|
Forepaper by
Francisco
Jose Ricardo and Elin Johanne Sjursen
for Messenger
Morphs the Media 99
BEHIND THE NARRATIVE INTERFACE
-- ASPECTS OF STRUCTURE IN HYPERFICTION
In hyperfiction, certain computational
authorial techniques have been de rigueur, for instance,
anchors that link to different story elements as a function
of time or prior reader traversal; links whose names are
themselves are a form of description of story intentions;
re-visited lexias whose literal composition remains the
same, but whose implication changes as the story evolves;
etc.
With the repetition of certain
prominent techniques, it is clear that the envelope of
representation is being pushed to limits that are not
infinite. Authors are increasingly interpolating greater
cleverness into content anatomy to compensate for the
relative fixity of the interface and what it offers to the
writer -- limitations of features that are as true of
authoring systems like StorySpace as for web browsers. How
much farther can we go?
In light of this question, another
perspective is relevant to the design of hyperfictional
content. It involves managing the central role of what
happens relatively far behind the interface, as a function
of authorial design. In this back-end mode, the management
of lexias becomes a management of objects and the story
evolves, then, almost as a grammar, with the selective
interaction of the objects under specific conditions -- this
perspective no different than the approach of the database
designer and the object oriented programmer who organize and
manage an archive of objects. Here, we totally blur the line
between the poetic in programming -- as the creation of
autonomy in a program -- and hyperliterary authoring -- as
the creation of autonomous themes, plots, and actants in a
descriptive account. Both would join in aesthetic
unification designed to promote the possibility of
traversing a landscape between the unraveling and coupling
of symbolic patterns.
This also brings up an important
distinction -- that between author time and reader time
(like design time and run time modes in programming). In
author time (analogous to design time) the material
connections are made such as the reader will encounter them.
This includes the entering of fixed text lexias, links to
determinate locations, etc. In reader time, the connections
evolve (analogous to run time) as a result of the reader's
interaction with the work. In this mode (which is what we
are presenting for consideration), the work is as much
determined by realtime experience in conjunction with
algorithms for organizing, matching, and presenting story
contents -- the work is a greater manifestation of author
time and reader time components than is currently the case
with authoring and browsing systems. This is, again, the mix
of content and structure in actual performance for the
reader.
What, then, are the aspects of story
and structure that permit this unified approach to be
envisioned, and its intersection to be implemented? We set
up a matrix of comparison that links factors of the author
function with variables in the back-end management of story
elements -- that is, techniques and questions of character,
setting, and plot that map onto computational techniques not
present, prominent, or practical in current hyperauthoring
venues. These include three areas of computational
hermeneutics, story structure, and narrativity:
Computational Hermeneutics
For further consideration in the
workshop
Consideration of various forms of
hermeneutic recall and computational storgage of elements in
stories -- many as posed by the workshop participants in
their own work.
__Technique 1: Memory as archival.
How are elements of the story made
persistent? Discussion of how to divide the story and what
archival methods can be implemented to for selective and
simple recovery. The concept of recall in a story can be
implemented by given variables and saved state features in
authoring languages.
__Technique 2: Association as
retrieval.
How are elements of the story matched?
Beyond author time links, and keyword searches, there are
semantic information retrieval techniques that can implement
hermeneutic modes of historical relevance in a story. This
has major implications for the play of possible
interpretations, and is the inverse of technique 1
above.
Computational Story
Structure
For further consideration in the
workshop
The implication of how story
linearities relate to deeper mythic intentions of the work
-- for instance, via epic or sequel constructions. How
mythemes as lexias are the basis for hyperfiction. Also, the
philosophical question of where lies (and what remains of)
authorial intention in a system where the user can generate
her own techniques of traversal in reader time. And to the
degree that this is a desirable state, what formalist
narrative techniques are best implemented in author time
versus reader time.
__Technique 1: Formalist techniques of
defamiliarization
(e.g., Shklovskii's 1929 Theory of
Prose).
These include repetition, parallelism,
framing, embedding, juxtaposition in the story. Each of
these has a programmatic equivalent in the back end in
either a realtime (reader time) or "precooked" (author time)
mode of authorship. In reader mode, parallelism (for
example) can be determined by a semantic match of key terms
in a given lexia to other lexias. In author time, these
links pre-exist as part of the author's original content
design.
__Technique 2: structuralist
Mythemes
(Lévi-Strauss's 1958 the
structural Analysis of Myth).
These are basic and rearrangeable
units of signification which can map into lexial sequences
in hyperfiction. The idea here is that the story is a Markov
chain of successive states like a raiload track. At specific
points, the track can deviate toward another track. The
mytheme metaphor means that interchangeability is always
present, but also that certain units must proceed others, in
order for a given sense of order to emerge. The mythemic
metaphor is a high level version of the Todorov narrative
grammar, which focuses on singular elements as factors in
the outcome of the story. Being high level, the mythemic
metaphor is already a part of StorySpace, for
instance.
Computational Narrativity
For further consideration in the
workshop
What, for individual authors, are the
major narrative components of hyperfiction, how do authors
treat these, and what would be desirable if these components
could be enhanced programmatically?
__Technique 1: Narrative
Grammar
(e.g., Todorov's 1969 Grammaire du
Décameron).
In this analytic model, actions are
assimilated into verbs, characters to nouns, attributes to
adjectives. This is a virtual homology with objects in the
programmatic sense. What a character is given to do, for
instance, by the reader (who can, say, decide whether the
character will go to a specific location, interact with a
given person, or undertake some special act), reconfigures
not only the subsequent story development, but also other
characters and their attributes. Again, this level of high
granularity provides a programming onus, but also a
correspondingly much higher degree of interactive
variability.
__Technique 2: Greimas's syntactic
structures and Barthes's narrative codes.
(e.g., : Greimas's syntactic
structures and barthes's s/z).
Greimas's syntactic structures
(characters being classified by function in the story) and
Barthes's narrative codes (where story lexias are what is
thus classified) are equally promising. For instance, the
hermeneutic code of Barthes (cf. S/Z) which represents the
imposition of an enigma in need of resolution in the story,
can be explored as its own substory, whereby the user can
follow the links of the enigma, to the omission of other
plot sequences. This resembles the mytheme organization,
although the mytheme presupposes that the story was created
with a given underlying structure, whereas the narrative
codes of Barthes do not seek a larger structure, and instead
re-configures the story around reader-assigned strands of
signification.
Summary
We will explore how these theoretical
aspects of the story can be implemented programmatically,
rather than at the level of what the interface alone offers
(e.g., precooking the links in a story at "author" time) by
pointing out areas of overlapping convergence between the
narratological and the computational, so that the
hyperauthor can look beyond the tools endemic to current
environments and instead approach the production of more
individual creations from the ground up.
| 
