English–Latvian Toponym Processing: Translation Strategies and Linguistic Patterns

Proceedings of the 13th Annual Conference of the EAMT, pages 81–87, Barcelona, May 2009 English-Latvian Toponym Processing: Translation Strategies and Linguistic Patterns Tatiana Gornostay Tilde, Latvia tatjana.gornostaja@tilde.lv Inguna Skadiņa Tilde, Latvia inguna.skadina@tilde.lv Abstract The paper presents a study of a challeng- ing task in machine translation and cross- language information retrieval – transla- tion of toponyms. Due to their linguistic and extra-linguistic nature, toponyms de- serve a special treatment. The overall translation process includes two stages of processing: dictionary-based and out-of- vocabulary toponym translation. The lat- ter is divided into three steps: source string normalisation, translation, and tar- get string normalisation. The translation process implies an application of transla- tion strategies and linguistic toponym translation patterns. Possible translation strategies, including transliteration and translation per se along with combined strategies, and linguistic toponym transla- tion patterns, including multi-word pat- terns as well, were investigated and im- plemented for English-Latvian machine translation. 10,000 The UK-related topo- nyms from Geonames were selected for a development set. The evaluation of out- put quality on basis of a test set has showed 67% accuracy in out-of- vocabulary translation: 58% on a set con- taining one-word toponymic units and 81% on a multi-word test set. 1 Introduction The paper presents a study of a challenging task in machine translation (MT) and cross-language information retrieval (CLIR) – translation of toponyms. Due to their linguistic and extra- © 2009 European Association for Machine Translation. linguistic nature, toponyms deserve a special treatment. In general toponyms are studied by toponymy and represent names of places comprising the following types: hydronyms (names of bodies of water: bays, streams, lakes, lagoons, oceans, ponds, seas, etc., e.g. Thames as a river); oronyms (names of mountains, cliffs, craters, rocks, points, etc., e.g. Bexhill as a mountain); geonyms (general names for streets, squares, lines, avenues, paths, alleys, roads, embankments, etc.); oeconyms (names of populated places: an administrative division, country, city, town, house or other building). The first part of the paper overviews the concept and nature of toponyms along with existing toponym translation strategies (TS). The second part of it focuses on the developed and implemented English-Latvian toponym MT approach, including a description of TSs and linguistic toponym translation patterns (LTTP). 2 Concept and Nature of Toponyms After Geoffrey Leech (1981) we can accept a special status of toponyms as proper names with- out a conceptual meaning as we cannot perform any componential analysis for them. However, we cannot but admit the fact that many toponyms are at least meaningful etymologically, e.g, Cambridge – bridge over the river Cam (Leidner, 2007), and, as Leidner pointed out, this etymolo- gy might or might not be apparent to a speaker. This feature makes toponyms difficult for processing. 81 Besides, toponyms are not unambiguous. Leidner (2007) describes three types of the topo- nymical ambiguity: morpho-syntactic ambiguity: a word it- self may be a toponym or may be a com- mon noun in a language, e.g. Hook as the populated place in the UK versus hook as a common …

Content truncated to 500 words