This dissertation investigates second language (L2) speech learning challenges by testing advanced L2 Mandarin Chinese learners’ tone and word knowledge. We consider L2 speech learning under the scope of three general hypotheses. (1) The Tone Perception Hypothesis: Tones may be difficult for L2 listeners to perceive auditorily. (2) The Tone Representation Hypothesis: Tones may be difficult for L2 listeners to represent effectively. (3) The Tone Processing Hypothesis: Tones may be difficult for L2 listeners to process efficiently.

Experiments 1 and 2 test tone perception and representation using tone identification tasks with monosyllabic and disyllabic stimuli with L1 and advanced L2 Mandarin listeners. Results suggest that both groups are highly accurate in identification of tones on isolated monosyllables; however, L2 learners have some difficulty in disyllabic contexts. This suggests that low-level auditory perception of tones presents L2 learners with persistent long-term challenges. Results also shed light on tone representations, showing that both L1 and L2 listeners are able to form abstract representations of third tone allotones.

Experiments 3 and 4 test tone representation and processing through the use of online (behavioral and ERP) and offline measures of tone word recognition. Offline results suggest weaknesses in L2 learners’ long-term memory of tones for specific vocabulary. However, even when we consider only trials for which learners had correct and confident explicit knowledge of tones and words, we still see significant differences in accuracy for rejection of tone compared to vowel nonwords in lexical recognition tasks. Using a lexical decision task, ERP measures in Experiment 3 reveal consistent L1 sensitivity to tones and vowels in isolated word recognition, and individual differences among L2 listeners. While some are sensitive to both tone and vowel mismatches, others are only sensitive to vowels or not at all. Experiment 4 utilized picture cues to test neural responses tied directly to tone and vowel mismatches. Results suggest strong L1 sensitivity to vowel mismatches. No other significant results were found.

The final chapter considers how the three hypotheses shed light on the results as a whole, and how they relate to the broader context of L2 speech learning.