TABLE OF CONTENTS
ABSTRACTS
"Automatic Morphological Analysis of Basque"
Inaki Alegria, Xabier Artola, and Kepa Sarasola
Abstract
This paper describes the components of a robust and wide-coverage
morphological analyser for Basque. The analyser is based on the two-level
formalism and has been designed in an incremental way with three main
modules: the standard analyser, the analyser of linguistic variants, and
the analyser without lexicon which can recognize word-forms without having
their lemmas in the lexicon. Using lexical transducers for our analyser we
have improved both the performance of the different components of the
system and the description itself. The analyser is a basic tool for
current and future work on automatic processing of Basque and its first
two applications are a commercial spelling corrector and a general purpose
lemmatizer/tagger.
"Feature-Finding for Text Classification"
Richard S. Forsyth and David I. Holmes
Abstract
Stylometrists have proposed and used a wide variety of textual features or
markers, but until recently very little attention has been focused on the
question: where do textual features come from? In many text-categorization
tasks the choice of textual features is a crucial determinant of success,
yet is typically left to the intuition of the analyst. We argue that it
would be desirable, at least in some cases, if this part of the process
were less dependent on subjective judgement. Accordingly, this paper
compares five different methods of textual feature finding that do not
need background knowledge external to the texts being analysed (three
proposed
by previous stylometers, two devised for this study). As these methods do
not rely on parsing or semantic analysis, they are not tied to the English
language only. Results of a benchmark test on ten representative
text-classification problems suggest that the technique here designated
Monte-Carlo Feature-Finding has certain advantages that deserve
consideration by future workers in this area.
"Tampering with the Text to Increase Awareness of Poetry's Art"
Estelle Irizarry
Abstract
Theoreticians have linked the act of poetic creation inextricably to the
principle of linguistic 'play'. A number of Hispanic poets have
experimented with transformational and permutational creativity of the
type that computers can accomplish quite easily. Such computer-induced
play enhances the study of poetry by imbuing the poetic text with a new
and dynamic dimension in which on-screen manipulation destabilizes the
text, allowing the reader to explore it more thoroughly than is possible
in the fixed printed medium and to appreciate it as a unique blend of
word, structure and pattern. Well-known poems from writers who have
themselves experimented with textual alteration, as well as works of
others who have not, serve to illustrate diverse modalities of textual
alteration, which are grouped by the types of transformation carried out
by the computer.
"Virginia Woolf's 'The Waves' in French and German Waters: Computer
Assisted Study in Literary Translation"
Jan-Mirko Maczewski
Abstract
The case study analyses the first chapter of Virginia Woolf's novel The
Waves and its two French and three German translations with the help of
the PALIMPSEST suite of programs. Created specifically for such tasks, the
software provides assistance with viewing the texts in an interlinear
format and offers facilities for the automatic generation of multilingual
and -textual word and phrase based concordances and statistics. Pursuing
the typical aims of literary translation studies, the investigation
focuses
on an analysis of the relationships between the translations and the
original text as well as on a consideration of the influences that can be
identified within the corpus of translations; apart from encoding,
technical matters are not elaborated upon. The new ways of assessing
literary translations offered by PALIMPSEST yield noteworthy results which
contribute new empirical evidence to the critical debate on translation in
general and on the translations of The Waves in particular. As a result,
computer assisted literary translation studies appear as a field of
research worth exploring further.
"A Hybrid Disambiguation Model for Prepositional Phrase Attachment"
Haodong Wu and Teiji Furugori
Abstract
Prepositional Phrase (PP) attachment is a major cause of structural
ambiguity in natural language. Many proposals have increasingly relied on
large-scale corpus to resolve this problem. However, this approach
encounters the notorious sparse-data problem that produces poor results on
disambiguation. We in this paper offer a hybrid method which integrates
corpus-based approach with knowledge-based techniques for PP attachment
disambiguation. It explores a wide-variety of information, including
co-occurrence frequencies from annotated corpora, conceptual relationships
and conceptual features from a machine-readable dictionary, and syntactic
clues from our linguistic observations. We use dictionary definitions and
human knowledge to overcome the sparse-data problem. An experiment shows
an accuracy rate of 87.7% of our method over 3043 sentences in real
English text that contain ambiguous PPs. This result is better than those
of any existing methods.
The full text of the articles to which these abstracts refer will be published in Literary & Linguistic Computing, Vol 11, No 4, which is due to be published in December 1996. If you would like further details about the journal, including details of subscription rates, please contact:
Caroline Lock
Journals Marketing (H-CLC)
Oxford University Press
Walton Street
Oxford OX2 6DP
UK
Tel: +44 (0)1865 56767
Fax: +44 (0) 1865 267782
lockc@oup.co.uk
Further information on all Oxford journals can be found on the OUP Web site at: http//www.oup.co.uk/
Copyrights to preprints remain with Oxford University Press. Users may download preprints and reproduce them for their own personal use, but downloading of preprints for any other activity, including reposting to other electronic bulletin boards or archives, may not be done without the writen consent of Oxford University Press. It is the responsibility of Oxford University Press to notify the H-CLC editors when they wish to have abstracts removed from the H-CLC WWW site.