CHAPTER 4 RESULTS
4.1 Experiment 1: one-segment disruption
results of experiment 3 are demonstrated with brief discussions. The reasons why the
results occurred are also explained.
4.1 Experiment 1: one-segment disruption
(Position=the position replaced by the hiccup noise RT=the average reaction time
Acc=accuracy
1C=the initial consonant in the first syllable replaced by the hiccup noise 1Pre=the prenuclear glide in the first syllable replaced by the hiccup noise 1V=the vowel in the first syllable replaced by the hiccup noise
1Po=the postnuclear glide in the first syllable replaced by the hiccup noise 1N=the final nasal in the first syllable replaced by the hiccup noise
2C=the initial consonant I in the second syllable replaced by the hiccup noise 2Pre=the prenuclear glide in the second syllable replaced by the hiccup noise 2V=the vowel in the second syllable replaced by the hiccup noise
2Po=the postnuclear glide in the second syllable replaced by the hiccup noise
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 1 shows the results of experiment 1 (high-frequency words). According to
the table, the first row designates the position replaced by the hiccup noise. The first
column illustrates the reaction time (written in millisecond), the number of the test
items which are successfully recognized by the subjects (Pass), the number of the test
items which weren’t recognized by the subjects (Fail), the total number of the test
items (Total), the rate of the test items which can be correctly recognized (Acc%), and
the number of the invalid responses (Invalid). For instance, the number situated in the
second row and the second column is 596. This means that in average subjects need
596 milliseconds after the end of the targets to recognize the targets whose initial
segments of the first syllable are replaced by the hiccup noise. The data located in the
fourth row and the second column is 7. This means that there are 7 test items whose
initial segment of the first syllable is replaced by the hiccup noise not able to be
recognized or correctly recognized. In addition, the table displays that 1Po has the
longest reaction time. 1V and 2V have the second longest reaction time. 2N has the
shortest reaction time. Last but not least, the lowest accuracy of the test items is
84.44%, nestled in the 1V column.
In this table, it is obvious that the vowel in the first syllable is the most important
segment in the processing of spoken words. The words whose first vowel is replaced
by the hiccup noise need 659 milliseconds after the end of the stimuli to be
‧
hiccup noise is just slightly faster than that of the stimuli whose first postnuclear glide
is replaced by the hiccup noise (661 ms) and the same as that of the stimuli whose
second vowel is replaced by the hiccup noise (659 ms). Although the reaction time is
the second longest when the first vowel is replaced by the hiccup noise, there are 28
test items that cannot be correctly recognized by the subjects. The accuracy for 1V is
84.44%, which is much lower than the accuracy for 1Po (100%) and the accuracy for
2V (93.89%). Consequently, the first vowel in the disyllabic word is the most
important in spoken word recognition in Taiwan Mandarin.
Table2. Low frequency words: one-segment disruption
number of the test items which are successfully recognized by the subjects (pass), the
number of the test items which cannot be recognized by the subjects (fail), the total
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
number of the test items, the rate of the test items which can be correctly recognized,
and the number of the invalid responses. For instance, the data situated in the second
row and the second column is 632. This means that in average subjects need 632
milliseconds after the end of the targets to recognize the test items whose initial
segments of the first syllable are replaced by the hiccup noise. The data located in the
fourth row and the second column is 18. This means that there are 18 test items whose
initial segment of the first syllable is replaced by the hiccup noise not able to be
recognized or correctly recognized. In addition, the table displays that 1V has the
longest reaction time; 2N has the shortest reaction time. Last but not least, the lowest
rate of the unrecognizable test items is about 79%, nestled in the 1V and 2V columns.
The results of the low frequency words show the similar results as the high
frequency words, which indicates that the vowel in the first syllable is the most
important for spoken word recognition and the vowel in the second syllable is the
second crucial segment in the processing of Mandarin words. The results display that
the 1V stimuli need 715 milliseconds to be correctly recognized by the subjects,
which takes the longest reaction time. The 2V stimuli need 706 milliseconds to be
successfully recognized, which takes the second longest reaction time. The results
also illustrates that there are 37 test items (Accuracy: 79.21%) which cannot be
identified correctly because of the disruption of the first vowel and 36 test items
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
(Accuracy: 79.66%) which cannot be recognized successfully when the vowel in the
second syllable is corrupted. The longest reaction time and lowest accuracy for both
1V and 2V indicate that the first vowel and the second vowel are very important, so
subjects need more time to identify the word whose first and second vowel are
disruptive. Furthermore, similar to the results of high-frequency words, the results of
low-frequency words show that the segments in the first syllable are more important
(longer reaction time and lower accuracy) than their corresponding segments in the
second syllable. This finding suggests that the perceived order in time has some effect
on the spoken word recognition in Mandarin.
As for the frequency effect, the results depict that subjects have more difficulties
in recognizing the low frequency words. The results show that the disruptive
segments of the low-frequency words cause longer reaction time and lower accuracy
compared with their corresponding disruptive segments of the low-frequency words.
This result shows that frequency effect appears here. Therefore, it can be inferred that
frequency effect exists in spoken word recognition in Mandarin.
Table 3. Incorrect responses of tones (high-frequency words): one-segment disruption
Position 1C 1Pre 1V 1Po 1N 2C 2Pre 2V 2Po 2N
Fail 0 0 7 0 0 0 0 3 0 0
Total 3 0 13 0 4 1 1 5 0 0
Percentage 0 0 53.85 0 0 0 0 60 0 0
‧
Table 4. Incorrect responses of tones (low-frequency words): one-segment disruption
Position 1C 1Pre 1V 1Po 1N 2C 2Pre 2V 2Po 2N
Incorrect 0 0 4 0 1 0 0 2 0 0
Total 7 2 14 0 6 9 0 10 0 0
Percentage 0 0 28.57 0 16.67 0 0 20 0 0
Table 3 and Table 4 display the incorrect perception of tone among the incorrect
responses in experiment 1. The incorrect responses here mean that subjects did say a
word when they heard the particular stimulus, but the tone of the response to the
particular stimulus was wrong. From these two tables, we know that vowels carry
most tonal information in Mandarin, so when the vowels are replaced by the hiccup
noise, the percentages of the incorrect responses of tones are higher. It is also
noticeable that there is one misperception of tone of 1N. Although coda nasal dose not
occupy a long period of time in words, it still carries tonal information because it
belongs to rime. Therefore, tone can still be misperceived when the coda nasal is
replaced by the hiccup noise.