Abstract

Thanks to the use of computational means, there are different lines of research nowadays in the field of the so called corpus linguistics, which describe authentic texts from linguistic features. The criticism that has been made to such studies is founded on the fact that they only reach the word level with scarce impact on the textual level. In this work, we critically revised different feature matrices and methodologies that have been used for studying Spanish and English, in order to propose a matrix that can deal with part of the criticism made to previous researches. Concretely, the proposal is based on the following components: the organization modes of discourse, the sequences or lexical bundles, the Model of Event indexing and the Appraisal Theory. In addition, a component associated with grammatical categories, a component of syntactic units, and a component of statistics measures will be included. Finally we will also consider a component derived from a previous project (the dimensions established in the project Fondecyt 1020786). In order to verify its descriptive potential, the matrix, based on these components, is applied in an exploratory way to a small set of texts.