The ongoing advancement of computer hardware and software technology over recent decades has attracted a lot of interest from foreign language educators and researchers. As new and improved technologies become available, educators are looking at a multitude of ways to leverage these technologies to enhance language learning within the limitations of the traditional classroom environment. Among the technologies of interest, virtual worlds, AI, and speech-to-text (STT) or automatic speech recognition (ASR) are three that have the potential to be combined to provide learners with situated learning in immersive simulations of real-life scenarios. This chapter reports on a pilot study involving 6 learners of Chinese as a foreign language (CFL) who while engaged in two lessons in the 3D virtual township of Chinese Island in Second Life interacted with non-player characters (NPCs) in both text-based chat and voice chat. The purpose of the study was to illicit feedback from participants on perceived or actual differences, challenges, benefits, or disadvantages of both forms of communication. Analysis of this feedback is intended to provide the basis for making necessary adjustments to lesson design and technology implementation for a future larger scale study comparing the relative efficacy of voice-based and text-based interaction with the Chinese Island NPCs. This chapter first gives some background on the technologies used and on the larger teaching and learning context in which they have been combined and utilised. Also discussed are the application of situated learning and simulation in a 3D virtual environment, as well as the key differences between chatbots and NPCs. Finally, some of the key findings from the pilot study are presented, which will serve as a reference for future study designs and pedagogical practice and design.