A methodological approximation to the study of linguistic variation in digital interactions

Abstract

The goal of this paper is to investigate the usefulness of different corpora of online Spanish for the study of linguistic variation. In order to do so, we compare five different linguistic phenomena in two corpus of computer mediated communication, namely, esTenTen, which consists of various types of online texts, and the Twitter corpus of the project Proyectando la variación lingüística de Internet. The five linguistic phenomena under study are the loss of intervocalic /d/, the use of 2nd plural personal pronouns, the non-referential use of ello, the plural agreement of existential haber and the colloquial use of the superlative suffix -érrimo. The analysis shows that automatically compiled macrocorpora that do not differentiate among digital genres might show a high level of statistical noise. Since they favour monological contexts, they might be problematic for the documentation of highly marked phenomena, be it diatopically, diastratically or diaphasically. The major advantage they show is the large quantities of data they offer. On the contrary, more controlled corpora that also privilege more dialogical contexts are preferable for the study of linguistic variables typical of communicative immediacy, despite the fact that they are more difficult to collect and consult.