Abstract
The Palimpsest poster presentation is meant to
supplement the theoretical discussion provided by the session on Computer
Assisted Literary Translation Studies (CoALiTS). A first example of a Windows
application realizing aspects of the theoretical investigation presented in the
session, the presentation offers the opportunity to try out the latest version
of the Palimpsest software and to discuss practical and
theoretical issues on the background of this "hands-on" experience.
Drawing on established models and methods of comparative literary and linguistic
studies and fusing them with computer scientific resources, Palimpsest has interdisciplinary discourse at its heart and
explores new technical as well as literary issues: the recursive process of
software design for literary purposes and the re-formulation of these purposes,
leading to a re-writing of the software, is expected to grant new insights into
theoretical and practical issues of multilingual text processing that are of
interest beyond literary translation as such.
A decidedly experimental and open application, Palimpsest currently divides into two parts: the pre-processors for
the creation of word- and phrase-based tables, and the main application with the
Palimpsest Viewer and the statistics and
concordancing facilities. The presented version recurs to an already
pre-processed corpus of six texts, the first italicized passage of Virginia
Woolf's The Waves and its three German and two French
translations.
The central feature the Palimpsest Viewer offers is the
multilinear display of the source text (ST) and its various destination texts
(DTs), based on the unit of the phrase as discussed in the session. The Viewer
offers to browse phrase by phrase through any of the texts, and to access a
specific phrase directly by searching for a particular word or lemma it
contains. Via an "Align" button, all other texts can be aligned to the selected
phrase. Apart from this, the Viewer screen offers a display of the existing
links between ST and DT which can be edited by a mouse click.
Accessible via a pull-down menu, the statistical routines produce dBase tables
containing the word- or phrase-based, sentence or not-sentence-aligned output.
These files trace the interrelation of the texts by means of measuring the
relation of the word- or phrase-order of one text to the others, giving either
the accumulative effect in the not-sentence-aligned data or the sentence-aligned
results.
Also activated via a pull-down menu, the concordancing routines produce a dBase
table containing a word- or phrase-based concordance relating the respective
units of the currently selected text to all others.
Looking towards future developments, the as yet star-like organization linking
all DTs to the ST, but not to one another, will surely undergo reconsideration,
and the conversational form of the presentation during the ACH-ALLC conference
seems particularly appropriate to discuss such matters in interdisciplinary
conversation; also, the yet quite time-consuming task of manual pre-processing
might be eased significantly by co-operation with tagging projects which, in
return, could profit from some of the data a Palimpsest
tagged multilingual corpus offers.