Analysing of human-human and human- chatbot dialogues in the context of the speech act theory

The aim of this study is to examine which speech acts and strategies are used in the chats with the human and chat robot via the virtual chat application. There have been different interpretations of human-human communication over the years. Speech act theory which is revealed by Austin(1962) removed speech from a mechanical process. According to Grice's Cooperative Principles and Maxims (1975) speech is proceeded within the framework of a certain cooperation and the rules of cooperation. Then, the Relevance Theory and the maxims of Grice were melted in the relevance pot and it was suggested that the basic rule of all speech was the principle of relativity (Sperber ve Wilson, 1995). Conversational analysis and verbal interaction were seen as a process built between participants and it was mentioned that the speakers used certain strategies during the conversation (Sacks, Schegloff ve Jefferson, 1974). In this study, the use of the rules developed on interpersonal verbal interaction in human-chat robot interaction will be examined. Virtual chat applications are types of computer-mediated communication. In virtual chat applications that allow Quasi-synchronous communication, interaction is realized by correspondence. Unlike oral communication, computer-mediated communication has its own rules and practices (Baron, 2004). It can be human-human interaction in virtual chat applications as well as chatbot wich is a kind of dialogue system. Although according to the technology they have their abilities are different , the chatbots are responsible for understanding the person they talk and producing the appropriate answer (Fraser, 1997). In the data collection process, 50 university students were contacted via the virtual chat application using the Wizard of Oz. Interactions were carried out by operators called the wizard, who were responsible for simulating the dialogue systems. The virtual chat dumps performed throughout the research are divided into two main groups according to the identity of the person to whom the conversation is being made and whether the conversation conforms to the maxim of relation as defined by Grice. Chat transcripts were analyzed by conversational analysis method. As a result of the analyzes, it was observed that the speech acts used by the participants were based on whether the chat is relevant or not. In cases where there is a violation of the relevance, it is revealed that the participants turn to strategies such as repair, correction, ignoring, changing the content of the message and forming new contexts in order to continue the chat. The lines of speech taken during the interaction were examined in terms of the Speech Act Theory put forward by Searle (1979) significant difference was reached in terms of the declaration and assertive. With the questionnaire distributed at the end of the application, the attitudes and behaviors of the participants about the chat were questioned. It was found out that the participants in the partially relevant group thought the person they talked as a chatbot and judged them as low-skilled.