Web-Scale Expressivity and Ambiguity in a Translation Interlingua: The DLT Intermediate Language

An exploration of 32 controlled natural languages, based on English and eight other languages, revealed one, the DLT Intermediate Language, a controlled variety of Esperanto, that was designed for multidomain use and was thoroughly enough documented so a domain author could draft a realistic range of Web content in it. A test of this language on a sample of sentences from Web documents in the health and human rights domains found it expressive enough to represent all sentences in the sample, with low structural ambiguity, but some structural and much semantic ambiguity remained that could interfere substantially with human and machine comprehension.