This article reports on a study investigating the use of 2 synchronous computer-mediated communication tools: text and voice chat. The experimental design employed 3 groups (text, voice, and control), each consisting of 30 novice-level secondary school learners of English as a foreign language. Over a 4-week period, the participants in the experimental groups engaged in 40-45-minute-long chat sessions in dyads, guided by a total of 8 tasks. The data were collected through preanxiety and postanxiety scales and speaking tests, and the participants' perspectives were investigated through the use of closed and open-ended questionnaires. The results showed that the speaking proficiency of both experimental groups increased, whereas there was a decrease in the anxiety levels only for the text chat group. The results were then interpreted by taking into consideration the data from the questionnaires, and a protocol was proposed for the selection of tools for different learner groups.