parallel text

A parallel text is a text in one language together with its translation in another language. Parallel text alignment is the identification of the corresponding sentences in both halves of the parallel text.

Large collections of parallel texts are called parallel corpora (see text corpus). Alignments of parallel corpora at sentence level are prerequisite for many areas of linguistic research. During translation, sentences can be split, merged, deleted, inserted or changed in order. This makes alignment a non-trivial task.

Bitext

In the field of translation studies a bitext is a merged document comprised of both source- and target-language versions of a given text.

Bitexts are generated by a piece of software called an alignment tool, or a bitext tool, which automatically aligns the original and translated versions of the same text. The tool generally matches these two texts sentence by sentence. A collection of bitexts is called a bitext database or a bilingual corpus, and can be consulted with a search tool.

History

The idea of the bitext is attributed to Brian Harris, who first wrote a paper on the concept in 1988, and has been promoted by the Université de Montréal-based RALI (Recherche appliquée en linguistique informatique, or Applied Research in Computational Linguistics), a group of computer scientists and linguists who study natural language processing. Pierre Isabelle and Claude Bédard are noted promoters of the concept of the bitext.

Bitexts and Translation memories

The concept of the bitext shows certain similarities with that of the translation memory. The main difference between a bitext and a translation memory is that a translation memory is a database in which its segments (matched sentences) are stored in a way that is totally unrelated to their original context; the original sentence order is lost. A bitext retains the original sentence order.

Note that the standard format for exchanging translation memories between CAT programs is TMX, an XML vocabulary published by LISA (Localisation Industries Association). TMX allows preserving the original order of sentences, so the previous paragraph should be taken with a grain of salt.

Bitexts are designed to be consulted by a human translator, not by a machine. As such, small alignment errors or minor discrepancies that would cause a translation memory to fail are of no importance.

Parallel Corpora in the Web

• The JRC-Acquis Multilingual Parallel Corpus of the total body of European Union (EU) law: Acquis Communautaire with 231 language pairs.^[1]
• The Opus project aims at collecting freely available parallel corpora
• COMPARA - Portuguese/English parallel corpora
• LILABAR - English/Russian parallel corpora
• Nunavut Hansard - English/Inuktitut parallel corpus

See also

• Natural language processing
• Machine translation
• Computer-assisted reviewing

External links and references

Documentation

• Parallel text processing bibliography by J. Veronis and M.-D. Mahimon
• Proceedings of the 2003 Workshop on Building and Using Parallel Texts
• Proceedings of the 2005 Workshop on Building and Using Parallel Texts

Donate to Wikimedia