LINGUIST List 13.1051

Tue Apr 16 2002

Diss: Computational Ling: Al-Asali "Computer..."

Editor for this issue: Karolina Owczarzak <karolinalinguistlist.org>


Directory

  1. al-asali, Computational Ling: Al-Asali "Computer-Assisted Translation..."

Message 1: Computational Ling: Al-Asali "Computer-Assisted Translation..."

Date: Mon, 15 Apr 2002 17:47:56 +0000
From: al-asali <al-asaliun.org>
Subject: Computational Ling: Al-Asali "Computer-Assisted Translation..."


New Dissertation Abstract

Institution: AI-Mustansiriya University
Program: Department of Translation
Dissertation Status: Completed
Degree Date: 2000

Author: Sa'ad I. Al-Asali 

Dissertation Title: 
Computer-Assisted Translation of English Scientific Texts into Arabic:
Designing a computer prototype for cohesive ties

Linguistic Field: Computational Linguistics

Dissertation Director 1: Majeed Al-Mashta 
Dissertation Director 2: Akram M. Othman 


Dissertation Abstract: 

The present study focuses primarily on translation problems related to
semantic ambiguities of conjunctions which are, in most cases, solved
through adapting a textual analysis. It aims at designing a
computerised semantic analyser for MT purposes and examining the
possibility of incorporating linguistic models into this analyser. To
achieve these tasks, certain theoretical and practical approaches are
followed in the present work. They can be divided into two: linguistic
and computational.

The former involves the exploration of conjunctive ties on two textual
levels: intersentential and interclausal. To set up the framework of
this exploration, two interrelated models of cohesive patterns are
adapted. The first is Halliday and Hasan's model of lexico-grammatical
relations (1976); the second is Hallidayan model of logico-semantic
relations (1985). 

This study is built on the assumption that the adoption of the above
models of cohesion can cope with ambiguities related to choosing
appropriate target language (TL) equivalents since the textual ties
have an essential cohesive effect due to the continuity of meaning in
the text. Hence, translation problems can partially be solved through
textual processing. A frequent problem that may occur is related to
the misinterpretation and the subsequent misrepresentation of text
cohesion; that is, the misinterpretation of the ties or conjunctions
within a text and which partially provide the semantic unity of that
text.

Computationally, the rendition of cohesive devices for the purposes of
the computer-assisted translation (CAT) requires a proper programmable
model which operates both intersententially and interclausally. For
this aim, a prototype parser, equipped with such a powerful grammar as
definite clause grammar (DCG), is developed to carry out the analysis
of the SL syntactic cohesion. The parser employs a set of DCG
formalisms with newly adapted approaches to overcome the syntactic or
semantic ambiguities of conjunctions.

Having to decide on a primary criterion for arranging the colourful
mixture of multidisciplinary topics involved in this work, the current
thesis is grouped into four chapters, ranging from cohesion-based
conjunctions to MT systems.

In the first chapter, a major linguistic-oriented subject on textual
ties in scientific English represented by conjunctions is
described. It also reviews main characteristics of scientific
English.

The next chapter surveys artificial intelligence (AI) concepts, NLP
methods and MT systems. It describes a variety of different
computational techniques which can be applied to NLP, in general, and
MT, in particular. 

Chapter three discusses the prototype in action. It demonstrates
current problems in the MT area, especially those that overlap with
textual ties. It focuses on the components of the programs and
outlines the work of the parser developed to this project. 

The final chapter presents conclusions and recommendations.




Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue