Ad hoc compounds in English to Finnish machine translation

Abstract

Language is in a continuous development process, which poses a challenge to machine translation. People also tend to have their only styles of writing, which is often difficult to translate. A special problem is the so called ad hoc compounding, where the writer decides to construct such compounds, which do not exist in any dictionary. By compounds I mean here such clusters of words, which are written together, using '-' as a connecting diacritic, not the normal compounding, where members of a compound are written as separate words. In addition to ad hoc compounds, there are also established compounds using the same connecting method. This report describes the methods for handling these compounds in English to Finnish machine translation