This data was gathered as part of a collaborative research project between researchers at Griffith University, Australia (Chiharu Tsurutani) and Nanjing Normal University, China (Wentao Gu, Ting Zhang, Ping Tang, Lei Liu and Liangping Zhong). The researchers investigated segmental and tonal errors made by Australian learners of L2 Mandarin. The study and its supporting data may be of interest to speech scientists and teachers of Mandarin as frequently observed errors can be treated purposefully in research and teaching. They might also assist with the development of computer-assisted language learning programs for L2 Mandarin. L2 Mandarin speeches by the Australian beginner-level learners were collected on the basis of a phonetically balanced corpus. File 1 contains 22 sentences spoken by Australian learners of Chinese and their annotation. File 2 was compiled in attempt to identify prosodic differences between neutral and polite speech. Three files, 1) Utterances by Chinese native speakers (Text grid and sound files), 2) Utterances by Australian learners of Chinese (sound files) 3) Text grid files of 2) are placed in File 2.