Lexical Allocation in Interlingua-Based Machine Translation of Spatial Expressions
Given a spatial expression, or its computational semantic form, how is the expression’s spatial semantics to be allocated lexically, that is, among the expression’s entries in the lexicon? In interlingua-based machine translation (MT) research, lexical allocation is the problem of allocating or subdividing a linguistic expression’s full interlingual (IL) structure into the substructures that are lexical IL forms, that is, in the lexicon. Here we present our work developing IL forms and an IL lexicon for translating English spatial expressions into Turkish. We describe different ways in which spatial information is allocated to lexical IL forms in our MT system, preruntime during lexicon construction and at runtime during lexical selection.