Code-switching is one of the aspects to study the immediate results of language contacts which supposes that intra-sentential analysis of morpho-syntactic structures could explain the causes and reasons of language borrowings in synchronic linguistics. It also involves psychological and social aspects of human behaviour, but confines itself to linguistic analysis. At present three main theories which are language rooted and the most developed ones govern code-switching research, namely, linear order constraint model of S. Poplack et al.; linear order non-constraint model of N. Chomsky and E. Woldorf and Matrix Language Frame model elaborated by C. Myers-Scotton. All the three proceed from the idea of a monolingual grammar regulating code-switching process, though in Chomskyan model such monolingual grammar is substituted with the Universal Grammar in combination with generative grammar rules. The C. Myers-Scotton Frame model seems more complicated as it is organized as a double-headed instrument: dominant Matrix Language doesn’t eliminate the factor of Embedded Language which in course of their interaction creates a kind of collaboration. All the three models basically deal with a sentence structure and thus are in fact intra-sentential studies, though the conclusions are also made on the morphological and even phonological aspects of the utterance. Lexicon of languages in contact is practically neglected in the studies but N. Chomsky, and J. MacSwan and C. Myers-Scotton pay attention to lexical borrowings; they prove the importance of those by developing the notion of nonce borrowings specific representation of culture and tradition. In fact the present day state-of-the art in the domain of code-switching base itself upon observation and language samples accumulation in course of analysis of structural features of contacting languages. The shortcomings of the approach might be motivated by the fact that typological language features are not applied, so the analysis lacks a solid background bringing in random data without any classification possible. Thus one may assume that main results at a present level of research could be found in the sphere of methodology and methods. As to code-switching terminology it’s in course of development: even the notion of ‘code’ in our opinion needs more precise definition as it originally refers not to language itself but to semiotic systems of various origins and nature. The directions for future research do not confine themselves to bilingual or multilingual data, but mean to deepen the multidisciplinary studies of language contacts and verify the reasons for code-switching through the complex of humanities and natural sciences applied to linguistic items and codes.