Towards a model for dynamic text editions

Creating digital editions so far is devoted for the most part
to visualisation of the text. The move from mental to machine
processing, as envisaged in the Semantic Web initiative, has not
yet become a priority for the editorial practice in a digital
environment. This drawback seems to reside in the almost
exclusive attention paid until now to markup at the expense of
textual data models. The move from “the database as edition”
[Thaller, 1991: 156-59] to the “edition as a database” [Buzzetti
et al., 1992] seems to survive only in a few examples. As a way
forward we might regard digital editions to care more about
processing textual information rather than just being satisfi ed
with its visualisation.
Here we shall concentrate on a recent case study [Rehbein,
forthcoming], trying to focus on the kind of logical relationship
that is established there between the markup and a database
managing contextual and procedural information about the
text. The relationship between the markup and a data model
for textual information seems to constitute the clue to the
representation of textual mobility. From an analysis of this
kind of relationship we shall tentatively try to elicit a dynamic
model to represent textual phenomena such as variation and
interpretation.
I.
The case study uses the digital edition of a manuscript containing
legal texts from the late medieval town Göttingen. The text
shows that this law was everything else but unchangeable.
With it, the city council reacted permanently on economical,
political or social changes, thus adopting the law to a changing
environment. The text is consequently characterised by its
many revisions made by the scribes either by changing existing
text or creating new versions of it. What has come to us is, thus,
a multi-layered text, refl ecting the evolutionary development
of the law.
In order to visualise and to process the text and its changes, not
only the textual expression but, what is more, its context has
to be regarded and described: when was the law changed, what
was the motivation for this and what were the consequences?
Answers to these questions are in fact required in order
to reconstruct the different layers of the text and thereby
the evolution of the law. Regarding the text nowadays, it is
however not always obvious how to date the alterations. It is
sometimes even not clear to reveal their chronological order. A simple example shall prove this assumption. Consider the
sentence which is taken from the Göttingen bylaws about
beer brewing
we ock vorschote 100 marck, de darf 3 warve bruwen
together with 150 as a replacement for 100 and 2 as a
replacement for 3. (The meaning of the sentence in Low
Middle German is: one, who pays 100 (150) marks as taxes, is
allowed to brew beer 3 (2) times a year.) Without additional
information, the four following readings are allowed, all
representing different stages of the textual development:
R1: we ock vorschote 100 marck, de darf 3 warve bruwen
R2: we ock vorschote 100 marck, de darf 2 warve bruwen
R3: we ock vorschote 150 marck, de darf 3 warve bruwen
R4: we ock vorschote 150 marck, de darf 2 warve bruwen
With some more information (mainly palaeographical) but still
limited knowledge, three facts become clear: fi rstly, that R1
is the oldest version of the text, secondly that R4 is its most
recent and thirdly that either R2 or R3 had existed as text
layers or none of them but not both. But what was, however,
the development of this sentence? Was it the path directly
from R1 to R4? Or do we have to consider R1 > R2 > R4 or
R1 > R3 > R4? In order to answer these questions we need
to know about the context of the text, something that can
not be found in the text itself. It is the external, procedural
and contextual knowledge that has to be linked to the textual
expression in order to fully analyse and edit the text.
Textual mobility in this example means that, to a certain extent,
the textual expression itself, its sequence of graphemes, can be
regarded as invariant and objective, the external knowledge
about its context cannot. It is essential in our case study not
only to distinguish between the expression and the context of
the text but what is more to allow fl exibility in the defi nition
and reading of (possible) text layers. It became soon clear,
that for both, visualising and processing a dynamic text, a new
understanding of an edition is needed, and, as a consequence,
the mark-up strategy has to be reconsidered. This new
understanding would “promote” the reader of an edition to
its user, by making him part of it in a way that his external
knowledge, his contextual setting would have infl uence on the
representation of the text. Or in other words: dynamic text
requires dynamic representation.
The way chosen in this study is to regard textual expression
and context (external knowledge) separately. The expression
is represented by mark-up, encoding the information about
the text itself. Regarding this stand-alone, the different units of
the text (its oldest version, its later alterations or annotations)
could indeed be visualised but not be brought into a meaningful
relationship to each other. The latter is realised by a database
providing structured external information about the text,
mainly what specifi c “role” a certain part of the text “plays”
in the context of interest. Only managing and processing both,
markup and database, will allow to reconstruct the different
stages of the text and consequently to represent the town law
in its evolutionary development.
Using the linkage mechanism between mark-up and database,
the whole set of information is processable. In order to create
a scholarly edition of the text, we can automatically produce
a document that fulfi ls TEI conformity to allow the use of the
widely available tools for transformation, further processing
and possibly interchange.
II.
The case study just examined shows that in order to render an
edition processable we have to relate the management system
of the relevant data model to the markup embedded in the text.
But we cannot provide a complete declarative model of the
mapping of syntactic markup structures onto semantic content
structures. The markup cannot contain a complete content
model, just as a content model cannot contain a complete
and totally defi nite expression of the text. To prove this fact
we have to show that a markup description is equivalent to
a second-order object language self-refl exive description and
recall that a second-order logical theory cannot be complete.
So the mapping cannot be complete, but for the same reason
it can be categorical; in other words, all the models of the text
could be isomorphic. So we can look for general laws, but they
can provide only a dynamic procedural model.
Let us briefl y outline the steps that lead to this result. In a
signifi cant contribution to the understanding of “the meaning of
the markup in a document,” [Sperberg-McQueen, Huitfeldt, and
Renear, 2000: 231] expound it as “being constituted,” and “not
only described,” by “the set of inferences about the document
which are licensed by the markup.” This view has inspired the
BECHAMEL Markup Semantics Project, a ground breaking
attempt to specify mechanisms “for bridging [...] syntactic
relationships [...] with the distinctive semantic relationships
that they represent” [Dubin and Birnbaum, 2004], and to
investigate in a systematic way the “mapping [of] syntactic
markup structures [on]to instances of objects, properties,
and relations” [Dubin, 2003] that could be processed trough
an appropriate data model. Following [Dubin and Birnbaum,
2004], “that markup can communicate the same meaning in
different ways using very different syntax”, we must conclude
that “there are many ways of expressing the same content,
just as there are many ways of assigning a content to the same
expression” [Buzzetti, forthcoming].
The relationship between expression and content is then
an open undetermined relationship that can be formalized
by taking into account the “performative mood” of markup
[Renear, 2001: 419]. For, a markup element, or any textual
mark for that matter, is ambivalent: it can be seen as part of the text, or as a metalinguistic description/ indication of a certain
textual feature. Linguistically, markup behaves as punctuation,
or as any other diacritical mark, i.e. as the expression of a
refl exive metalinguistic feature of the text. Formally, markup
behaves just as Spencer-Brown’s symbols do in his formal
calculus of indications [1969]: a symbol in that calculus can
express both an operation and its value [Varela, 1979: 110-
111].
Markup adds structure to the text, but it is ambivalent. It
can be seen as the result of a restructuring operation on
the expression of the text (as a textual variant) or as an
instruction to a restructuring operation on the content of the
text (as an interpretational variant). By way of its ambivalence
it can act as a conversion mechanism between textual and
interpretational variants [Buzzetti and McGann, 2006: 66]
[Buzzetti, forthcoming].
Markup originates a loop connecting the structure of the
text’s expression to the structure of the text’s content. An
implementation of the markup loop would considerably enhance
the functionality of text representation and processing in a
digital edition. To achieve implementation, markup information
could be integrated into the object (or datatype) ‘string’ on
which an application system operates. Extended strings, as a
datatype introduced by Manfred Thaller [1996, 2006], look as
a suitable candidate for the implementation of the markup
loop.
Markup originates a loop connecting the structure of the
text’s expression to the structure of the text’s content. An
implementation of the markup loop would considerably
enhance the functionality of text representation and processing
in a digital edition. To achieve implementation, markup
information could be integrated into the object (or datatype)
‘string’ on which an application system operates. Extended
strings, as a datatype introduced by Manfred Thaller [1996,
2006], look as a suitable candidate for the implementation of
the markup loop.
Bibliography
[Buzzetti, 1992] Buzzetti, Dino, Paolo Pari e Andrea Tabarroni.
‘Libri e maestri a Bologna nel xiv secolo: Un’edizione come
database,’ Schede umanistiche, n.s., 6:2 (1992), 163-169.
[Buzzetti, 2002] Buzzetti, Dino. ‘Digital Representation and
the Text Model,’ New Literary History, 33:1 (2002), 61-87.
[Buzzetti, 2004] Buzzetti, Dino. ‘Diacritical Ambiguity and
Markup,’ in D. Buzzetti, G. Pancaldi, and H. Short (eds.),
Augmenting Comprehension: Digital Tools and the History of Ideas,
London-Oxford, Offi ce for Humanities Communication,
2004, pp. 175-188: URL = <http://137.204.176.111/dbuzzetti/
pubblicazioni/ambiguity.pdf>
[Buzzetti and McGann, 2006] Buzzetti, Dino, and Jerome
McGann. ‘Critical Editing in a Digital Horizon,’ in Electronic
Textual Editing, ed. Lou Burnard, Katherine O’Brien O’Keeffe,
and John Unsworth, New York, The Modern Language
Association of America, 2006, pp. 51-71.
[Buzzetti, forthcoming] Buzzetti, Dino. ‘Digital Editions and
Text Processing,’ in Text Editing in a Digital Environment,
Proceedings of the AHRC ICT Methods Network Expert
Seminar (London, King’s College, 24 March 2006), ed. Marilyn
Deegan and Kathryn Sutherland (Digital Research in the Arts
and Humanities Series), Aldershot, Ashgate, forthcoming.
[Dubin, 2003] Dubin, David. ‘Object mapping for markup
semantics,’ Proceedings of Extreme Markup Languages 2003,
Montréal, Québec, 2003: URL = <http://www.idealliance.
org/papers/extreme/proceedings//html/2003/Dubin01/
EML2003Dubin01.html>
[Dubin and Birnbaum, 2004] Dubin, David, and David J.
Birnbaum. ‘Interpretation Beyond Markup,’ Proceedings
of Extreme Markup Languages 2004, Montréal, Québec,
2004: URL = <http://www.idealliance.org/papers/extreme/
proceedings//html/2004/Dubin01/EML2004Dubin01.html>
[McGann, 1991] McGann, Jerome. The Textual Condition,
Princeton, NJ, Princeton University Press, 1991.
[McGann, 1999] McGann, Jerome. ‘What Is Text?
Position statement,’ ACH-ALLC’99 Conference Proceedings,
Charlottesville, VA, University of Virginia, 1999: URL =
<http://www.iath.virginia.edu/ach-allc.99/proceedings/hockeyrenear2.
html>
[Rehbein, forthcoming] Rehbein, Malte. Reconstructing the
textual evolution of a medieval manuscript.
[Rehbein, unpublished] Rehbein, Malte. Göttinger Burspraken
im 15. Jahrhundert. Entstehung – Entwicklung – Edition. PhD
thesis, Univ. Göttingen. [Renear, 2001] Renear, Allen. ‘The descriptive/procedural
distinction is fl awed,’ Markup Languages: Theory and Practice,
2:4 (2001), 411–420.
[Sperberg-McQueen, Huitfeldt and Renear, 2000] Sperberg-
McQueen, C. M., Claus Huitfeldt, and Allen H. Renear.
‘Meaning and Interpretation of Markup,’ Markup Languages:
Theory and Practice, 2:3 (2000), 215-234.
[Thaller,1991 ] Thaller, Manfred. ‘The Historical Workstation
Project,’ Computers and the Humanities, 25 (1991), 149-62.
[Thaller, 1996] Thaller, Manfred. ‘Text as a Data Type,’ ALLCACH’
96: Conference Abstracts, Bergen, University of Bergen,
1996.
[Thaller, 2006] Thaller, Manfred. ‘Strings, Texts and Meaning.’
Digital Humanities 2006: Conference Abstracts, Paris, CATI
- Université Paris-Sorbonne, 2006, 212-214.