Multilingual Lexicons

A unidirectional bilingual lexicon with lemmas in language $S$ and definitions in language $T$ is described like a unidirectional MT system. The OLAC solution is to designate $S$ as the Subject.language and $T$ as the Language. A fully bidirectional lexicon intended for use by speakers of either language would be described in the same fashion as a bidirectional MT system:

<Language code= X/>
<Language code= Y/>
<Subject.language code= X/>
<Subject.language code= Y/>

As before, this solution has lost some structure, but retrieval behavior is correct nonetheless. Of course, this metadata could equally be the collection-level metadata for a set of two monolingual dictionaries, one in language $X$ and one in language $Y$. However, the Type.linguistic element would distinguish these cases by having different vocabulary items for monolingual and bilingual lexicons.

Multilingual lexicons may also exhibit star configurations: one-to-many (lexicons with definitions in multiple languages); or many-to-one (comparative wordlists). Here the treatment is analogous to the corresponding MT systems discussed above.

Finally, multilingual lexicons may map between all pairs of a set of languages, as in the case of some termbanks. In this case, all the languages are designated both using Language and Subject.language.