Machine Translation
Machine Translation
Dateline: 03/13/00
The dream of building an universal translator - the equivalent of the one found in Douglas Adams' The Hitchiker's Guide to the Galaxy - is driving the research efforts around the world from the early 1930s. Serious efforts to develop Machine Translation systems were under way soon after the ENIAC was made in 1946, and the first known public trial took place in January 1954 at Georgetown University. We've made a remarkable progress in the past fifty years, but MT involves so many complex tasks that current systems give only a rough idea or indication of the topic and content of the source document. These systems tend to reach a "quality barrier" beyond which they cannot go, and work best if the subject matter is specific or restricted, free of ambiguities and straightforward - the typical example for this type of text are computer manuals. We'll need more advanced systems to handle the ambiguities and inconsistencies of the real world languages - no wonder that translation, whether performed by machine or by human, is often regarded as an art discipline, and not as an exact science.
Today we have an entire range of translation methods with varying degrees of human involvement. Two extreme ends of the translation spectrum are fully automatic high quality translation (FAHQT) which has no human involvement and traditional human translation which has no machine translation input. Human aided machine translation (HAMT) and machine aided human translation (MAHT) lie in between these extremities, with HAMT being primarily a MT system that requires human aid, whereas MAHT is human translation method that is utilizing MT as an aid or tool for translation. The term Computer Assisted Translation (or Computer Aided Translation) - CAT - is often used to represent both HAMT and MAHT. For an excellent discussion on these topics see this article written by Steven Pang.
So, what's so hard about Machine Translation? There's no such thing as a perfect translation, even when performed by a human expert. A variety of approaches to Machine Translation exists today, with direct translation being the oldest and most basic one of them. It translates texts by replacing source-language words with target-language words, with the amount of analysis varying from system to system. Typically it would contain the correspondence lexicon - list of source-language patterns and phrases and mappings to their target-language equivalents. The quality of the translated text will vary depending on a size of the system's lexicon and on how "smart" are the replacement strategies. The main problem with this approach is its lack of contextual accuracy and the inability to capture the real meaning of the source text.
Going a step further, syntactic transfer systems use software parsers to analyze the source-language sentences and apply linguistic and lexical rules (or transfer rules) to rewrite the original parse tree so it obeys the syntax of the target language. On the other hand, interlingual systems translate texts using a central data representation notation called an interlingua. This representation is neutral to any languages in the system and breaks the direct relationship that a bilingual dictionary approach would have. Statistical systems use standard methods for translation, but their correspondence lexicon are constructed automatically, using advanced alignment algorithms from a large amount of text for each language, usually available in online databases.
For a detailed introduction to this field see "MACHINE TRANSLATION: An Introductory Guide" by Doug Arnold et al.
Going back to the practical applications of the Machine Translation technologies, the following is a list of well known MT packages and links to the related resources. Please do not hesitate to contact me with additions to this list, or any other comment or suggestion. Most of the modern systems mix several of the approaches described above, so it is often difficult to sort them out according to the translation strategy. That leaves us with the simplest sort criteria - alphabetical order:
- AltaVista Translations
Translate your texts or URLs online: English -> French, German, Italian, Portuguese, Spanish and vice versa.
- Dja Vu
CAT system that stores your translations in a "memory database" and reads over your source text, instantly retrieving applicable translations whenever it finds something equal or similar to the sentence you are working with.
- Language Engineering Corporation
Full range of consultancy, integration, and development services, providing translation and knowledge management solutions for major global enterprises. LEC has developed the LogoVista Translation Model to provide greater flexibility and quicker implementation for its customers. The model promotes cooperation between vendors, allowing translation components to be developed independently while ensuring compatibility.
- L&H; iTranslator
A client/server system with a browser-based user interface, L&H; iTranslator Enterprise; L&H; iTranslator Publish for an instant, gist-level translations of Internet or intranet Web pages; a limited form of iTranslator, L&H; iTranslator Wireless, is available at no cost to users of Palm Computing's Palm VII connected organizer.
- Linguatec
PT 2000 home, PT 2000 Office and other translation programs + Business English with 25,000 English sentences for 1 million possible combinations. Includes English dictionary and dictionary of idioms, voice output and Word integration.
- Logos Corporation
Logos Corporation was one of the first to introduce a machine translation (MT) production system in the early 1970s. LogosServer, the Logos machine translation engine, translates new text and returns draft-quality translation ready for post-editing with the original formatting preserved. LTMTM, the translation memory component, maintains a database of previously translated sentences and applies them automatically when similar or identical text is found in new documents. LogosClient, the user-friendly translation interface, sends documents from your desktop to the Logos machine translation engine for translation or terminology verification. ALEX, the Automatic LEXicographer, builds customized dictionaries by adding new terms to existing subject-matter dictionaries through the use of simple and intuitive prompts and questions.
- MT at IBM
Machine Translation within IBM has a long history. Born in the IBM U.S. Scientific Centers, the idea of having written language automatically translated by the computer has been pursued in various applications developed in IBM Software Laboratories. IBM TranslationManager is used for machine assisted human translation for professional translators; TransLexis is the tool for company wide terminology management; MachineTranslation is the machine translation engine.
- Project MT
PROject MT company is specialized in the development of computer-aided translation systems for the main European languages, including online translation systems.
- Smart Translator
The SMART Translators are a series of computer programs that use Artificial Intelligence to translate Controlled English and Simplified English texts into other languages.
- Star
STAR's Computer-Aided Translation System is developed by taking advantage from the practical long term experience of one of the largest and most competitive translation service groups in the world.
- SYSTRAN
One of the best known MT companies, delivering products for Windows 95, 98, NT, Linux, HP UX, Sun Solaris, Sun OS, and DEC Unix.
- Tolken99
Tolken99 is a complete word processor that also can perform very fast translations of documents, e-mail and web pages, etc. to and from English and Swedish.
- Trados
Trados products use translation memory (TM) technology by storing previously translated sentences and building up a large repository of source text and foreign language equivalents over time. With the assistance of fuzzy matching functionality the system also recognizes similar, not just identical, translation matches. This technology is very efficient for translating repetitive texts (such as standardized forms, operating manuals, contracts, product updates and successive versions of other texts).
- Transcend
Transcend enables you to get fast, accurate translations of business documents, e-mail, Web pages and more.
- TransLinGO!
TransLinGO! Plus translates users' English keywords into Japanese and conducts the search with the search engine of their choice. Search results are translated into English, allowing users to collect Japanese information freely without knowing Japanese at all.
- Transparent Languages
Online translation services, instantaneous translation of unlimited length documents using the topical dictionaries for law, business and more.
- World Language Resources
World Language Resources provides foreign language products for over 487 languages. Software types include: language systems, dedicated word processors, language learning, translation and more.
Other interesting links:
- American Translators Association
The purpose of this site is to help you learn more about the American Translators Association and the translation and interpretation professions.
- Computing Research Laboratory
CRL, at New Mexico State University, is a non-profit, self-supporting research enterprise committed to basic research and software development in advanced computing applications, including AI and MT.
- CMU LTI Projects
There are currently a number of active projects at the CMU's Language Technologies Institute in the fields of machine translation, information retrieval, and speech.
- European Association for Machine Translation
EAMT is an organization that serves the growing community of people interested in MT and translation tools, including users, developers, and researchers of this increasingly viable technology. The EAMT is one of three regional associations of the International Association for Machine Translation (IAMT), which counts an increasing number of members worldwide. The EAMT is the only organization of its kind in Europe.
- MT respository
This directory contains machine translation software and
related materials.
- Language Dictionaries and Translators
An excellent resource from Rivendell International Communications.
- USC/ISI
The Natural Language Processing group at the Information Sciences Institute of the University of Southern California (USC/ISI) is currently involved in various aspects of computational linguistics/natural language processing, including MT.