The purpose of the present paper is to describe a system which automatically analyses pauses within the communication between the participants to a collaborative learning instant messenger (chat) conversation. Learners use chat as an assignment, for example, to have a debate in which each participant takes the role of a representative for one of the given subjects: "chat", "forum", "blog", "wiki". The utterances used inside this automatic analysis are question-answer type speech acts, and the pauses in collaborative learning chats are classified based on the used utterance types and conversation times. In addition, it is presented the link between the number of used utterances and the pause numbers, causing in this way, the implication of each participant. The last part of the paper identifies the link between the utterance type and the pause type used by each participant, based on the automatic analyses. An important characteristic of collaborative learning chats is highlighted by the overlap of the utterances in the same time. By using this characteristic we can say that more participants can issue more utterances in the same time. The participants exchange speech acts and, depending on everyone's involvement, there may be established types of interchangeable utterances and types of pauses for each participant. There are also implicit and explicit (stated by the participants) references in the conversation. The paper will determine several statistics like the number of words for each participant, the number of utterances, the number and types of pauses, time intervals between the utterances of each participant and between participants who have the same reference. At the end, we will present the results obtained automatically, a comparison between them and a parallel with the results analysed manually.

Computer-Supported Collaborative Learning, chat, utterances, pauses, learning analytics