossible to speak of 'variants' if there is no 'invariant' (de Biasi 2000
, 20). No matter how valuable this argument is, the notion of 'variants' can still be employed to conceptualize a genetic edition. From a pragmatic perspective it seems useful to employ the notion of 'variants' as an umbrella term. Even if one does not wish to employ the term 'variants' to designate 'rewritings', one can only visualize them by contrasting them with another version and highlighting those instances that 'vary'. Although in genetic criticism there is not an absolute invariant (i.e. not a fixed point against which the 'rewritings' can be measured), it is possible to apply a system of relative calibration. If there is no invariant to compare the variants with, it is always possible to compare a variant with another variant, on condition that the edition indicates which variant serves as a 'temporary invariant' (for instance the previous version in the chronology of composition).
Every literary genesis is characterized by a dialectic of copied and new elements, identity and variance (Ferrer 2002
, 48). To visualize this dialectic it seems important to indicate not just the isolated 'variants' (as in a traditional apparatus variorum) but to indicate or highlight these modifications in their syntactical context, i.e. leaving the structure of the whole sentence intact.
The 'top layer' option can be employed for digitally supported collation. The syntactical context of each segment remains intact, but to highlight the variants, they have to be encoded first. In view of the large amount of manuscript materials in several modules of the BDMP, the project would not be able to include the option of manually encoding an apparatus in all of the transcriptions.
As an alternative, the Centre for Manuscript Genetics at the University of Antwerp is working together with Ronald Dekker, Bram Buitendijk and Joris van Zundert at the Huygens ING institute in The Hague to test the possibilities of digitally supported collation by means of the CollateX algorithm (as part of the European COST Action IS0704 'Interedition: An Interoperable Supranational Infrastructure for Digital Editions').[9
One of the complicating elements of the test case Stirrings Still / Soubresauts is the rather large number of versions in combination with the presence of deletions and additions. The latter problem can be solved by working with the top layer of each draft version. We tried to apply this top layer system to test the first research results of CollateX. All top layer versions of a segment are prepared in the JSON syntax required by CollateX. This data is transferred through a REST service to the CollateX algorithm, which runs as a Java web archive within the architecture of the BDMP. The collation happens on the fly and the output takes the form of a table, aligning invariant passages and variant passages. The segmentation of the textual material not only reduces the danger for the user to get lost in the jungle of manuscripts, but also determines the speed of the almost instant collation.
Usually, digital 'archives' are distinguished from 'editions' because the latter offer a critical apparatus. An interoperable tool such as CollateX can enable any user - not necessarily an editor - to transform a digital archive into an electronic edition. As a consequence of these developments in scholarly editing, the strict boundary between digital archives and electronic editions is becoming increasingly permeable, resulting in a continuum rather than a dichotomy. At the moment of writing, the CollateX program and its embedding in the BDMP is still in development. Nonetheless, the first module of the BDMP already includes it as a tool that will be improved and updated in the following modules.
Non-equivalent instances between the English and French versions are called 'mismatches' by Magessa O'Reilly. A vertical bar 'indicates the position that would most likely have been occupied by the matching segment, were it present' (O'Reilly in Beckett 2001
, xiii). O'Reilly also acknowledges that these are obviously not the only instances of non-equivalence. He mentions shifts of verb tense, of singular and plural, of person or register, etc., and adds that 'it would be impossible to point out such an endless array of mismatchings' (xiii). In his edition of Comment c'est
O'Reilly points out the difficulty of establishing criteria to decide which variants should be marked or not. Minor instances of non-equivalence are not always signalled, in order not to overshoot the target.
In 'L'Évolution du sujet s'auto-traduisant: L'imaginaire de Beckett face à Malone meurt, Happy Days
et Stirrings Still
à traduire', Linda Collinge points out that in the case of Soubresauts 'la traduction est extrêmement proche de l'original' (2000
, 199). In the vast majority of cases, mismatches or translation variants are also genetic variants, i.e. they may be regarded as rewritings and may therefore benefit from a similar treatment as the one applied to genetic variants. This implies that the publication of Beckett's self-translation is not necessarily to be regarded as the end of the genesis, but as its continuation.
The BDMP functions as a research environment that is non-hierarchical in the sense that each text can be compared to any other text and that no text is singled out as being more important or 'definitive' than the other versions. In the underlying markup, this system of relative calibration is based on a numbering system keyed to a so-called 'base text'. The CSE Guidelines for Editors of Scholarly Editions define a 'base text' as a 'text chosen by an editor to compare with other texts of the same work in order to record textual variation among them': 'Unlike a copy-text, it is not assigned any presumptive authority and may not even be used to construct a critical text, serving instead only as an anchor or base to record textual variants.' (http://www.mla.org/cse_guidelines
According to the same guidelines, a copy-text is 'the specific arrangement of words and punctuation that an editor designates as the basis for the edited text and from which the editor departs only where deeming emendation necessary. Under Greg's rationale the copy-text also has a presumptive authority in its accidentals (...) Its selection is based on the editor's judgment that the authority of its accidentals is on the whole superior to other possible texts that could be chosen for copy-text.' To avoid any confusion because of the implicit link with W. W. Greg and Fredson Bowers' copy-text theory, which presupposes a very specific editorial approach, the term 'copy-text' is not employed in the present genetic edition. The base text is merely a tool in the underlying markup as a basis for the numbering of sentences.
The choice of the base texts in the BDMP is linked to the moment the author was on the verge of presenting his text to the public. In Pierre-Marc de Biasi's 'Functional Typology of Genetic Documentation' (1996), the 'bon à tirer' moment (pass for press) is presented as a dividing line, marking 'the decisive moment when what had been in a pliable and mobile manuscript state up to that point becomes fixed in the frozen shape of a published text' (Biasi 1996
, 37). The metaphor of the 'frozen shape' is somewhat problematic since the genesis of the text often continues even after the first publication (as for instance the editions of the theatrical notebooks and the revised texts illustrate), but the 'bon à tirer' does mark an important transition from the private atmosphere of the manuscripts to the public representation of the text. Since, on the Beckett Estate's request, the BDMP focuses on what precedes the published versions, the base text is the text of the last extant document preceding the 'bon à tirer' moment. Even though this project focuses on the creative process rather than on the product, the choice of a base text does imply the recognition that the notion of a finished product plays an important role in the genesis since every project implies a form of teleology.[10
] The BDMP can therefore function perfectly well in tandem with critically edited texts or with the reading texts as published by Editions de Minuit, Grove Press or Faber and Faber, together representing the dialectics of Beckett's works between completion and incompletion.
In the case of Stirrings Still
, the choice of the base text is marked by the delay between the writing of the first two sections and the third one. The base text for the first two sections is the text of document UoR MS 2859
; the base text for the third section, that of UoR MS 2935/5/3
. The first sentence of the second section in typescript UoR MS 2859
is marked by an open variant. The sentence starts as follow: 'As one in his right mind when at last out again he knew not...' The words 'he knew not' are underlined (not cancelled), and Beckett has added the alternative 'no knowing' above the line.[11
] So even at this late stage in the composition process Beckett kept revising his text. Since this hesitation is an integral part of the composition process, the base text does not represent the author's 'final intention'; it simply serves as an anchor to record composition variants and translation variants, and make the versions comparable.
For a bilingual comparison, this base text can be compared to the computer print-out of the translated version (UoR MS 3543
) on which Beckett has written 'Final'. This version differs in some instances from the text published by Les Éditions de Minuit (both in a limited edition of 99 + 10 copies and in a paperback edition). The nature of the changes suggests that Beckett was open to suggestions by his French publisher Jérôme Lindon (e.g. 'lorsqu'amortis' / 'lorsque amortis'; 'se clouer debout' / 'se figer debout'; 'cloué sur place' / 'figé sur place'). Even during earlier stages of the translation process (for instance on 9 May 1989, in reaction to the translation of the first part of Stirrings Still
) the publisher not only expressed his admiration for the new work and for the translation, but also felt free to point out a few minor errors, such as the absence of a double m in the word 'sufisamment' or the wrong gender of the word 'horloge'.
The base text of Comment dire
in the BDMP is the text of the typescript preserved at the Beckett Archive in Reading, dated '29.10.88' (UoR MS 3317
). The base text for the English version is UoR MS 3506
The advantage of encoding the transcriptions is the resulting flexibility of the textual material, based on Ted Nelson's principle of 'transclusion': the same piece of information can have meaning in a variety of contexts and in each of these contexts the shared data can be retrieved without duplicating them. The codex format could not include full transcriptions, otherwise the user would get lost in the abundance of textual material.
The flexibility of the transclusive electronic medium allows the edition to offer full transcriptions, retrievable either in their entirety or piecemeal, in different contexts:
1. The documents can be studied in the order of their catalogue numbers. The digital facsimiles of these documents are presented with different tools, such as a digital magnifying glass:
If the passage is hard to decipher, a transcription may be of help:
2. The chronology of the composition process can be visualized in the form of a genetic map. Dead ends in the genetic dossier (passages that did not make it into the published text) are indicated as 'abandoned sections'. The language is indicated by means of shades of grey: dark grey for English, light grey for French.
3. The versions of sections that made it into the published text can be arranged according to their position in the narrative structure and the textual variants between them can be studied with the option 'Compare versions'.
4. The rearrangement per language shows that Beckett often switched between French and English during the composition process. Translations (i.e. authorial translations) can be examined separately.
The genetic edition offers the possibility (based on a numbering system in the XML encoding) to choose the size of the textual unit one wishes to compare: large (a section); medium (a paragraph); small (a sentence or segment). For instance, Stirrings Still consists of three sections, the first of which is subdivided into seven paragraphs. The sections are encoded with a <div> tag, carrying the number of the section; the paragraphs with a <p> tag, numbered according to the following system: e.g. the third paragraph in the first section:
Not all of Beckett's texts can be subdivided in sections and paragraphs. The most useful unit is usually the smallest one: the sentence or segment, marked with a <seg> tag, is followed by a stable identifier consisting of the catalogue number of the relevant holding library and - between square brackets - the number of the sentence or segment in the 'base text':
Only segments that made it into the published text are numbered (for instance the sentences of the abandoned sections 'Before Stirrings Still
' are not numbered). Usually the size of the segment <seg>
is a sentence. In cases such as Not I / Pas moi
, where Beckett did not work with full sentences, the segment consists of a few lines of text, a unit of text that can easily be compared to other versions. By means of the segment numbers, any version of a segment that made it into the published text can be compared to any other version of that segment (for more details, see the Manual
The option 'Bilingual comparison' (under 'Language') facilitates the systematic comparison of the author's translation with the base text. The variants are highlighted by means of a colour code. The option 'Early translations' enables the comparative study of the translations Beckett started during the composition process.
Bilingual Dynamic Comparison
One special visualization is only applicable to the unique form of Beckett's last work. In Comment dire / what is the word the genesis of the text becomes its theme; the text (product) presents itself as a process of writing (production). At first sight, the text looks like a poem, but it can also be read as an attempt to write one single sentence. The option 'Dynamic Text' visualizes this attempt. Every 3 seconds it automatically shows the next step in the fictionalized composition of the sentence; or one can pass through the stages manually at one's own pace. The purpose of this option is to visualize the dynamics of this text between completion and incompletion.
'Your Comments' and 'Revision History'
The Beckett Digital Manuscript Project is updatable. To give readers and researchers the opportunity to contribute to the project, for instance by suggesting a correction or a transcription for a passage that could not be deciphered up to now, there is a permanent link ('Your comments', in the top right corner of the screen). This permanent link leads to a comments page, where readers can comment on any document, to suggest for instance an alternative reading of a hardly legible autograph passage. The editors receive this comment along with the automatically generated URL of the digital facsimile it relates to. The comment is presented to the editorial board and if necessary to the advisory board. If the editorial board decides the correction is a valuable contribution, the transcription in the XML document will be changed. In the header of the XML document, the change is recorded in a <change> tag: person A is responsible for the change of 'X' to 'Y' at a particular date. In the case of a modified transcription, a box entitled 'Revision history' will subsequently appear at the top of the linear transcription of the document. This box lists all the revision information in such a way that a state of the transcription at a particular point in time can always be reconstructed. Since all changes are documented, the transcription can be quoted without the risk of any passage being silently changed.