표 3. | Table 3. 조음 정확성 평가를 위한 음향 자질별 최소변별치 기준 | Just noticeable difference (JND) thresholds for articulatory accuracy

Feature	JND-based threshold
Vowel duration (ms)	LM≥NM±20 ms
F1 (Hz)	|zLM–zNM|≥1.0 (i.e., |ΔzF1|≥1.0)
F2 (Hz)	|zLM–zNM|≥1.0 (i.e., |ΔzF2|≥1.0)
F3 (Hz)	|zLM–zNM|≥1.0 (i.e., |ΔzF3|≥1.0)
VSA (kHz2)	|ΔM / SDN|>1

LM, learner mean; NM, native speaker mean; zLM / zNM, z-score normalized values for LM / NM; ΔzF1, difference between LM and NM in z-score for F1 (same for F2, F3); SD^N, standard deviation of the native speaker group (N = native speaker).