Interlingua in machine translation

Abstract

The concept of interlingua is of seventeenth century provenance, and it gained the attention of computer scientists in the context of MT. The idea was that if there are n source languages (SLs) then n(n-1) algorithms are required to translate each SL to every SL, except itself. Having an interlingua (IL), only 2n algorithms are needed: to translate from SL to IL and from IL to target language (TL). For n3 interlingual approach needs less algorithms. There are two general interlingual approaches: syntactic and semantic; the first tries to pay attention mainly to syntactic aspects of languages, the second is interested in the information their texts convey.