Automated translation

One of the pioneers of the idea of ​​automated translation is the Soviet scientist Pyotr Petrovich Troyansky , who proposed in 1933 a “machine for selecting and typing words when translating from one language to another.” The car consisted of a table with an inclined surface, in front of which was fixed a camera synchronized with a typewriter. On the surface of the table there was a “glossary field” - a freely moving plate with words printed on it in three, four or more languages ^[1] .

The practical implementation of the idea of ​​automated translation began with the advent of computers : translators have always opposed the standard concept of the MP in those years, to which most research in computer linguistics was directed, but supported the use of computers to help translators.

In the 1960s, the European Coal and Steel Association (the predecessor of the modern European Union ) began to create terminological databases under the common name Eurodicautom ^[2] . In the Soviet Union for the creation of such bases was created VINITI .

In its modern form, the idea of ​​AP was developed in an article by Martin Kay in 1980 ^[3] , who put forward the following thesis: "The computer takes on routine operations and frees a person for operations that require human thinking ").

Currently, the most common ways of using computers in written translation are working with dictionaries and glossaries, translation memory ( English Translation Memory, TM ) containing examples of previously translated texts, terminological databases, as well as using so-called corpuses , large collections of texts on one or several languages, which gives a concise description of how words and expressions are actually used in the language as a whole or in a specific subject area.

Specialized tools are often used to localize software, for example, Passolo , which allow translating menus and messages into software resources and directly into compiled programs, as well as testing the localization correctness. For the translation of audiovisual materials (mainly films), specialized tools are also used, such as Aegisub , which combine some aspects of the translation memory, but additionally provide the possibility of the appearance of subtitles in time, their formatting on the screen, following video standards, etc.

In the case of simultaneous translation, the use of automated translation tools is limited due to its specificity. One example is the use of dictionaries downloaded on a PDA . Another example is the semi-automatic extraction of lists of terms in preparation for simultaneous translation in a narrow subject area ^[4] .

In narrow subject areas, with a large number of source texts and well-established terminology, translators can also use machine translation, which can provide good quality translation of terminology and stable expressions in a narrow area. The translator in this case performs post-editing of the received text. More than half of the texts within the European Commission (mainly legal texts and current correspondence) are translated using the MP.

Automated translation is a broad and not quite accurate concept covering a wide range of simple and complex tools. These may include:

• Spelling checkers that can be embedded in text editors or additional programs;
• Grammar checker programs that are also embedded in text editors or additional programs;
• Terminology management programs that allow translators to manage their own terminological base in electronic form. This could be a simple table created in a text editor, or a spreadsheet, and a database created in FileMaker. For more time consuming (and more expensive) solutions, there is special software, for example, LogiTerm, MultiTerm, Termex, TermStar, etc.
• CD dictionaries , monolingual or multilingual;
• Terminology databases stored on CDs or connected via the Internet, such as The Open Terminology Forum or TERMIUM;
• Full-text search programs (or indexers ) that allow the user to query the previously translated texts or various kinds of reference documents. In the translation industry, there are indexers such as Naturel, ISYS Search Software and dtSearch;
• Concordance programs that allow you to find examples of words or expressions in the context in use in a monolingual, bilingual, or multilingual package, such as a bitext or translation memory, such as Transit NXT ;
• Bitext , one of the recent innovations, is the result of the merging of the source text and its translation, which can later be analyzed using full-text search or concordance programs;
• Project management software that allows linguists to structure complex translation projects, transfer various tasks to different employees and monitor the process of performing these tasks;
• Translation memory management programs (TMM), consisting of a database of text segments in the source language and their translations into one or more target languages, for example Transit NXT ;
• Almost completely automatic systems resembling machine translation, but allowing the user to make certain changes in doubtful cases. Sometimes such programs are called machine translation with human participation.

• Translation memory
• Interactive machine translation
• Machine translate
• Linguistic software

• John Hutchins (1998), The origins of the translator's workstation. // Machine Translation , 13: 287-307, (PDF)