Central theme of this work would be the category of discourse markers, which were produced and analysed in the context of cyber communication. In order to adequately analyse them, it was necessary to rely on theorethical and methodological bricolage assembled of various pragmatic, sociolinguistic and semiotic approaches. Finally, category of discourse markers was analysed both from synchronic as well as diachronic perspective. Before we go into explanation of basic problems appearing in this work, here is a brief look into its structure. First chapter is introductory in its nature, here the reader is being introduced with the basic structure of the material, as well as main concerns and topics dealt throughout the dissertation. Second chapter aims to connect its central topic – the category of DMs, with the context of investigation – the cyber context. The following chaper (chaper 3) deals with Cyber pragmatics – a new field proposed by the authors, in order to provide an adequate theorethical and methodological frame for the analysis of cyber communication. Chapter 4 investigates main problems of discourse markers which appear in synchrony. Those problems are the following: terminological problems, definition problems and criterial problems (which in turn define the category). With the fifth chapter, the focus of investigation moves to diachrony. Specifically, to the area of historical pragmatics, gestural theories of language evolution and changes in language on a discourse level. Investigation of language change in discourse was central here and it was exemplified with formation and evolution of DMs. The following chapter (chapter 6) aims to investigate the influence of technology on communication and its efficacy. Here we have brought about the hypothesis verification, a short discussion on creativity and a list of principles of Cyber pragmatics as a new disciplinary field. Finally, seventh chapter summarizes the most important results and new findings obtained by the analysis, and it brings about some main difficulties and ideas for the future. Discourse markers (DMs) are defined here as a functional category of connective devices which can have various different roles in utterance interpretation. Although scholars do agree on some of their main characteristics, such as the cohesive force which enables them to tie together different sections of discourse, a consensus yet needs to be reached on how to delimit the very category in question. If syntax is considered, it's easy to see that the class of DMs is extremely heterogenous as it may contain: interjections, adverbs, conjuctions, adverbial phrases, verbs, sentence clauses and even entire sentences. That could be one possible source of the terminological mess occupying this area of inquiry, where discourse markers have also been referred to as: discourse particles, discourse operators, pragmatic markers, discourse connectives, connective phrases etc. For the reasons mentioned, the problem of defining discourse markers will be dealt with the help of functional criteria, since they are considered to be a functional pragmatic category. Nevertheless, there are several formal features of discourse markers which are being mentioned in various theories, and are imposing themselves as main and obligatory attributes of this category. Those are connectivity, syntactic optionality, non-truth-conditionality, utterance initiality and multicategoriality. Listed formal criteria have been shown to be useful for diachronic approaches to DMs, whereas the functional ones work better in synchrony. Next problem addressed here and tightly connected to the evolution of DMs would be the problem of their meaning. For the purposes of this work, we have taken two main types of meaning into consideration. First is the core meaning which represents the basic semantic content of a specific discourse marker and for which is also claimed to exist even in those DMs considered semantically empty. This concept of meaning has proved to be a useful tool for diachronic approches. Nevertheless, while discussing language change on a discourse level, we'll be approaching DMs from much broader perspective of historical pragmatics (not to be limited by syntactic and semantic constraints), giving the process of grammaticalization (Lehmann 1995) and pragmaticalization (Diewald 2011; Heine 2013) a valid account in DM evolution. In accordance with this, we’ll be dealing with yet another type of meaning, namely with procedural meaning. This type of meaning represents information encoded in linguistic constructions which tells us how to use conceptual representations in the phase of inferring and comprehension (Blakemore 2002). To be even more precise, procedural meaning is metacommunicative, metapragmatic, metadiscursive, instructional and it is a primary type of meaning for this category, as well as a key tool for synchronic and diachronic approaches of our topic. The context of this investigation also deserves a few explanatory words. Namely, Computermediated discourse is a type of communication which emerges when people interact by sending messages via networked computers (Herring, 2001: 612). Stated definition is satisfactory, but terminology leaves us with problems. Concepts such as informatic, informational, computer, computational, do not correspond to the content they should be referring to: some are too narrow while others are too broad in their meaning coverage. This is the main reason why we are opting for the prefix/adjective cyber while discussing online communication and discourse. Other reasons would also be of pragmatic nature: economy of the term (cyber discourse is much shorter than computer-mediated discourse), lexical productivity (the term is very prolific in the domain of new word formation, e.g. cyberspace, cyberfeminism, cybersociology etc.) and frequency in everyday language use (people use it more and understand it better). Cyber discourse would be a term with especially broad spectrum of meaning because it covers all types of communication on the Internet: from blogs, e-mails, through forum discussions, virtual reality interactions, to chats and instant messaging. Research dealing with cyber discourse is a part of a larger field known as Computer-mediated communication (CMC). Here, the focus is put on online language use by implementing methodology of discourse analysis, conversation analysis and interactional sociolinguistics. To complement that, we have added a functional approach to grammar (Functional Discurse Grammar) as well as some aspects of philosophy of language (Coyne 1998, Wittgenstein 1953, Vlăduţescu 2013) by which we define cyber pragmatics as a multidisciplinary research of online communication. This theoretical and methodological frame has been chosen because of the nature of the media of communication under scrutiny. To be more precise: the Internet is very fluid (prone to change, intertextuality, multimodality and open access), it doesn't hinder language use and it encourages langauge creativity. On the other hand its lack of paralanguage, its asinchronicity and spatial decontextualization (non-existence of a shared physical context between speakers) hinder coherence and cohesion, language economy and signalization of language functions. All those factors have created ideal surroundings for the developement and proliferation of discourse markers, as an inexhaustible source of original everyday laguage use which hasn't been subjected to alteration, niether for methodological nor linguopolitical reasons. Main goals of this research would be the following: a) delimiting the category of discourse markers, b) emphisizing the epistemological problems in the science of linguistics, c) indicating diachronic (evolutionary) processes which occur on a discourse level, d) accentuating the necessity for functional language theories in order to expand traditional grammar. Finally, the very analysis has been conducted on a private corpus holding 600 online conversations in a form of instant messaging (via Hangouts, Facebook messenger, Whatsapp applications) and on the examples chosen from a public Croatian Web Corpus (hrWaC). Methodologically and theoretically, the analysis has been supported by conversation analysis, pragmatics, language philosophy, discourse analysis and functional (discourse) grammar.