Chapter 2 Literature review
2.4 Syllable acquisition in cross-language studies
2.4.4 Beijing Mandarin
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
frequency in the domain of syllable types of children aged 1;2-4;4 acquiring Taiwan Southern Min. The study was based on the longitudinal data from Taiwan Child Language Corpus (Tsay, in preparation). The results showed that CV was the most
frequently used syllable, followed by CVC, CVV, and V. More than 82% to 86% of children’s speech was these four syllable types. CV syllable is the core syllable and is
considered as the most unmarked syllable across languages. The findings showed that the most unmarked syllable type CV was the most frequent syllable in children acquiring Taiwan Southern Min. However, frequency did not always have a positive
correlation with accuracy. For example, CVC was the second most frequently used syllable in children’s speech; however, the error rate of children producing CVC
syllable was 98%. The most common-error in syllable structure involved coda dropping, which was a regular type in phonological acquisition (So & Dodd, 1995) and as mentioned by McCarthy and Prince (1994), children’s early productions were governed by highly-ranked No-Coda constraints, which predicts that CV syllable types appear to be the most common output of syllable errors.
2.4.4 Beijing Mandarin
Zhu and Dodd (2000) studied the phonological acquisition of Putonghua and found that the children’s errors suggested that Putonghua-speaking children mastered
syllable elements in the following order: tone was acquired first; and followed by
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
syllable-final consonants and vowels; and syllable-initial consonants were mastered
last. In the study, vowels emerged early in the development. Both syllable-final nasals [n, ŋ] appeared in the children’s inventory at their 1;6, while the syllable-initial
consonants was completed by 3; 6 for 75% of children. It is proposed that the saliency of the components in the language system determines the order of acquisition. Tone is more salient than the three other syllable components, so it is acquired by children earlier. Since syllable-initial consonants are optional, they have the lowest saliency of the four syllable components. So, children would master syllable-initial consonant later.
In sum, at children’s early phonological developmental stage, the unit of lexical representation is the syllable or the whole word rather than the phonemes. Inter-and intra-word phonological variability of children serves as evidence of this statement.
This study will then focus on intra-word variability produced by children acquiring Taiwan Mandarin. Two aspects are including in the following: the phonological variability of words and the syllable types composed the words. A longitudinal observation was conducted. The overall variability pattern, the frequency and variability rate of syllable type, and the substation pattern of syllable type will be discussed.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
31
Chapter 3 Methodology
There are two parts in this methodology section. The first part includes the data collection, and the second part contains the data analysis. The data have been collected by the author and the research team in the Phonetics and Psycholinguistics Lab at National Chengchi University for many years. The whole study has been sponsored by the NSC research projects, “Consonant Acquisition in Taiwan Mandarin (NSC 100-2410-H-004-187- )” and “Consonant acquisition in Taiwan Mandarin:
Evidence from longitudinal and experimental studies (NSC 101-2410-H-004-182- )”, both investigated by Professor I-Ping Wan.
Section 3.1 involves how I recruited the participants and their background information. Furthermore, the procedure and the recording equipments used during the data collection would also be detailed in this section.
For data analysis in section 3.2, I would present the methods of data transcription, the criteria for choosing target words, the formulas used in obtaining syllable type frequency and variability rate, and how the substitution pattern of different syllable types has been organized.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
3.1 Data collection
The participated families were recruited from an advertisement posted on a non-profit parent forum called Babyhome (http://www.babyhome.com.tw/). An article was posted on the forum explaining the academic research purpose, the information of the NSC research project, and the age of recruiting children. Parents who wanted to
participant in the research filled out the registration form we designed on “Google doc spread sheet,” which is an online questionnaire and can be customized in several ways.
Sixteen families were enrolled under the study.
Some of the children in the NSC research project lived with their grandparents who spoke Taiwan Southern Min, so the children might produced Taiwan Southern Min during the observation. Furthermore, some of the parents used English to communicate with their children, so the children might also produced English sometimes. In order to rule out the influence of languages other than Mandarin Chinese, these children would not be included in this study. At the end, only six children fit in this study, from which I collected four children, and the other two were collected by the research team. There were in a total number of 5868 tokens produced by six participants, among which 2088 tokens were transcribed by the assistants in the research team.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
3.1.1 Participants
The background of the six children enrolled in this study shared several similarities. All of them were the only child in their family. They only lived with their parents, and they were all taken care by their mothers in the day time. All mothers used Mandarin Chinese to communicate with their children, so these children‟s first language was Taiwan Mandarin.
All of them were from middle class family in Taipei. Two of them were boys and
the other four were girls. Their ages were between 0; 11 to 1; 1 at the beginning of observation. Since every child‟s phonological development was inconsistent, I
selected the age when they were in non-reduplicated babbling stage and have already produced their first meaningful word. The observation continued for twelve months.
At the end, the participants‟ ages were between 1; 10 to 2; 0. All participants were healthy and appeared to have normal hearing, as determined through parental
interviews and observation of children during data collection.
The participants‟ background information is presented below.
Table 3.1 Participants‟ age and recording duration
Participants Gender Age range Duration
#1 M 1;1-2;0 12 months
#2 F 1;0-1;11 12 months
#3 F 0;11-1;10 12 months
#4 M 0;11-1;10 12 months
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
#5 F 0;11-1;10 12 months
#6 F 1;1-2;0 12 months
3.1.2 Procedures
The data collection started from December 2011 to August 2013. There were eight research assistants in the research team. Every other week, two assistants were sent to a child‟s house in order to record the spontaneous speech between the child and the mother.
On average, the recording was about sixty minutes for one time. Sometimes the recording time might be shorter if the children were tired, hungry, or cried. The activity during recording was not limited. It could be share-book reading, eating, or playing with toys. During the recording, in order to create a more natural context, the mothers were encouraged to play with their children just like the way they always did
when they were home by themselves. Sometimes the assistants would also interact with the children, using toys to attract children‟s attention and encourage them to talk.
Children played with their mothers for the most of the time, since they were the people children were more familiar with. According to Lewedage et al. (1994), children produce well-formed syllables more frequently in home environment when familiar adults are present than in lab settings.
No specific or systematic planning of elicitation was done during the recording, except for natural elicitations in daily life. For example, when the mother and the
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
child were doing share-book reading activities, sometimes mother would ask the child to name pictures. The goal was to ensure that the data of this study was elicited from spontaneous speech, a method which increases the chances for variability since it involves the planning of longer utterances, including the need for syntactic planning and communicative intent (Dodd et al., 1989). The target words for analysis were selected in these spontaneous speeches. Therefore, the number and type of errors might be larger than a carefully controlled experimental task. However, it could reflect processes of phonological acquisition that occur in a more natural context. As
Ingram (2011) have claimed that syllables are best studied from words taken from a spontaneous sample because they more directly reflect a child‟s preferred usage.
Children‟s vocalizations were audio recorded during observations of their natural daily activities in their homes. One of the assistants held the video recorder and the
other held the sound recorder during the recording. The assistant who held the video recorder had to make sure to film the children‟s face, mouth, and the objects they
played with. The assistant who held the sound recorder had to stay near to the children.
The participants were all paid volunteers and have had signed the human subject consent forms. At the end of the research project, the families would receive an album of video recordings as a souvenir. The rewards and cost were supported by the NSC
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
research projects (NSC 100-2410-H-004-187- and NSC 101-2410-H-004-182- ).
3.1.3 Recording equipments
Video-recording and sound-recording equipments were both used in this study.
Sony DCR-SR40 Handycam digital video camera recorder and the Sony ICD- UX513F digital voice recorder were used during the recording. The sizes of these equipments were small, so it is easy to carry. The equipments were provided by the
lab and also sponsored by the two NSC research projects..
The video camera helped us record children‟s gestures, lip movement and things
they played with. The video could provide us some clues to decode the referential meaning of children‟s utterance. The sound recorder provided us high quality sound
files which could help us distinguish the sounds children uttered.
3.2 Data analysis
The participants in the observation period were around one-word stage, but their utterances were longer than two syllables, so the study will analyze children‟s monosyllabic and disyllabic words separately in different sections. The following section includes how the data have been transcribed, coded, classified and analyzed.
3.2.1 Transcription and coding
The data from the recording were transcribed by the author and the assistants of the research team. If there were disagreements, the tokens would be discussed or
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
checked by another team member. All of the assistants are native speakers of Taiwan Mandarin and have good training in transcribing children‟s speech.
Inter- rater reliability and intra-rater reliability were assessed for the identification of participants‟ consonant and vowel productions in IPA broad transcription. The inter-transcriber and intra-transcriber reliability of the transcription reached a percentage higher than 90% under the study, as shown in Table 3.2
Table 3.2 Inter- and intra- transcriber reliability Inter-transcriber
(2088 tokens)
Intra-transcriber ( 3780 tokens)
Consonant 97.5% 99.4%
Vowel 91.6% 94.1%
The current study did not deal with the acquisition of segments, so only broad transcription applied to all words and phrases children produced during the recording.
Transcriptions included the standard set of symbols used for adult Mandarin phonemes.
The utterances of words and phrases would be transcribed into four parts: actual produced words in IPA transcription, tone, possible meaning and number of occurrences. The transcribed examples are shown below in Table 3.3.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 3.3 The sample of coding
IPA transcription Tone Possible meaning Occurrence
[tshatɕhi] [55-] fork 3
[tshatɕi] [55-] fork 6
[pijɑw] [35-51] no 7
[tatɕi] [51-] here 3
[tɕhjotɕhjo] [21-35] ball 3
The first column represented children‟s actual produced words which were transcribed with IPA symbols. The second column marked the tone of produced words.
The tones were coded with [55], [35], [21], [51], representing level tone, rising tone, falling rising tone, and falling tone respectively. The neutral tone was coded without any tone number. The third column marked the possible meanings of each word
children produced which could be inferred by contexts, children‟s gestures, or the repetition of adult‟s speech. If the utterance was meaningless or unable to infer from
the context, we would leave this column blank. The meaningless token would not be included in the study. The last part was the number of occurrence of each word. For example, in the first and second row of the sample, this child produced „fork‟ as [tsha55tɕhi] for 3 times and as [tsha55tɕi] for 6 times.
3.2.2 Criteria for target words
The following are some criteria for choosing target words that children produced for analysis. The criteria were adopted from Sosa and Stoel-Gammon (2012). Firstly,
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
the sound quality of words must be fair and clear. Whispered speech and overlapping speech of adults would be excluded. Background noise from toys and rustling noise from contact with the sound recorder resulted in blurred and fuzzy sound would also be excluded. Secondly, the meaning of the words must be clear. Words that would only be considered for analysis if a Mandarin gloss could be identified, or if the meaning of words could be inferred by careful examination from the context as well as the reaction or repetition of adults‟ speech. For example, if a child pointed at a toy car and uttered [tɤ55 tɤ55], we would suggest that its intended meaning is „a car‟.
Thirdly, target words with fewer than three useable tokens, although initially transcribed, would not be included in the final analysis.
Imitated words would be included in this study, as is often done in this type of study (Ferguson & Farwell, 1975; Macrae, 2013; McLeod & Hewett, 2008). Ferguson and Farwell (1975) argued that a high percentage of what young children say is imitated and children can imitate words spoken by adults with a considerable separation in time, so it is difficult to exclude imitation from analysis. Imitated words were defined as productions that occurred within 2 seconds of immediately preceding adult utterances that contained the same target words (Sosa & Stoel-Gammon, 2012).
Since the study focuses on children acquiring Taiwan Mandarin, English and Taiwanese produced by children, although meaningful, would not be included in the
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
study.
3.2.3 Variability rate
To answer the research question concerning variability rate, the analysis of production variability rate would be undertaken.
Phonological variability refers to the different realizations for multiple productions of a same lexical item produced by children. Whether productions of an individual form would be considered the same or different would be determined by the phonetic transcription of the word. Words that displayed consonant, vowel, or overall syllable structure differences would be considered as different phonetic forms, but tone difference is not included. Demuth (1993) mentioned that every mother‟s motherese would be slightly different since the way mothers adopted to simplify and reduplicate words would have individual style, especially in the acquisition of tone.
For example, care-taker may use [pa21pa35] or [pa51pa] to refer to „father‟.
Children‟s production of tone is correct or not would depend on their different motherese. Therefore, tone difference in the variable repeated form of the same word would not be included in the calculation of variability rate of individual word. The sample of phonological variability is shown in Table 3.4
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 3.4 Phonetic forms and tokens of the word [khaj55]
Target word IPA description Tokens
khaj 55 „open‟ tha55 1
taj55/taj51 3
tej55 1
khaj55 8
4 distinct forms 13 tokens
Table 3.4 shows the multiple production of the same word [khaj55] „open‟.
Although [taj55] and [taj51] have different tones, they were still counted as the same form in this study. However, [taj55] and [tej55] would be counted as different forms since their vowels were different.
Variability rate would be measured by the percentage of word variability which is calculated for each child at each age. Measures of overall variability for each child at each age, variability of each word as well as variability of each syllable type would be calculated. The method was adopted from Ingram‟s (2002) Proportion of
Whole-Word Variation measure.
Variability for individual words would be calculated by dividing the number of different phonetic forms of each word by the total number of productions of that same word during the same age. The number was then multiplied by 100 to determine the percentage of variability rate. The formula is presented below.
Different phonetic forms (forms)
Variability rate of word = X100%
Total productions (tokens)
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
In table 3.4, the child used 4 different phonetic forms for 13 production of the word „open‟, so the variability rate for the word „open‟ was 30.7% (4/13x100). Since
the term in the numerator reflects the number of different phonetic forms of a word, it can never be equal to one; there are either zero different form (complete consistency),
or two different forms or more. So, if the child used only one phonetic form for 13 productions of the word „open‟, then the number of different forms was zero. The
percentage of variability rate for that word would be 0% (0/10x100). Thus, variability rates may range from 0% to 100%.
Variability for individual syllable type would also calculated by dividing the number of different syllable types of target syllable type by the total number of
production of the target syllable type. The formula is presented below.
Different syllable type
Variability rate of syllable type = X100%
Total productions of target type
Overall variability for each child at each age would be calculated by dividing the number of different phonetic forms by the total number of word tokens produced during the recording at that age. For example, 17 word tokens were included in the analysis for participant #1 at age 1;2, in which he used 11 different phonetic forms in producing these 17 tokens, resulting in an overall variability level of 64.7 % (11/17x100) at this age.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
The variability rate could help us understand how prevalent variability is in children‟s speech. By estimating the variability rate we could know whether a certain
type of syllable is mastered by the children or not.
3.2.4 Syllable type frequency
Evaluating the frequency of different syllable types in children‟s speech could reveal children‟s preference and show the mastery of each syllable types. If the
frequency of a certain syllable type is low, it might be explained that the syllable type is more problematic to children and has not been acquired yet. For example, if the frequency of CGVN (C = consonant, G = glide, V = vowel, N = nasal) was observed to be lower than other syllable types when children were one year old, it would imply that the children have not acquired syllable type CGVN yet and they might use other syllable type to replace it, such as CV or CGV, resulting in production variability and the low frequency of CGVN. So we could predict that syllable types with lower frequency would have higher variability rate.
The frequency of each syllable type children produced would be computed by the formula presents below.
Tokens of a syllable type
Percentage = X100%
Tokens of all syllable type
The numerators would be the tokens of a particular syllable type. The fraction would be presented as a percentage. The frequencies of all syllable types could be
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
ranked into an ordering. The ordering of syllables‟ frequencies could be used to
ranked into an ordering. The ordering of syllables‟ frequencies could be used to