This paper is concerned with the design and compilation of an electronic textual corpus. The criteria for the selection of the texts to be included in the corpus, as well as the principles of compilation and modes of use in Theoretical, Applied and Computational Linguistics and Lexicography are presented. Emphasis is given to the structural and linguistic annotation of the texts according to a set of features representing the levels of linguistic analysis (phonology, morphology, syntax and semantics). Exemplifying cases taken from the project on a Modern Greek Reference Corpus undertaken by ILSP (Institute for Language and Speech Processing) illustrate the theoretical issues visited.