Compiling an Estonian-Slovak Dictionary with English as a Binder

DENISOVÁ, Michaela. Compiling an Estonian-Slovak Dictionary with English as a Binder. Online. In Electronic lexicography in the 21st century: Post-editing Lexicography. Proceedings of the eLex 2021 conference. Brno: Lexical Computing CZ s.r.o., 2021, s. 107-120, 13 s. ISSN 2533-5626.

Základní údaje

• Originální název: Compiling an Estonian-Slovak Dictionary with English as a Binder
• Název anglicky: Compiling an Estonian-Slovak Dictionary with English as a Binder
• Autoři: DENISOVÁ, Michaela.
• Vydání: Brno, Electronic lexicography in the 21st century: Post-editing Lexicography. Proceedings of the eLex 2021 conference, od s. 107-120, 13 s. 2021.
• Nakladatel: Lexical Computing CZ s.r.o.

Další údaje

• Typ výsledku: Stať ve sborníku
• Utajení: není předmětem státního či obchodního tajemství
• Forma vydání: elektronická verze "online"
• WWW: Plný text
• Organizační jednotka: Fakulta informatiky
• ISSN: 2533-5626
• Klíčová slova anglicky: bilingual dictionaries; (semi)automatic compilation; intermediate language; Estonian; Slovak

Anotace

For such a rare language combination as Estonian-Slovak, it is complicated to find study materials designated for Slovaks learning Estonian, especially a bilingual dictionary, an essential language study resource. However, building a bilingual dictionary from scratch requires a lot of work and effort. The half-automatic computational methods and vailable open-source language resources offer a possible solution for this complicated task. One approach is to merge two already existing dictionaries that share a common language to derive a new language pair dictionary. However, as words are polysemous, many mistakes could occur while attempting so. Therefore, it is required to edit the aligned translations afterwards. This article describes the process of compiling the Estonian-Slovak dictionary created from English-Estonian and English-Slovak dictionaries. English was chosen as an intermediate language, as it is a well-resourced language, and all materials are easy to find. Various automatic techniques were applied in the editing step to decrease the number of incorrectly aligned translations. Finally, the techniques used and quality of the dictionary were manually evaluated on a random sample of 1,000 translations. The final version of the dictionary consists of 138,779 translations, and the Estonian headword list covers about 85% of basic Estonian vocabulary, which contains around 5,000 lemmas. The correct translations form approximately 40% of the dictionary. Additionally, a web application is being developed for this dictionary.