國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
(2) Regarding the syllable type: what is the general syllable type used among children? What is the most frequent syllable type used by the participants?
What is the rank-order of frequency in different syllable types? What are the variability rates of different syllable types? Do syllable types that have higher frequencies present lower variability rates? Is there a relationship between the frequency of syllable types and its variability rate?
(3) Regarding substitution pattern: are there any syllable substitution patterns in children’s production of variable forms of same words? Which syllable types
are more unstable and are replaced by other types more often? Which syllable types are more likely to be chosen to replace the unstable ones? What kind of strategy would participants use to replace the syllable types they are not mastered yet? Would they replace the syllable types they are not mastered with more frequent ones? Are there obvious individual differences among children's substitution strategies?
1.4 Organization of the thesis
The thesis is organized as follows. Chapter one has laid out the introduction to the present study and the questions pertinent to the data analyses. In Chapter 2, firstly I will introduce the development speech production in first-language acquisition in 2.1, and the factors of phonological variability will be introduced in 2.2. Introduction to
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Mandarin syllable will be introduced in section 2.3. Syllable acquisition in cross-language studies will be presented in 2.4. Chapter 3 contains the methodology of this study. Section 3.1 is the data collection methods. How I obtained the speech tokens will be introduced in this section. Section 3.2 is the data analysis which explains how the data were arranged. Chapter 4 will present the results and analysis in tables and graphs. The overview of the overall data will be presented in section 4.1.
Section 4.2 will present the pattern of phonological variability pattern of each participant. Section 4.3 will present the results of syllable analysis, including the general syllable type participants produced, the frequency of syllable types in different syllable positions, and the variability rate of different syllable types. Section 4.4 will present the substitution pattern between different syllable types. The relationship between frequencies of syllable types and its variability rates as well as the relationship between frequencies and substitution pattern will be summarized in section 4.5. The discussion and explanation is provided in chapter 5. Section 5.1 summarizes the findings in chapter 4. Section 5.2 presents the discussion on variability pattern. Section 5.3 presents the discussion on syllable type analysis.
Section 5.4 will be the concluding remarks.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
11
Chapter 2
Literature review
In this section, I will introduce the universal process of children’s speech
production in 2.1. Second, I will introduce the factors of production variability of children in 2.2. Introduction to Mandarin syllable will be introduced in 2.3. Syllable acquisition in cross-language studies will be presented in 2.4.
2.1 The development of speech production
Children’s productions of sounds begin with simple cries at birth, and they progress through several stages until they can produce complex babbling and adult-like intonation patterns. Many researchers proposed that children across the
world acquire different languages by similar steps (Lenneberg, 1967; Kaplan &
Kaplan, 1971; Stark, 1980). The six stages of children’s early vocalization are shown
in the following (Stark, 1980):
Stage 1: Reflexive vocalizations (0; 0-0; 2) -Most of the production are crying, fussing sounds, and vegetative sounds like coughing and sneezing. Some vowel-like sounds may occur.
Stage 2: Cooing and laughter (0; 2-0; 4) - Infants interact with adults or older kids by using cooing sound and laughter.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Stage 3: Vocal play (0; 4-0; 6) - Infants begin to test their articulatory organs and use them to produce sounds.
Stage 4: Reduplicated babbling (0; 6 and older) -The sequences of consonant-vowel (CV) syllables and adult-like intonation begin to appear at this stage.
Stage 5: Non-reduplicated babbling (0; 10 and older) - Strings of sounds and syllables uttered with a rich variety of stress and intonation patterns are appeared. In addition to consonant and vowel sequence, other syllable types also appear such as consonant-vowel-consonant (CVC) syllable.
Stage 6: Single word production - Infants begin to produce protowords, and
words used as symbols and refer to recurring objects or events.
Although researchers commonly refer children’s different vocalization period as
‘stages’, these vocalization types typically overlap from one stage to another. For
example, consonant and vowel sequence has become a unit in the speech production in stage 4. The production of CV sequence may appear in the previous stages such as in vocal play stage; the production of CV sequence may also continue to appear into stage 5 (Stark, 1980).
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
2.1.1 Whole-word representation
The following section will present the whole-word representation and its measurement.
2.1.1.1 Ferguson and Farwell (1975)
Ferguson and Farwell (1975) proposed that the minimal unit of lexical representation at early developmental stages is the whole word or syllable rather than the segments or the phonemes. In other words, young children are able to be aware of
relatively large phonological units, such as syllables or words, at early stages of phonological development. Ferguson and Farwell proposed this ‘whole-word’ system
of phonological representation based on two observations. The first is the variability of individual phonemes in different contexts. For example, a child may have produced initial [b] correctly in a specific set of words; however, when the initial [b] occurs in contexts other than the specific set, the child may produce the initial [b] in many different forms. This pattern of production suggests that the child has not mastered /b/
as an individual phoneme, but rather only mastered [b] when it occurs in specific words.
The second evidence supporting the whole-word representation is the prosodic variability (Ferguson, 1986). Ferguson and Farwell described a young girl who used ten different pronunciations of the word ‘pen’ in a one half-hour session. They
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
suggested that this variation, in which multiple tokens of the same word are produced differently at the same point of time, may be referred to as intra-word variability.
Aslin and Smith (1988) also supported the whole-word representation system of young children. They proposed that young children’s representation of lexicon is
holistic in nature. Only later can they analyze a string of new sounds based on
phonemic units. For example, in early phonological process, the representation of [dɔg] ‘dog’ is not organized as a sequence like [d] + [ɔ] + [g]. Instead, words are
represented as overall acoustic shape. The common assumption regarding the development of phonemic categories is that as the amount of vocabulary in children’s
lexicon grows, there is a need to discriminate the speech sounds efficiently in
production and also in perception. As the result, the growth of children’s lexicon leads to the development of phonemic representation. Therefore, children’s phonological
development can be viewed as a gradual process from a more holistic level to a more segmental level; that is, from whole-word representation to phonemic representation (Nittrouer et al., 1989; Waterson, 1971; Walley, 1993). When, precisely, the transition begins is not known, but assumption has been made that phonemic representation begins to emerge when a child has between 50 and 100 words, and the process may not be complete until much later in childhood, perhaps as late as 7;0 or 8;0 (Leonard et al., 1980).
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
2.1.1.2 Ingram (2001)
Ingram (2001) proposed four measures of whole-word productions to estimate children’s whole-word abilities. Firstly, the Phonological Mean Length of Utterance measures the length of a child’s words and the number of correct consonants.
Secondly, the Proportion of Whole-Word Proximity may capture how well a child approximates the target words. Thirdly, the Proportion of Whole-Word Correctness determines the proportion of a child’s words that are produced correctly out of the entire production. And fourthly, the Proportion of Whole-Word Variation provides a method for quantifying the amount of intra-word phonological variability exhibited by children.
In order to obtain the phonological variability of children, the fourth measure proposed by Ingram (2002), the Proportion of Whole-Word Variation, was adopted in this study. The relation between variability and phonological representation is still not clear. However, one hypothesis is that a decrease in intra-word variability would reflect the emergence of a segmental phonological representation. On the other hand, an increase in variability might reflect that the phonological system is not stable when it starts to undergo reorganization, a transition from whole-word representation to segmental representation (Sosa & Stoel-Gammon, 2006).
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
2.1.2 Syllable
Jakobson’s early work (1947, 1968) proposed a universal order of acquisition of
syllable structure. He indicated that children begin the processes of phonological acquisition with the CV or CV reduplicated syllable and gradually followed by more
complex syllable such as CVC and CVCV (where the second consonant-vowel combination is different from the first one). He also proposed ‘markedness theory of syllable’ which was summarized below:
(1) The open syllable is more unmarked than the closed syllable.
(CVCVC, CVV)
(2) The syllable with onset is more unmarked than those without onset.
(CVV, CVCVC)
(3) Syllable contained consonant cluster is marked.
(CVCCV, VCVCC)
Stark (1980) proposed the order of stages in the speech production before infants begin to produce their first word. In the stage called canonical babbling, infants in their 6-month-old start to produce sequences of identical CV syllables with adult-like timing such as [mama] or [baba]. At around 12 or 13 months, syllable strings with varying consonants and vowels, like [bagidabu] emerge as the more frequent type in infants’ speech. When infants are 10-month-old, syllables like V, VC, and CVC start
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
to emerge in infants’ production.
Allen and Hawkins (1978, 1980) proposed that young children acquiring English
tend to use the form of disyllabic trochaic feet. They observed that children often use augmented (CVC→CVCV) or truncated words. Furthermore, this early syllable
structure might be a universal tendency; that is, children all over the world acquire
languages with uniformity. Demuth and Johnson (2003) also found that children acquiring different languages use similar rules to truncate adults’ target forms. They
examined the phonological acquisition of French in longitudinal data from one French-speaking child, aged 1; 3-1; 5, and found that the child’s early words were all reduplicated CVCV forms. The examples of English and French children are
presented as below:
English [bənæ nə] → [‘nænə] ‘banana’
French [paˈtat] → [pəˈtæ:] ‘potato’
Child acquiring English produced [bənæ nə] as [‘nænə], and child acquiring French produced [paˈtat] into [pəˈtæ:]. Both English and French children truncated trisyllabic
word into disyllabic word.
2.2 Factors of phonological variability
Variability appears frequently in the developmental phonology literature, and is often used as a diagnostic marker of phonological disorders. However, less is known about normal patterns of variability. The following section introduces the factors of
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
phonological variability in typically developing children. Studies concerning phonological variability will also be presented.
Production variability has been attributed to a number of different factors which summarized into the following categories: physical factor and phonological factors.
2.2.1 Physical factor
The development of neuromotor control for speech that occurs during the period of early language acquisition can influence children’s speech production. Young
children have been found to demonstrate high levels of variability in many different aspects of motor control (Green, Moore & Reilly, 2002; Holm, Crosbie & Dodd, 2007;
Macrae, 2013; Sosa & Stoel-Gammon, 2012). In general, motor development might be summarized as a process of increasing accuracy and decreasing variability (Stoel-Gammon, 2006)
Green et al. (2002) investigated the sequential development of the upper lip, lower lip, and jaw movement of 1-, 2-, and 6-year-olds and adults during speech. The findings revealed that 1- and 2-year-old children’s jaw movements were significantly similar to adults’. However, 1-year-olds’ upper and lower lip movement patterns
exhibited high variability, which would become more adult-like with maturation.
These findings suggested that children’s early sound acquisition might be influenced
by the inconsistent development of articulatory control, with the jaw preceding the
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
lips. For example, it is easier for children to produce sounds formed by using mandible as primary mover like /b/ than those tend to be associated with lip control like /f/. According to Green et al. (2002), young children’s phonetic inventory was constrained by their dependence on the mandible to approximate adult-like speech targets resulting in the production of predictable speech errors and distortions.
2.2.2 Phonological factors
There are a number of phonological factors of phonological variability to be discussed. The first is phonetic context. The position of sounds in a word may affect the accuracy of production. Kenney and Prather (1986) examined the speech consistency of children aged 2;5 to 5;0. They found that children produced phonemes /t, l, f/ more accurately in word initial than in word final position.
The second phonological factor is phonological overload. It is not surprising that words that are difficult for a child to pronounce will display greater variability.
Leonard et al. (1982) examined 8 typically developing children ranging from 1; 10 to 2;2 years of age. They found that variability is most likely to occur when more than
one phonological or structural feature of the target words that show instability in the child’s linguistic system. Furthermore, words with higher variability rates are most
often those which have more advanced forms, sounds, or word shapes. Thus, variability can be seen as the result of phonological overload, which results in the
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
simplification or substitution of sounds that are difficult for children to produce.
The third factor of phonological variability is phonological complexity. McLeod and Hewett (2008) examined variability and accuracy in the production of words containing consonant clusters in typically developing children, aged 2; 0 to 3; 4. They found that the children in the study exhibited extensively variability when producing words that contained consonant clusters.
Macrae (2013) investigated word variability and accuracy in children aged 1; 9 to 3; 1. The study used Word Complexity Measure (Stoel-Gammon, 2012) to assign score to a word based on the three levels of complexity: word pattern, syllable structures and sound classes. The results showed that phonological complexity has a significant positive effect on word variability. Words with more complex speech sounds are produced with more variability than those with less complex speech sounds. These studies are consistent with the study of Ferguson and Farwell (1975) which also indicated the effect of phonological complexity on word variability.
The last factor of phonological factor is reorganization of phonological system.
Sosa and Stoel-Gammon (2006) investigated the patterns of intra-word production variability of English-speaking children during their first year of lexical acquisition (1;0-2;0). The variability pattern observed by Sosa and Stoel-Gammon (2006) showed peaks and valleys. Three of the four children showed a very noticeable peak in
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
variability. These peaks appeared when these children aged 1; 9 to 2; 0. It was also the time when two-word utterances were first observed in children’s speech. The results indicated that an increase in variability might correspond to instability in the phonological system when it undergoes reorganization: a movement from lexically-based system of phonological representation to segmental system. This view is consistent with dynamic systems theory, which proposed that variability is associated with transitions between developmental stages and is a potential force of developmental change (Thelen & Bates, 2003).
2.3 Introduction to Mandarin syllable
There are two parts in this section. The possible syllable types and the tone in Mandarin will be introduced.
2.3.1 Syllable types in Mandarin
All consonants except nasal [ŋ] in Mandarin can appear in the onset position, and
only nasals can be in the coda position. Prenuclear glides occurred in syllable initial position serve as an onset of the syllable, or it can occur before the syllable nuclear position. They are not considered as phonemes in Mandarin but can be treated as phonetic variants of high front vowels since the prenuclear glides do not contrast with the corresponding high vowels (Duanmu, 2007, Lin, 2007, Wan, 1999). Throughout this study, the Mandarin phones are presented by IPA system. Table 2.1 shows the
‧
description of possible consonant inventory, based on the studies from Lin (2007) Table 2.1 Mandarin consonants
Bilabial Labio- except for aspiration. The one on the left is voiceless unaspirated which is shaded, and the one on the right is voiceless aspirated.
Taiwan Mandarin allows at most four segments in a syllable, and is analyzed as having twelve syllable types: V, CV, VG, GV, VN, CVG, CVN, CGV, GVG, GVN, CGVG, and CGVN. The maximal syllable is CGVX, in which C is a consonant, G is a glide, V is a vowel, and X can be a glide or a nasal. Possible syllable types and examples are shown in Table 2.2.
Table 2.2 Possible syllable types in Mandarin (Wan, 1999:36) Syllable type Phonetic Transcription Gloss
V i55 dependent
CV ma55 mother
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
GV ja55 push
VG aj51 love
VN an55 safe
CVG maj214 buy
CVN tjŋ214 top
CGV ɕjɛ35 shoes
GVG jɑw35 shake
GVN wan51 ten thousand
CGVG tjɑw51 drop
CGVN thjɛn55 sky
In traditional analysis, Chinese syllable contains three parts: the first part is the Initial, which is optional and could be a consonant, a glide, or a nasal; the rest of the syllable after the initial consonant is the Final, which contains Medial and Rime. The medial is the glide before the main nuclear vowel. Rime can be further divided into two parts: the Nucleus, which is the main vowel in a syllable, and the Ending, which could be a glide or a nasal. The third part is the Tone, which is considered a property of the whole syllable (Cheng, 1973).
The phonological status of the prenuclear glides [j], [w], and [ɥ] in Mandarin Chinese has been well-studied but still remains as a controversial issue. Traditionally, the prenuclear glides are considered to be part of the final (Cheng, 1973), while Duanmu (1990) suggested that the prenuclear glides are not rime segments but secondary articulation on the onset. Duanmu (2007) pointed out that the
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
CG(C=consonant; G=glide) sequence, like [s] and [w] sounds in [swan] ‘sour’, is actually a single sound in Chinese, due to the fact that the lip rounding of [w] starts at the same time as [s]. In contrast, [sw] in English words like ‘sway’ are two sounds because the rounding of [w] occurs after [s].
Zhu and Dodd (2000) examined the phonological acquisition of Putonghua of 134 children aged 1;6-4;6 in Beijing. From the speech error pattern of syllable-initial consonant deletion, Zhu and Dodd pointed out that children acquiring Putonghua always delete the syllable-initial consonants before the vowels [i], [u], and [y]. This pattern may reflect the flexible function of these three vowels: these vowels have their variants [j], [w], and [ɥ], as mentioned before. Deletion occurs before these vowels suggested that children acquiring Putonghua have the tendency to cluster the prenuclear glides with the nucleus; that is, prenuclear glides tend to group with rime instead of onset in child phonology.
2.3.2 Tone
Syllable is a tone-bearing unit in Mandarin. People speaking tonal languages use tones to distinguish lexical meanings. Mandarin has four lexical tones and one neutral tone. Chao (1956) have provided a 5-point scale to specify the tone values in Standard Mandarin which were widely used and cited in most research on tone languages.
Table 2.3 presents the five-point scale proposed by Chao (1956).
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 2.3 Five-point scale of Mandarin tone
Tone Chinese Name Tonal feature Pitch value
Tone 1 Yinping High level 55
Tone 1 Yinping High level 55