Abstract: In this study a number of various programs has been considered to reveal opportunities and distinctive features of modern automatic translation systems and applications based on machine translation. In particular three different kinds of translator’s program toolkit are being considered.
Keywords: computer-assisted translation, machine translation, linguistic ambiguity, linguistic processor.
First of all it is should be stressed that there is an essential difference between two terms – “a machine translation” (MT) and “a computer-assisted translation” (CAT). It is known that the MT is a translation technology of the texts from one natural language into another by a special computer software. In other words the text is completely translated by the use of machine translation system. In case of the using of CAT system thetranslation process is fully realized by a person-translator; the software only assists the translator in achievement of his goal.
In addition, there can be no doubt that MT and CAT, on the one hand, and traditional translations by a translator, on the other hand, they supplement each other. Let us consider one simple explanation. In the conditions of a global informational space two categories of people that hold interest in execution of a translation of large amount of information, presented in the most various languages:people are spreading information and people are consuming information.
Thus these categories of people can be classified depending on their goals:
• formulation of a text or its synthesis,
• analysis of textual information or understanding the one.
The synthesis of the texts where the cognitive side of communication is especially important (a technical documentation, the agreements, the note of the Foreign Ministry, a scientific labour) is traditionally conducted by a translator.
While translating without the use of machine translation systems for such aims as the search of information for writing an abstract or an analytical note, correspondence of the technical support of the company with a foreign customer, turn out economically disadvantageous and require excessive time. Such information is necessary here and now for quick analysis and an application. It is just that very case when the machine translation systems are irreplaceable.
In this study a number of various programs will be considered to reveal opportunities and distinctive features of modern automatic translation systems and applications based on machine translation. In particular three different kinds of translator’s program toolkit will also be considered:
• MCHI system, which is an application based on machine translation;
• OmegaTtranslation memory application, which is a computer-aided system.
• ETAP-3 linguistic processor;
MCHI (Multilingual Chat with Hint Images) is a multilingual chat system designed and developed by scientists from Tokyo University of Agriculture and Technology. This system is based on machine translation and provided with “a presentation function of images related to the contents of the messages by utterers so that listeners are able to notice mistranslation”.[ 1 ] MCHI system has such merits as supporting multilingual communication, short enough response time, small user’s cognitive load. The current version of MCHI supports six languages – English, French, Japanese, Korean, Vietnamese and Chinese. The second feature of the system is ensured by a combination of programming language PHP and AJAX technology. Finally, the use of images as hints allows one to achieve the third merit of MCHI, because the user can easily understand these images.
OmegaT is a computer-assisted translation (CAT) tool, which is based on Translation Memory technology. It is an application intended for professional translators. A customer can modify OmegaTfreelyand/or re-distribute it in accordance with the user license. The OmegaT toolhas the following features:
• Translation Memory: OmegaT stores your previous translations and then transfer translation of reasonably similar segments of text you are currently working on.
• Terminology management: OmegaT uses simplified bilingual dictionaries for a special domain that is a glossary containing translations of small phrases or single words.
• Working with complex file structure: OmegaT provides the possibility to translate the materials from a single file to a folder containing subfolders each with a number of files in a variety of formats.
ETAP-3 is a multi-language linguistic processor developed by the Computational Linguistics Laboratory of the Institute for Information Transmission Problems of RAS. It is based on I. Melčuk’s Meaning «a rule-based machine translation system;
1. a system of synonymous paraphrasing of sentences;
2. a workbench for syntactic annotation of text corpora;
3. a Universal Networking Language translation engine;
4. a natural language interface to SQL-type databases;
5. a grammar checker.
The current ETAP-3 machine translation options contain the Russian-to-English and the English-to-Russian translation pair; also there are several prototypes: Russian – French, Russian – German, Russian –Korean and Russian – Spanish.
A problem of linguistic ambiguity is a great obstacle to the scientists engaged in the development of machine translation systems. Let us consider the following example of lexical and syntactic ambiguity: “He made a general remark that everything was normal”. Thissentencecan betranslatedin two ways:
• ‘Он сделал общее замечание, что все нормально’ and
• ‘Он заставил генерала заметить, что все нормально’.
Another non-trivial example of ambiguity in the Russian language looks as follows: “Тритипасталиестьнаскладе”. ‘There are three types of metal in stock’ or ‘three men began to eat in a warehouse”. It is obvious that in such cases a translator should interfere in the working process of a machine translation system, and he should choose one or another variant of translation.
Software tools based on machine translation are created to facilitate everyday processing of natural language texts, and they cope with it on a satisfactory level. Of course, there are problems in the field of computational linguistics that need to be solved. However one cannot but agree with the fact that the prospect of creating state-of-the-art machine translation system looks very promising.