Abstract:

Since the late 1950s, linguistic corpora have been used as a pedagogical tool for teaching English. Example of this is the pioneering work carried out by Randolph Quirk and his colleagues around the Survey of English Usage project. Quirk's work has given rise to two theoretical and methodological drifts: (i) on the one hand, the teaching of English to native students, especially at the university level; (ii) on the other hand, the teaching for non-native students who acquire it as a second language. In this last drift, it is worth noting the work of Sylviane Granger, who has achieved considerable advances in the implementation of textual corpora in English, generated by non-native university students, which are useful for recognizing patterns of linguistic learning.
Complementary to these drifts on the use of electronic corpora for teaching, it is worth noting that today there are several computational tools for their analysis, as is the case of Sketch Engine, one of the most complete platforms out there, developed by Adam Kilgarriff. The use of this kind of platforms allows us to formulate a research question for applied linguistics: does the use of electronic corpora, together with their analysis tools, configure a new theoretical and methodological perspective to approach the linguistic learning?
Bearing this question in mind, linguists such as Alex Boulton or Peter Crosthwaite consider that the use of linguistic corpora is the basis for a new learning model, which is guided by empirical data that reflects —in the Austin’s pragmatic point of view— the things that speakers do with their language. Therefore, the named Data-Driven Learning (or DDL) is an approach that focuses on pattern identification in large linguistic data sets, in order to discover regularities and relationships between linguistic units and structures.
In line with such ideas, this paper develops a brief evaluation regarding the impact that linguistic corpus have on the teaching of English, considering some results from a pilot study carried out with 19 Mexican students, assigned to an English Language Degree, located in the middle and higher levels, who were proposed to perform a identification of specialized terms —belonging to the areas of engineering and medicine— using Sketch Engine, in order to evaluate if any improvement is observed regarding their abilities to understand and translate these terms from English to Spanish. Our pilot study was divided in two phases: (a) in the first, they carried out such identification of terms without considering any corpus tool, with a view to determining their competence in English; (b) in the second, they were trained in the use of the Sketch Engine platform, which they used to automatically recognize these terms, as well as their meaning within their context of use.
Thus, the results of our pilot study allow us to reflect on the following points:
i.To establish the theoretical framework that defines the DDL as a linguistic learning model, to then explain its relationship with the use of electronic corpora.
ii.To recognize and analyze the use of computational tools in the process of acquiring terminology extracted from specialized corpora in English, specifically engineering and medicine.
Finally, based on the results obtained by our pilot study, propose a teaching methodology based on the use of linguistic corpus, supported by the defined theoretical framework to explain what DDL is.

Keywords:

Applied Linguistics, Language Learning, Corpus Linguistics, Data-Driven Learning, Term Extraction.