Go to Usage Data

F13AH Usage Data

Usage data for a single subject trying out SpeakGoodChinese for an extended period.

Evaluate the tone productions of F13AH

The results are here

Subject:

• Female
• Native language: Dutch
• 13 yo
• Previous exposure: ~ 10 weekly lessons of 45 minutes at school in a choice subject introductory course program

Procedure

• Subject was paid for her work and knew about the project
• Short instruction on use of SpeakGoodChinese
• Every daily session starts with a Pretest and ends with a Posttest each about 30 words long (shuffled lists)
• Both pre- and posttests were done without audio examples or playback, but with visual feedback
• Practise sessions were done on shuffled lists, preferably not including the words of the pre- and posttests (but this could not always be prevented, especially between sessions)
• Subject worked alone in quiet room.
• Sessions last for 3-4 hours where the subject spends up to an half hour on Chinese and an half hour on an unrelated task with a 10 minutes break. (subject did not follow this very precisely)
• In some sessions, there was no other task, but the subject was stimulated to take longer breaks.
• There were two kinds of sessions, on alternate days, practising either monosyllabic words or mixed mono- and bisyllabic words. The major difference was in the nature of the audio examples:
• Monosyllabic words with prerecorded real audio examples from the MandarinSounds.sgc wordlist
• Mixed mono- and bisyllabic words with hummed examples from the mixed mono- and bisyllabic wordlists

During session 1, F13AH AudioMonoSyllable 090707, the Audio I/O settings were wrong initially. The microphone boost was activated and the recording level wrong. This lead to high noise levels and blowing noises. This was reset at 12:35:11 (word xian2).

Results

There are three error conditions visible to the user, each with its own error message. In order of decreasing severity:

Low:

Top of intonation contour is more than 3 dB below the top of the model tone

Narrow:

The distance between the top and bottom of the intonation contour is more than 3 dB less than the corresponding difference in the tone model

Wrong:

The best fitting tone model is not the target tone combination (after post-DTW rules)

1. F13AH Audio MonoSyllables 090707

Result	Pretest	Posttest	Practise	Total
Wrong:	14	3	29	46
Low:	15	6	20	41
Narrow:	1	4	12	17
W+L+N:	21	10	45	76
Correct*:	9	21	91	121
Total:	30	31	136	197

2. F13AH Hummed Bisyllabic 100707

Result	Pretest	Posttest	Practise	Total
Wrong:	6	4	28	38
Low:	7	5	0	12
Narrow:	9	6	12	27
W+L+N:	16	11	34	61
Correct*:	18	14	119	151
Total:	34	25	153	212

3. F13AH Audio MonoSyllables 110707

Result	Pretest	Posttest	Practise	Total
Wrong:	5	3	75	83
Low:	4	4	22	30
Narrow:	3	1	9	13
W+L+N:	10	7	101	118
Correct*:	24	26	288	338
Total:	34	33	389	456

4. F13AH Hummed Bisyllabic 100707

Result	Pretest	Posttest	Practise	Total
Wrong:	8	2	23	33
Low:	4	3	0	7
Narrow:	2	2	3	7
W+L+N:	13	6	23	42
Correct*:	27	25	73	125
Total:	40	31	96	167

5. F13AH Audio MonoSyllables 170707

Result	Pretest	Posttest	Practise	Total
Wrong:	4	8	27	39
Low:	3	3	2	8
Narrow:	1	1	3	5
W+L+N:	7	11	31	49
Correct*:	23	23	83	129
Total:	30	34	114	178

6. F13AH Hummed Bisyllabic 180707

Result	Pretest	Posttest	Practise	Total
Wrong:	12	3	31	46
Low:	0	0	0	0
Narrow:	2	0	1	3
W+L+N:	13	0	32	48
Correct*:	17	37	85	129
Total:	30	30	117	177

7. F13AH Audio MonoSyllables 190707

Result	Pretest	Posttest	Practise	Total
Wrong:	8	2	25	35
Low:	1	2	5	8
Narrow:	0	0	0	0
W+L+N:	8	4	30	42
Correct*:	22	27	53	102
Total:	30	31	83	144

8. All sessions combined:
   Automatically determined error rate: 28% overall, 39% pretest and 24% posttest (p < 0.002, X² = 13.39, DoF = 2)

Result	Pretest	Posttest	Practise	Total
Wrong:	57	25	238	320
Low:	34	23	49	106
Narrow:	18	14	40	72
W+L+N:	88	52	296	436
Correct*:	140	163	792	1095
Total:	228	215	1088	1531

Recorded Data

1. F13AH Audio MonoSyllables 090707
The Audio I/O settings were wrong initially. The microphone boost was activated and the recording level wrong. This lead to high noise levels and blowing noises. This was reset at 12:35:11 (word xian2)
• Pretest
• Posttest
• Practise
2. F13AH Hummed Bisyllabic 100707
• Pretest
• Posttest
• Practise
3. F13AH Audio MonoSyllables 110707
• Pretest
• Posttest
• Practise
4. F13AH Hummed Bisyllabic 160707
• Pretest
• Posttest
• Practise
5. F13AH Audio MonoSyllables 170707
• Pretest
• Posttest
• Practise
6. F13AH Hummed Bisyllabic 180707
• Pretest
• Posttest
• Practise
7. F13AH Audio MonoSyllables 190707
• Pretest
• Posttest
• Practise

Questionnaire

1. Have you ever used a software program to practice your Chinese pronunciation?
   B. No, I have never used any.
2. What is your general impression of the application SpeakGoodChinese?
   C. Very useful, because you can easily learn Chinese tones with it
   D. The score I give to SpeakGoodChinese is (1- 10): 8.0
3. What is your opinion about the synthetical tone examples?
   C. Very useful, because you can hear from them how it has to be done
   D. The score I give to the synthetical tone example is (1- 10): 8
4. What is your opinion about the graphical pictures of the tone contour?
   B. Useful, because it gives a good impression how high (or low) one has to speak
5. Comparing to the program that you have used, SGC is:
   ---
6. If you get SGC for free, will you use it:
   B. No, because I do not follow a Chinese course. If I would follow a course, I would like to use it to practise
7. If SGC should be improved what are the aspects that you would like to change?
   A. Recording should be shorter, eg, recording could stop after releasing the record button
8. Other comments:
   ---