Units of Meaning, Parallel Corpora, and their Implications for Language Teaching

Translation equivalence is a key issue for all who apply multilingual skills in a professional environment. This includes language teachers, translators, lexicographers and terminologists, as well as experts in computational linguistics. Translation equivalence has therefore to be dealt with by academic foreign language teaching. There are two reasons. The units of meaning are only rarely the traditional single words; much more common are larger chunks, compounds, multi-word units, set phrases and even full sentences. In corpus linguistics, these are called collocations. They are the true vocabulary of a language. Collocations are statistically significant co-occurrences of words in a corpus. But they also have to be semantically relevant. They have to have a meaning of their own, a meaning that is not obvious from the meaning of the parts they are composed of. Whether an English text chunk is a true collocation or just a chain of words can only be decided from the perspective of a source language. This is why a list of English collocations for students with other native languages would have to be compiled from a parallel corpus. I will show how an approach to translation equivalence based on collocations yields results that can be applied in language teaching.