Subject-object asymmetry in speech recognition

Abstract

The article gives a brief analytical review of deep structural syntax conceptions as ways to reveal deep syntax structures underlying the comprehension of information contained in the utterance. Syntax analysis of deep structures was originally treated as a set of instructions for utterance backward derivation and revealing the semantic content. But all theories under consideration which are based on Chomsky's and his followers' generative grammar - derivation theory of complexity, nucleus element projection theory - cannot stand experimental check and all the drawbacks are revealed by means of psycholinguistic experiment as an efficient way of studying the comprehension. It appears most obvious in the analysis of syntactically and semantically ambiguous utterances in languages with different structures. It is demonstrated by the examples of utterances with pause-filling dependencies and verbal nuclei with immanent amphiboly of transitivity/intransitivity in German languages (Dutch, German and English). The proposed integral model does not fully underestimate the accomplishments of universal grammar and other theories of communicative syntax. It rather aims to unite and develop universal characteristics in studying deep syntax structures by means of expanding its categorical set and the methodology of analysis through inclusion of philological phenomenological hermeneutics categories and transformation derivation models structured by metameans of comprehension process. The main components of integral "lexical expectation model" for deep syntax structure analysis are not only metaschemes, metameans and transformation principles, but also special prognostic strategies based on "lexical preference principle". It should be admitted that though integral "lexical expectation model" provides an opportunity to establish a precise deep hierarchical structure, its universal character can be observed only as regards language systems in which the syntactic structure can be presented according to the principle of nucleus element projection. The model requires further verification by empirical material of languages with a different morphological system which will be the object of the authors' further research