CLOZE TEST ON RUSSIAN TEXTS

 

In the spring of 1995 a linguistic experiment, a so-called "cloze test", was carried out in Moscow, Arkhangelsk and Severodvinsk. Native speakers of Russian, mostly students and teachers, were presented with short literary texts, in which every fifth word had been excluded. The task consisted in guessing what word had been excluded in each slot. As a result a large number of guesses were obtained, and the author's aim has been to compare the proposed variants, including the original word, of every slot, thus trying to describe the possible relationships from a purely linguistic point of view.

The relationship holding between two guesses is characterised according to three criteria: a) whether the words have the same syntactic function or not; b) whether the words belong to the same part of speech (word-class) or not; and c) whether the words are compatible in the same slot or not (in the case of compatibility there may by restrictions on word order).

The application of these three binary criteria gives eight groups, or types of relationship, all of which turn out to be represented in the material. The presentation of each type consists of two parts: a number of examples and commentaries containing additional grammatical and semantic analysis.

Publication: O gibkosti russkogo teksta. Opyt lingvisticheskogo eksperimenta. Nordlyd 25, 1997, pp. 36-46.