Controlled Language Translation
15th-17th May, 2003
The Joint Conference of the 8th International Workshop of the European Association for Machine Translation and the 4th Controlled Language Applications Workshop
Over the years, there have been many conferences on MT, involving rule-based approaches, statistical and example-based approaches, hybrid and multi-engine approaches as well as those limited to particular sublanguage domains. In addition, there has been an increased level of interest in controlled languages, culminating in the series of Workshops on controlled language applications. These have given impetus to both monolingual and multilingual guidelines and applications using controlled language, for many different languages.
Controlled languages are subsets of natural languages whose grammars and dictionaries have been restricted in order to reduce or eliminate both ambiguity and complexity. Traditionally, controlled languages fall into two major categories: those that improve readability for human readers, particularly non-native speakers, and those that improve computational processing of the text. It is often claimed that machine-oriented controlled language should be of particular benefit when it comes to the use of translation tools (including machine translation, translation memory, multilingual terminology tools etc.).
Experience has shown that high quality MT systems can be designed for specialized domains (e.g. METEO). However, the area of controlled translation has remained relatively unaddressed. This is rather strange given its undoubted importance. Such examples that exist use rule-based MT (RBMT) systems to translate controlled language documentation, e.g. Caterpillar's CTE and CMU's KANT system, and General Motors CASL and LantMark, etc. However, fine-tuning general systems designed for use with unrestricted texts to derive specific, restricted applications is complex and expensive.
There are several examples of using Translation Memory (TM) tools in a controlled language workflow, yet these have been primarily for combining TM and MT tools. Very few attempts have been made where Example-based MT (EBMT) systems have been designed specifically for controlled language applications and use. This is even harder to fathom: using traditional RBMT systems leads to the well-known `knowledge acquisition bottleneck', which can be overcome by using corpus-based MT technology. Furthermore, the quality of EBMT (and Translation Memory) systems depends on the quality of the reference translations in the system database; the more these are controlled, the better the expected quality of translation output by the system.
The primary aim of this unique conference, therefore, is to elicit papers on controlled translation, and provide a forum in which the problems may be outlined, possible solutions proposed, and in general to bring together developers, implementors, researchers and end-users from the publications, authoring, translation and localization fields to discuss how ideas from both the authoring and translation camps might be integrated in this common area. Some specific topics which might be addressed include:
In addition, we welcome contributions on MT as well as on controlled language which do not address the main theme per se. Suitable example topics include, but are not restricted to, the following:
- What is controlled translation?
- RBMT and controlled translation.
- TM/EBMT and controlled translation.
- Influence and interplay of controlled language upon both
source-language parsing and target-language generation in an MT system.
- Role of the lexicon in controlled translation.
- Can we expect better controlled translations from a hybrid
approach? Or from a multi-engine approach?
- Towards a Roadmap for controlled translation - the way ahead?
- MT for the Web;
- Practical MT systems;
- Methodologies for MT;
- Speech and dialogue translation;
- Text and speech corpora for MT and knowledge extraction from
- MT evaluation techniques and evaluation results;
- MT postediting.
- Examples of controlled languages: their definition, by whom, and
- Consequences for technical authors and implications
for Natural Language Processing;
- Practical experiences of teaching and using controlled languages;
- Application of controlled languages in speech systems.