Chapter 2 Literature review
2.5 Tone acquisition on Mandarin
2.5.3 Zhu (2002)
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
33
degree of development in the process of tone acquisition was not documented.
2.5.3 Zhu (2002)
A more recent work related to the tone acquisition in Mandarin set more specific criterion on stabilization of the acquired tones. Zhu (2002) conducted a longitudinal study in Beijing on four Mandarin-speaking children aged 0;10 to 1;2 in the beginning and 1;8 to 2;0 at the end. She provided the age of tone emergence and age of stabilization in each subject. The order of tone emergence was similar to that in tone stabilization. The criterion for deciding the tone emergence and stabilization was clearly cited in this study, and the tonal error patterns were presented in specific number of frequencies. The results showed that the high-level tone [55] was firstly emerged and stabilized. The second one was the falling tone [51], and rising [35] and falling-rising tones [214] were the last.
When tonal errors occurred, the most frequent tone that realized to replace the error was the high-level tone, and high-level tone seemed to be replaced by the falling tone when produced wrongly. However, the subjects Zhu studied were from Beijing, and the Mandarin was different from that in Taiwan. It is valuable to see whether the development of tones would be different in children exposed to dialects in Taiwan.
2.5.4 Summary
Based on the three studies reviewed above, researchers agreed that the high-level [55]
and falling [51] acquired earlier than the falling-rising [214] and the rising tone [35]. But
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
34
the substitution pattern for whether [35] was easier to replace [214] or vice versa did not gain consensus. There was also no precise document describing the tonal acquisition process in developmental stages. Thus, the topic is worth for further research.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
35
Chapter 3
Methodology
The methodology would include two parts: one is the data collection, and the other is the data analysis. The data were collected by the author and the research team in the Phonetics and Psycholinguistics Lab at National Chengchi University under the NSC project, “Consonant Acquisition in Taiwan Mandarin,” investigated by Professor Wan I-Ping (NSC 100-2410-H-004-187-).
For data collection in section 3.1, I will introduce how I recruited the participated families in 3.1.1, what my subjects’ backgrounds were in 3.1.2, how the observation proceeded in 3.1.3, and what recording equipments were used in 3.1.4. For the data analysis in section 3.2, I will illustrate how I transcribed the data in 3.2.1. From 3.2.2 to 3.2.4, I will show how I arranged the data in order to obtain the result of the tone emergence ordering, frequency, accuracy rate, and the substitution pattern in tonal errors.
3.1 Data collection
This section contains the process of recruitment in 3.1.1, the background information of the informants in 3.1.2, the observational procedures in 3.1.3, and the recording equipments in 3.1.4.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
36
3.1.1 Recruitment
The participated families were recruited through an advertisement on a popular parent forum called Babyhome (http://www.babyhome.com.tw/), on the behalf of the NSC project investigated by Professor Wan I-Ping (NSC 100-2410-H-004-187-). In the non-profit advertisement forum, an article was pasted to declare the academic research purpose, and to ask for recruiting children aged from 0;8 to 1;0 who was at the beginning stage of their language development. Parents who wanted to join the research could sign up by filling out the online registration form designed by the “Google doc spread sheet,”
which could be customized by users. There were totally 16 families enrolled in the NSC project, but only 6 children fit in this study.
3.1.2 Subject
The six children were all from middle class families in Taipei City or New Taipei City. These families were all core families that the children only lived with their parents, and the informants were all taken cared by their mothers for the whole day. All Mothers used Mandarin Chinese to communicate with their children, so these children’s first language was determined to be Mandarin.
Among the 6 subjects, three of them were males and three were females. From the beginning of the observation, their ages were between 0;10 to 1;1 (mean age= 0;11.67, SD= 0.8 months). The observation continued for eight months. At the end of the
‧
0.05 months). The six subjects were all healthy and had not detected with any hearing or intellectual impairment.
The subjects’ language development was around the one-word stage that some of them had not produced any meaningful words yet, but some of them had already produced some meaningful words with clear lexical tones. Among these six children, three of them were the only child in their family, including subject #1, #3, and #4. The other three were the second child, including subject #2, #5, and #6 whose older siblings were all brothers, and the age gaps between the first and second children were smaller than four years old. The subjects’ background information is presented below.
Table 3.1 The data collecting information on subjects and recordings Subject Gender Age range Duration
#1 M 1;1-1;6 6 months
The data collection started from January 2012 to the present. There were over six research assistants in the research team, and the team sent two assistants to an informant’s house to record the spontaneous speech between the child and the mother every two weeks. The recording was about sixty minutes long for one time, but it might be shorter if
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
38
the children felt tired and started to cry. During the recording, one of the assistants was in charge of the video-taping who had to step aside, stabilize the camera, and make sure to film the child’s face and the object he/she was playing; the other assistant had to hold the digital sound recorder, stay near the child, and interact with the child. The research assistants would not bring any toy or reading-material to the family, but might use their own toys to ask questions or attract the child’s attention. Because three of my informants had older siblings, sometimes the older child would join the free-play while recording.
Although the older sibling usually performed better language competence than the younger one and would interfere the recording, some Mothers told us that the younger child uttered more words when playing with their older siblings.
The participated families were paid NT$80 per visit, and they could receive an album of their own video recordings as a souvenir at the end of the term in the research project. The rewards, equipments and cost were all supported by the NCS project (NSC 100-2410-H-004-187-).
3.1.4 Recording equipments
We used both video-recording and sound-recording equipments, which were Sony DCR-SR40 Handycam digital video camera recorder and the Sony ICD- UX513F digital voice recorder. The sizes of these equipments were both very small and functioned well.
Both equipment provided high-quality digital files. The video files helped us decode the
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
39
utterance meaning by children’s gestures and eye movements, and the sounds file provided us high quality audio signals.
3.2 Data analysis
The subjects in the observation period were around one-word stage, and most of their utterances were short within two syllables, so the study plans to analyze only monosyllabic and disyllabic words. Although children in this stage could hardly produce perfect consonants and vowels, their tones developed better and earlier (Lenneberg, 1967;
Kaplan, 1970; Demuth, 1996). To analyze the tones, the utterances with clear tones were all included. However, the referential meaning of children’s utterances would sometimes be unclear. No matter the utterances had clear meanings or not, once the tones were recognizable, we would include them. In 3.2.1, I will introduce how I transcribed the utterances into speech tokens. The method used to track the tone emergence ordering will be explained in 3.2.2, the formulas to calculate frequency and accuracy rate will be shown in 3.2.3, and how to arrange the substitution pattern in tonal error will be presented in 3.2.4.
3.2.1 Transcription and coding
Each recording file was transcribed by two assistants at the same time. If there were disagreements, the token would be discussed or checked by another research assistant in the team. Though young children’s utterances were sometimes fuzzy and hard to
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
40
categorize, Yang (2010) has tested Mandarin native speakers’ perception of tones and found that native speakers had the ability to perceive four tones by pitch contour and register. In order to show the reliability of our perception, we extracted some examples from the sound files that represented the recognizable and unrecognizable tokens respectively. The pitch contours of the exemplified tokens were presented by the computer phonetic software, Pratt, in fundamental frequency (F0). The recognizable tones were classified into different tone groups in Figure 3.1, and there are 4 examples with unrecognizable tones presented together without classification in Figure 3.2.
Recognizable tones
T1 [55] T2 [35] T3 [21] T4 [51] T0
700Hz
75Hz
[ja55] [ma35] [ma21] [tu51] [mə]
700Hz
75Hz
[tɕɤ 55] [jɛ 35] [tɕ jo21] [pa51] [tə]
Figure 3.1 The pitch contour of recognizable tones
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
41
Unrecognizable tones
700Hz
75Hz
Figure 3.2 The pitch contour of unrecognizable tones
The pitch contours in the two rows of Figure 3.1 were recognizable tokens and were categorized into each tone group. The high-level tones [55] in the two examples were both flat and high in frequency. The contours of the rising tones [35] were climbing and the durations were long. The low-level tone [21] in the first row exactly presented a low-level contour. The second example in [21] showed a low-level contour preceded by a high-level neighboring tone in the context, so it started from a vertical line and then leveled off. The pitch contours in falling tones [51] went downward clearly, and the neutral tones were rather short than other tones. The neutral tones were examined to be significantly shorter than other lexical tones in Mandarin (Chen & Xu 2006). Thus, the short contours suggested that our ability of detecting the neutral tones was precise. The examples in Figure 3.2 were unrecognizable tones produced rapidly or sloppily. The movements of these pitch contours seemed more uncertain, so they were excluded. All in all, the pitch contours in Figure 3.1 and Figure 3.2 have testified our perception that our coding of the Mandarin tones was reliable.
All speech tokens were classified into two groups. The tokens without clear semantic
‧
meanings were grouped together and were transcribed into segments and tones. The other group contained the tokens with clear meanings and they were transcribed into four parts:
segments, produced tones, word meanings, and target tones. The transcribed examples are presented below in Table 3.2.
Table 3.2 The sample of coding
Segment Produced Tone Meaning Target tone
withoutFirstly, the consonants and vowels were taken down by International Phonetic Alphabet (IPA). Secondly, the produced tones were coded with [55], [35], [21], [51], and the neutral tones were coded with a capital N. Thirdly, the intended meanings of speech tokens were also transcribed. Although children’s articulation was underdeveloped, their meanings of utterancea were sometimes recognizable. The referential meaning could be identified by context or children’s gestures. For example, if a child pointed at a ball and uttered [tjo21 tjo35], we would take down its intended meaning as ‘a ball’ instead of ‘to throw.’ Although the segments in this utterance sounded more like the verb ‘to throw’ in
Mandarin, it was more reliable to determine the target tones from context. If the utterance yields meaninglessness, or we could not recognize it by context, we would leave it blank.
The meaningless tokens could only apply frequency analysis but could not apply other measures because it does not have target tones. Fourthly, in order to determine the
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
43
correctness of tones, we had to decide what the target tones were in every meaningful token. Basically, the target tones were determined by their care-takers’ tone model in motherese. However, every mother’s motherese would be slightly different (Demuth 1993). For example, some care-takers may use [wa21-wa35] to refer to ‘a doll’ but others would use [wa35-wa55] to refer to the same thing. Therefore, we had to make sure which tone was applied by each care-taker when they used motherese to talk to their children, and then we could determine what the target tones were and whether their children’s tones were correct.
3.2.2 Tone emergence ordering
The tone emergence ordering is a common issue for studies concerning first language acquisition, and it is also the first step researchers could investigate in children’s tonal development. In this study, I also applied the age-tracked method to take down the occurrences of every tone by ages. The criterion for tone emergence was defined by Vihman (1996) that a tone which was produced more than once in meaningful words would be considered an emerged tone. To compare whether there were individual differences between the subjects, I took down their individual ordering separately. For example, if a child first uttered [ma55] ‘mother’ twice at 0;11, had [ta21] ‘to hit’ and [tu51] ‘rabbit’ at 1;0, and finally produced [je35] ‘grandpa’ at 1;3, then, his tone emergence ordering would be [55]>[21],[51]>[35]. With this ordering, researchers could
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
44
examine the universal rules cross-linguistically.
3.2.3 Frequency and accuracy rate of tones
The tone emergence ordering could only help us observe which tone appeared first, but it could not present the mastery of each tone. Analyzing the number of tokens and checking their correctness are the better ways to picture children’s developmental process.
The frequency could reveal children’s preference of tones, and the accuracy rate could show the stabilization of tones. Li and Thompson (1977) revealed that children would avoid producing words that contain tones that they have not mastered yet. If the frequency of a certain tone is low, it may be explained that the tone is more problematic to children and has not yet been acquired. For instance, if a child’s frequency of the rising tone [35] was observed to be much lower than other tones and was often replaced with other tones, it would indicate that the child had not yet acquired [35], and tended not to use this [35] when reproducing adults’ speech. The tone frequency could be used to examine whether the more frequent tone would be acquired earlier, and whether the least frequent tone would be more problematic and would be acquired last.
When calculating the frequency of tones, because it is to calculate the number of occurrences, the utterance meaning does not matter. Thus, we included both tokens that with and without clear meanings in frequency analysis. The frequency of every tone would be computed by the formula presented below.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
45
Frequency =
The frequency would be applied five times for calculating four lexical tones and a neutral tone. The denominator should be the total number of syllables regardless of meaningful or meaningless tokens. The numerators would be the number of tokens of a particular tone.
The fraction then should be presented as a percentage. The frequencies of all tones could be ranked into an ordering, too. The ordering of tone frequency could be used to compare to those in tone emergence and in accuracy rate.
Regarding the accuracy rate, it could provide a percentage to show the degree of stabilization. As mentioned above, the coding of each utterance included the word meaning and target tones. The purpose of transcribing the target tones was to determine the correctness of tones. Therefore, in the measurement of accuracy rate, we only included tokens with clear meanings. To judge the correctness of tones, we tended to use care-takers’ target tones as the criterion. If children produce identical tones with their care-takers’, we would count it as correct tones. For example, a mother used [pej55 pej55]
to indicate ‘a cup,’ and if the child imitates the tones [55-55] correctly, it would be counted as a correct token; if the child used [pej51 pej51] or other tone patterns which are not identical to the mom, then the tones would be classified into incorrect tokens.
The tone accuracy rate will be calculated by the formula which had been applied by the number of tokens of a tone
the number of tokens of all tones
x 100%
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
46
many researchers (see Shriberg & Kwiatkowski, 1982; Sheiberg, et al., 1997)
Accuracy rate=
When calculating the accuracy rate, the denominator should be the number of targeted tone that the children intended to produce. And the numerator would be the number of correct tones which are determined both by children’s referential meanings and care-takers’ target tones. For example, given that the reduplication form of ‘a pen’ the children learned is [pi35 pi21], when the child uttered [pi55 pi21], the tone of the first syllable would be considered a tone error. Hua & Dodd (1995) provided a criterion that a tone was viewed stable when two-third (66.7%) of the tones were produced correctly.
With this criterion, we could precisely examine whether a tone is acquired or not by checking the accuracy rate. Zhu (2002) applied the 66.7% criterion of stabilization from Hua & Dodd (1995), and also applied a 90% criterion of stabilization to measure the advanced level of tone stabilization.
3.2.4 Substitution pattern in tonal errors
Meaningful tokens which were determined to be a tonal error would be further analyzed in this measure. Unstable and immature tones would sometimes be changed into other tones, and would be determined as a tonal error. Although the accuracy rate could pretty much depict the states for children’s tonal development, it could not tell us how
the number of correct tokens of a tone the number of targeted tokens of a tone
x 100%
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
47
children deal with immature tones. Therefore, the substitution pattern was used to demonstrate which tones were more likely to be realized in replacing the error tones.
The substitution pattern in tonal errors would be presented in a matrix that the row represents the target tones, and the column represents the actual realized tones. The example of the matrix is presented in Table 3.3.
Table 3.3 A sample matrix of substitution pattern in tonal errors
Target toneRealized tone
[55] [35] [21] [51] Total
[55]
24 19 17 60 50.8%[35]
3 13 10 26 22%[21]
4 7 11 22 18.6%[51]
0 5 5 10 8%Total
7 36 37 38 1185.9% 30.5% 31.3% 32.2%
The matrix could present the number of tokens and the frequency of substitution in each tone. If a tone should be produced in [55] but is realized as [35], it will be put into the first column and the second row. The percentages on the right column would tell us which tone is more frequently used in replacing others in tonal errors, and the percentages on the bottom row tell us which tone tends to make more tonal errors. In this sample in Table 3.3, there are 3 tonal error tokens that replace [55] to [35], and there are totally 26 tonal errors that are realized into [35], accounting for 22 % of the total errors. There are only 7 tonal errors whose target tones are [55], accounting for 5.9% of the total errors. With this
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
48
matrix, the substitution pattern in tonal errors could be displayed clearly.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
49
Chapter 4
Results and Analysis
This section will present the results and general analysis of the data, and the analysis will follow the method introduced in chapter 3. Section 4.1 will be the overview of the overall data including monosyllabic and disyllabic tokens. The results regarding the age of tone emergence will be presented in 4.1.1, and the tone frequency and accuracy rate will be shown in graphs in 4.1.2. In section 4.2, to obtain more specific results, the monosyllabic and disyllabic tokens will be analyzed separately. The monosyllabic tokens will be analyzed in 4.2.1, and the disyllabic tokens will be analyzed in 4.2.2.
We found that the unusual high frequencies of [21] in the first syllable and [35] in the second syllable in disyllabic tokens might result from the tone combination [21-35]
related to the reduplication of motherese, so the results of reduplication in motherese will be displayed particularly in 4.2.3. After we determined to exclude tokens produced in [21-35], the reanalysis of the modified disyllabic data will be shown in 4.2.4, and the reanalysis of the modified data combining both monosyllabic and disyllabic tokens will be put in 4.2.5. Then in section 4.3, the substitution patterns of tonal errors will be illustrated. Last but not least, the age of tone emergence and stabilization will be