Diachronic text traditions as "systems" Where the worries begin
	   
   Daniel Apollon University of Bergen Dept. of Literary,
   Linguistic and Aesthetic StudiesResearch Group for Digital
   Culture Email: Daniel.Apollon@uib.no

29 january 2010

During the last decades, three areas of innovations in the field of
mathematics and information science have encouraged textual scientists
to envisage, again, the feasibility of algorithmic reconstruction of
diachronic text traditions: 1. The wide adoption of graph theoretical
methods and multivariate scaling methods in all kinds of contexts,
2. The decisive role played by cladistics in evolutionary phylogenetics,
3. The development of various text encoding standards, parallel corpus
alignment, syntactic parsing, and collation techniques allowing
efficient preprocessing of huge collections of variants and other
textual characteristics. The key message of this paper will be that,
significant challenges persist in textual phylogenetics and that
fundamental questions about "textual traditions as systems" remain
unanswered. A few of these issues will be reviewed in this contribution:
What are the ambitions of stemmatology? We will contrast (a) prototypic
reconstruction by means of the most likely hierarchical binary graphs with
(b) dynamic morphogenetic modelling of diachronic textual change. We will
question the usefulness of the phylogenetic assumption to map adequately
all patterns of diachronic textual evolution. Is the binary branching tree
a form of epistemic myopia? We will challenge the general applicability
of hierarchical binary textual phylogenetics by contrasting it with
such patterns of textual evolution that cannot be mapped satisfactorily
by various brands of binary trees. Possible candidates for alternative
modelling are e.g., * Wikipedia's textual evolution, the largest
ever and fully observable textual tradition in the making, with full
critical apparatus supplied by its contributors and moderators, as well
as extensive discussions provided by the contributors calls for richer
entity-relation mappings, Avant-textes as understood by the French school
of textual genetics (Genetique Textuelle).

*

How far can the analogy between biological evolutionary systems and
textual traditions be stretched? The exploitability of the analogy between
biological organisms in evolutionary situations and diachronic textual
change can be questioned in several ways. While evolutionary science
exploits a consensual representation of what an organism is, there is
no such consensus on what constitutes a text or a textual tradition. As
a consequence, mapping of genetic variation and its consequences in
biology and of variants in text traditions differ considerably. The
former expresses a general theory, while the latter remains a collection
of fragmented insights and expert

rules (e.g. the lectio difficilior potior rule). The recent split between
textual genetics and text genealogy expresses a growing demand for
richer reconstruction models. To add insult to injury, the fundamental
remise en question by the New Critique on both sides of the Atlantic of
the status of a text, e.g., introducing the closure problem, pervasive
intertextuality, the opposition between discourse and inscription, etc.,
forces a plurality of potentially conflicting epistemic attitudes. Which
kind of âsystemâ is a textual tradition? Darwinian evolution (covering
all stages from molecular genetics to macro-morphology) offers a systemic
understanding of synchronic variation and diachronic change. There is,
yet no such general theory of âdiachronic textual systemsâ (DTS). The
benefits of computerassisted massive data collection of variants and
alignment methods do not include such theory as a bonus. The key challenge
for reconstructive and remodelling textual science is to explore the
systemic nature of textual transmission and diffusion. This implies
modelling not only which previous textual states generated new textual
states (phylogenetics), but also how such changes take place locally
and globally (dynamic morphogenetics). Such modelling would require
one 1. To describe the nature of the entities and the relations that
constitutes these textual states, 2. To organise these into functionally
interconnected layers (e.g. translation error layer, lacunae, redactional
interventions), 3. To explore and offer some mapping of the interaction
between endogenous processes (e.g. spontaneous mutations, mechanical
copy errors, lacunae etc.) and exogenous process (e.g. institutional
forces, cultural shift, political events, etc.). This raises the issue
of the permeability of the boundary region between the material text and
its world and the ability of current graph theoretical models to offer
such rich mappings. 4. To model the extent and configuration variational
space of textual traditions. All known textual traditions happen to vary
(as biological organisms do) only within certain strict boundaries and
not all variational sequences are likely to occur (it is unlikely that
any textual witness of, say, the Hebrew Book of Ruth, will evolve into
pure Dadaist jumble or into a text close to Vladimir Nabokov's Lolita
!) We will conclude by inviting discussion on two special issues: (a)
the potential of Wikipedia to provide new opportunities for modelling
DTSâs and (b) the utility of an artificially generated DTS as a
scientific playground for gauging diverse algorithmic methods.