Back to list

Detail of contribution

Auteur: Mikhail DYMARSKIY

Titre:
Syntax Modeling and Process of Generating an Utterance


Abstract/Résumé: From the point of view of the most noticeable trends in the theoretical syntax, the second half of the last century could have been called "Age of Modeling", so that now there is no lack in various versions of syntactic models (in particular: the Generative Grammar dependency trees, schemes by Igor A. Mel'čuk et al., sets of models by Natalia Yu. Shvedova, Prague School, and many others). Nevertheless, there hardly exists a specialist in this field who could definitely state that a system of syntactic patterns is completely elaborated which explains satisfactorily how, in what form the syntactic system is stored in the human brain and how it functions in the process of a spontaneous generation of an utterance. The reason for this state of affairs seems to be simple: the language syntactic system and an individual syntactic system of a person using this language are principally different essences. The first one is nothing but a construct of the highest level of abstraction, the fruit of the analytic-synthetic efforts of several generations of scientists who sought above all to build an adequate description of the language system. On the contrary, the second one is no virtual product but a really existing neurolinguistic mechanism which in common case works almost irreproachably. What is still more important, the latter requires no special knowledge about the previous in its owner's thesaurus. There are absolutely no reasons to suppose that the second is based on the first, and, consequently, to imagine the process of generating an utterance as a sequential scan of basic syntactic models followed by searching for appropriate words to be put in each slot etc., which requires hundreds of thousands mental operations per second. Despite its popularity, this supposition is extremely unlikely, and not because of the brain's inability to execute momentarily that amount of operations, but because of inefficiency of this way. A theoretical draft of a three-level system of syntactic models (following but not copying F.Daneš's idea), which includes not only core patterns but also speech models and utterance models, is to be presented in the proposed report. This system seems to make it possible to give more sensible answers to the questions stated above, since it outlines a much more efficient scheme of the process of generating an utterance.