Exploring a loan translation and its consequences in an oral bilingual corpus

Abstract

This work applies computational tools that have been used to model loanwords in newspaper corpora to an analysis of a loan translation in an oral bilingual corpus. The explicit goal of the contribution is to argue that a specific collocation found in a corpus of Spanish spoken in Texas, agarrar+NP (e.g., agarrar ayuda), is a loan translation that is calqued on English get+np support verb constructions (e.g., get help). We base our argument on the frequency and the linguistic distribution of the nonconventional usage within and between corpora and on the factors that favor its use. Our findings show that the overall frequency of agarrar is the same in Spanish in Texas as it is in the benchmark monolingual corpus of Mexican Spanish but that it is used differently in the two varieties, a difference that has grammatical, as well as semantic, ramifications.