Perform Context Independent Lexical Tonal Error Analysis

A Python script was written to generate the result of lexical tonal error produced by all the speakers in the corpus. The mispronunciation rates for each tone are listed in Figure 4. Among the 5 tones, Tone 3 is the most challenging for Mandarin learners (51.90% error rate), while Tone 1 and Tone 5 are the easiest (35.20% and 36.10% error rate, respectively). Among the Tone 3 errors, the majority (26.30%) are mispronounced as Tone 2. Over 15% of the Tone 2 and Tone 4 errors are mispronounced as Tone 1; 13.70% of Tone 1 errors and 11.50% of Tone 5 errors are mispronounced as Tone 4.

 Figure 4 Overall Tonal Error