Chapter 1 Introduction
1.5 The framework of this study
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
11
involved in the production of the children? What are the differences of glides
performances between the normally-developing children and
phonologically-disordered children?
1.5 The framework of this study
The framework of this study is presented as follows. Chapter one provides a
general overview of the research background of glides development in terms of
phonetic and phonemic approaches. Three research questions are addressed in this
chapter. Chapter two reviews on the phonological development of glides in both
normally-developing children and phonologically-disordered children, and will have a
literature review on Mandarin Phonology, including consonants, vowel inventories,
and phonological accounts in terms of phonological representation and phonotactic on
Mandarin glides. In addition, the phonological acquisition theories regarding
Markedness theory and Positional effects and Positional prominence hierarchy will
also be described. Chapter three contains two main parts. The first part is concerned
with data collection, including the background of the subjects, data collecting
procedures, and the recording equipment involved. The second part is concerned with
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
12
data analysis. The methods to data transcription and the criteria adopted for the
assessment of the development of the glides in terms of various positions are provided
under this section. Chapter four presents the results found in the study. Regarding the
normally-developing children, the emergence and the stabilization of the three glides
in terms of various positions and the accuracy rate of the glides in relation to the
various positions are shown. Furthermore, the accuracy rate of each glide of the
phonologically-disordered children and the preference of the phonological are
presented. Chapter five provides the theoretical account for the development of glides
and a brief summary of the results in response to the three research questions
addressed in this study.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
13
Chapter 2
Literature review
In this section, the phonological development of glides in both
normally-developing children and phonologically-disordered children will first be
reviewed in section 2.1. Mandarin Phonology will be presented in section 2.2,
including consonant and vowel inventories in 2.2.1 and phonological accounts in
terms of phonological representation and phonotactic on Mandarin glides in section
2.2.2 and 2.2.3. Finally, the phonological acquisition theories regarding Markedness
and Positional effects and Positional prominence hierarchy will be displayed in 2.3.
2.1 Phonological development of glides
In this section, the acquisition of glides in both normally-developing children
and phonologically-disordered children are displayed in 2.1.1. Section 2.1.2 presents
normally-development and phonologically-disordered children acquiring glides in
Mandarin.
‧
2.1.1 Glides acquisition of normally and phonologically-disordered children
Considering the acquisition of sounds, researchers have studied more on
consonant acquisition rather than vowels as the process of constructing consonant
system involves more complicated motor skills and normally takes up a long period of
time (Smith, 1976). Comparing with the research on the acquisition of stops,
fricatives, and affricates, fewer studies discussed on the acquisition process of glides.
Previous researchers who dedicated to examining the phonological development of
English phonemes carried out cross-sectional studies. The review on the development
of glides of the five studies examining English young children recruited from
different areas is listed in Table 2.1 (cf., Dodd et al., 2003).
Table 2.1 An overview of studies on English glide phonemes (Dodd et al., 2003) Wellman et al.
(1931)
Poole (1934) Templin (1957)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
15
Notes:
1. For the row word position, I, M, and F refer to word-initial, -medial, and –final positions.
2. % age group refers to the minimum percentage of children of an age group required in deciding the acquisition of phoneme.
3. In the results section, Smit et al. (1990) list different age of acquisition for some of the phonemes at different word positions.
The five studies adopted the phonemic approach in investigating sound development
in English children. The differences between them include subject number, age range,
numbers of word positions and the percentage of an age group considered for the
acquisition standard. For the three oft-cited studies (Wellman, 1931; Poole, 1934;
Templin, 1957), all word positions were taken into account, while in Prather et al.
(1975) and Smit et al. (1990), only two positions were considered in the studies. The
criteria of 75% age group in deciding the acquisition of phoneme were adopted in
Wellman (1931), Templin (1957), and Prather et al. (1975), whereas Pool (1934)
adopted a stricter standard for determining the acquisition of a phoneme, 100%. The
differences in the findings might be resulted from the various criteria and the methods
employed in these studies.
Early work on phoneme acquisition adopted the classical approach (Wellman,
1931; Poole, 1934; Templin, 1957). Wellman (1931) studied 204 children from the
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
16
University of Iowa Laboratory Preschools, aged from 2;0-6;0. In total, 133 sounds
were examined in the three positions, word-initially, word-medially, and word-finally.
The data were collected by spontaneous speech or imitation. The minimum
percentage of children of an age group required in deciding the acquisition of the
phoneme was 75%; in other words, 75% of the children of an age group were required
to have mastered a sound in three positions for a sound to be considered acquired. The
results showed that [w] was acquired earlier than [j], for the former was acquired at
the age of 3 and the latter was not acquired until the age of 4. Poole (1934) recruited
65 children between 2;6 and 8;6 of age at the laboratory schools of the University of
Michigan. In total, 23 consonants were tested in the three positions, word-initially,
-medially, and -finally. The speech mode includes both spontaneous speech and
imitations. In her study, the assignment of a sound to an age level required 100% of
the children to utter correctly in all three syllable positions. The results of the age
acquiring glides found in Poole (1934) were 6 months later than those of Wellman
(1931) in both [w] (3;6) and [j] (4;6). Templin (1957) conducted an experiment by
investigating 480 children, consisting of 240 boys and 240 girls, ranging from 3 to 8
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
17
years with similar socioeconomic status. Of the 480 subjects, 58 were excluded since
they did not complete all of the tests. The subjects were tested on 176 sound elements
in the initial, medial, and final positions. Moreover, different words were used for
children 3 to 5 and children 6 to 8. Both the repetition data and spontaneous picture
identification utterances were obtained for the data analysis. The assignment of the
developmental age level required 75% of the subjects of a group to have mastered a
sound. The finding of labiovelar [w] is identical with that of Wellman et al. (1931),
indicating that children acquired [w] at the age of 3, but in Templin’s (1957), the [j]
sound was acquired at the age of 3;6, earlier than the findings in Wellman et al.
(1931). Sander (1972) reanalyzed Templin’s (1957) and Wellman et al.’s (1931) data
for average age estimates and upper limits of customary consonant usage he proposed.
The former refers to the ages at which correct production reach 50% and the latter
refers to the ages at which correct production reach 90% in all positions. He found
that the average age estimates and the upper age limits for [w] sound were before two
years old and at three years old. In other words, the children reached 50% correct
production of [w] sound before two years old and did not attain 90% correct
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
18
production until age three. In addition, the average age estimates and upper limits for
the sound [j] were 2;6 and 4;0. Prather et al. (1975) adopted both the classical and
distinctive feature approaches attempting to inspect 147 subjects between the ages of
2 to 4 years. The subjects were selected from the study for Sequenced Inventory of
Communication Development in Seattle (Hedrick et al., 1975). The subjects were
asked to name pictures spontaneously and repetition was requested when needed.
Moreover, assignment of a sound to an age level demands the accurate usage in only
two positions by 75%, word-initial and word-final positions. In contrast to the
previous studies, Prather et al. (1975) reported that [w] and [j] were assigned to 2;8
and 2;4 respectively, differing from the order found in previous studies, in which the
labiovelar [w] was acquired prior to the palatal [j]. Smit (1990) analyzed the
production data of 997 children age 3 to 9 gathered from Iowa Nebraska. Two word
positions, word-initial and word-final were considered in the study. The assessing
instrument contained 80 photographs and 108 phoneme targets. The imitation
production was avoided unless the target sounds were not elicited. The study further
tested on the urban-rural distinction as well as gender differences. The results
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
19
suggested that the urban and rural distinction did not play a role in the acquisition
process whereas gender seemed to exert a great difference. The girls acquired the [j]
sound by 4 years old, but the boys did not acquire the [j] sound until they reached the
age of 5. In more recent studies, Dodd et al. (2003) collected 684 children, aged
between 3;0 to 6;11 years, from nurseries and schools in eight different areas of the
UK. The study reported on two aspects of speech development, inclusive of the age at
which a sound is correctly pronounced (phonetic acquisition) and the accurate
production of a phoneme in terms of word contexts and the percentage of children of
an age group reaching the level of accuracy (phonemic production). The subjects were
asked to name 30 pictures for articulatory assessment and 50 pictures for error type
assessment. Both word-initial and word-final positions were taken into account. A
90% criterion was adopted for the assessment of phonetic acquisition. Moreover,
three factors influencing the process of speech development were discussed, including
age, gender, and socioeconomic status. They found that [w] and [j] reached the
acquisition criterion at the group of 3;0-3;5. Furthermore, they concluded that age and
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
20
gender affect speech development while socioeconomic status did not play a
prominent role in speech development.
The acquisition process of sounds involves complicated processes and therefore
progress gradually within a long period of time (Olmsted, 1971). For the purpose of
tracking sequential development of particular children and identifying the individual
variations performed in different children, researchers preferred longitudinal
observations to cross-sectional experiments. Given much data on consonant
acquisition, the data of glide could only be drawn from few studies. Smith (1973)
conducted a case study via observing his son, Amahl, acquiring English from the age
of 26 months until four-year-old. In most cases the data were spontaneous utterances
produced by Amahl. The production forms of Amahl were clearly listed in Appendix
C, from which we found the error productions of [w] and [j]. The sound [w] occurred
relatively early with fewer errors like [vaif] for ‘wife’. In contrast, the [j] sound
appeared to display various forms, as in yellow [lɛlo]/[dɛlo]/[ɛlo] and yes [d̩ ɛt]/[rɛt].
Furthermore, he pointed out that the complexity of syllable structures might bring
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
21
about increasing error rate in reduction error type, suggesting that glides and liquids1
which attach to the first consonant in a syllable are subjected to reduction; in other
words, glides are likely to be omitted in word-medial positions (Smith, 1973).
Stoel-Gammon (1985) analyzed longitudinal samples of meaningful speech of 34
normally-developing children involving 19 boys and 15 girls from Seattle area. The
data were collected at 3-month intervals between 9 months to 24 months of age. The
spontaneous speech was limited to the subjects who had reached the onset of
meaningful speech. The onset of meaningful speech referred to the age at which
subjects reached 10 different word types during an hour recording session. In addition
to age of emergence, the order of appearance of phones in terms of different positions
and a given position was investigated. The results implicated that the [w] sound
occurred before age of 2 based on the criterion of 50% of the subjects producing the
sound. However, the [j] sound did not appear in half of the subjects’ inventories
before 2-year-old.
Linguistic theories on language acquisition should be able to account for both
normative data as well as disorder data. The importance of investigating and further
1 The glides will be the focus of this paper.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
22
comparing the data of normally-developing children and those who exhibit
phonological disorder help construct a better view for child development and provide
a better explanation for the reference in speech pathology. Regarding glides
development in phonologically-disordered children, very few reported on the
performance of glides. The few studies presented the performance of glides in English
phonologically-disordered children have shown that they tended to replace the
presumably earlier-developing sounds with the later-occurring sounds (Lorentz, 1974;
Grunwell, 1981; Weiner, 1981; Leonard, 1985). Lorentz (1974) analyzed a 4-year-old
child, who replaced the labiovelar glide [w] with a liquid [l] in the C2 position in a
sequence of C1C2 clusters such as [tlɛl] for twelve. Weiner (1981) also found this
unusual process from the observation of a child at the age of 4;3, replacing the
consonants in the initial position with the later acquired sounds, interdental voiceless
[θ] and voiced [ð]. For example, the [j] sound in the word young and the [w] sound in
the word wagon were replaced with interdental fricative [θ]. A similar process was
found in the data presented in Grunwell (1981), who studied a
phonologically-disordered child, age 6;3. Several unusual behaviors were found in the
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
23
child’s data. The child produced interdental fricative [ð] as an alternative for all adult
forms involving fricatives, liquids, and glides in the initial position as in the word
yellow [ðɛʔdʊŋ]. Stoel-Gammon & Dunn (1985) identified seven idiosyncratic
processes that occur in the speech of phonologically-disordered children and the one
that is relevant to the study is that glides were replaced with stops2 (i.e., [wɪl] ‘will’
[bɪl]).
There are a few cross-linguistics studies in relation to the development of glides
in phonologically-disordered children. Dodd & So (1995) discovered a few odd
substitution patterns of glides in 17 monolingual Cantonese-speaking children with
phonological disorder aged between 3;6 to 6;4. The results described that the palatal [j]
and labiovelar [w] were replaced by glottal fricative [h] and nasal stop [m]
respectively. The inconsistent initial consonant reductions were identified in velar
stop [k] and palatal [j] and cluster reductions did not show a consistent pattern. Two
cases such as [kw] [f] and [khw] being omitted were presented in the study. In the
first case, [f] replaced [kw] in the initial position and in the second case, the CG [khw]
2 The rest of the processes include Atypical Cluster Reduction, Initial Consonant Deletion, Glottal Replacement, Backing, Fricatives substituted for stops, and Sound preference.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
24
was omitted. Golstein (2005) examined the substitution patterns in 39 typically
developing Spanish-speaking children, ranging from 3;2-4;11, and 39
Spanish-speaking children with phonological disorders between the age of 3;1 and 4;9,
from Puerto Rico. The accurate percentage and substitutions tokens of each sound
were listed, in which we found that typically developing children performed glides
with high accuracy rate. The labiovelar [w] was produced without any errors and only
three error tokens for palatal [j], showing the accuracy rate of 97.89%. The substitutes
include the segments [t, h, ɾ]. However, the subjects with phonological disorders
demonstrated a few substitution patterns for labiovelar [w] and palatal [j]. They used
the segments [b, m, d] as the substitutes for labiovelar [w] once within all the data. On
the other hand, the substitutes for palatal [j] were labiovelar [w] (3 tokens), the
segments [b, t, d] (2 tokens each), and the segments [g, f, tʃ, m, n] (1 token each).
2.1.2 Glide development in Mandarin children
Unlike the glide development in English was mostly described in the acquisition
process of consonants, the acquisition process of glides in Mandarin was discussed
through the acquisition of vowels as the glides in Mandarin are widely considered to
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
25
be the allophonic variants of the three high vowels [i, u, y], creating vowel
combinations, that is, diphthongs and triphthongs. Li (1977) observed a boy, aged
between 2;0 and 3;0, and a girl, aged between 1;1 and 1;8. The findings showed that
the corner vowels [i] and [u] were acquired without errors. Jeng (1979) examined one
boy aged between 2 months and 1;8 and the other boy aged between 1;3 and 2;7 for
testing on the Jakobson’s laws of irreversible solidarity. The findings suggested that
high front vowel [i] was acquired earlier than rounded back vowels [u] and [y]. Su
(1985) studied two Mandarin-Taiwanese bilingual aged 1;5 to 2;4 and 1;2 to 1;11. The
results reported that the order of three high vowel development is as follows: [i] [u]
[y], in which high front unrounded [i] was acquired earlier than the high back
rounded [u], followed by the high front rounded [y]. These studies have agreed on the
development of diphthongs were acquired later than the monophthongs. Moreover,
the acquisition processes of diphthongs and triphthongs involving /i, u, y/ demonstrate
the stage of deletion, addition and were substituted with single vowels before their
stabilization (Chao, 1951; Li, 1977; Jeng, 1979; Su, 1985). Furthermore, children
employ deletion at early stages and the substitution occurred later. In contrast, the
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
26
addition strategy does not appear frequently in young children’s production errors.
Hsu (1987) found that children acquiring Mandarin developed monophthongs first,
followed by the diphthongs [aj, ej, ɑw, ow]. The last acquired were rhymes ending in
nasals [an, ən, ɑŋ, oŋ], which refers to the combinations of a vowel followed by a
nasal coda. Lin & Lin (1993), who investigated young children acquiring Mandarin in
Taiwan, suggested that 15 vowels, including monophthongs [i, u, a, ɔ, ɤ, ɛ, ɝ],
diphthongs [aj, ej, ɑw, ow] and rhymes ending in nasals [an, ən, ɑŋ, oŋ] were acquired
by 90% criterion before 3-year-old, except for the rounded vowel [y]. They reported
children do not master the vowel [y] sound until the age of 3;5.
Zhu (2000) adopted a view that Mandarin vowel system consists of nine surface
forms, which contain nine monophthongs /i, y, u, ɤ, o, ɑ, ə, ɛ, ɚ/, nine diphthongs /ae,
ei, ɑo, ou, ia, iɛ, uɑ, uo, yɛ/, and four triphthongs /iɑo, iou, uae, uei/. She examined
the phonological development of Mandarin young children by conducting both
cross-sectional study and a longitudinal one. In total, 129 normally-developing
children aged from 1;6 to 4;0 were recruited in Beijing for the cross-sectional study
and four Mandarin-speaking young children aged 1;0 to 2;0 were observed for the
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
27
longitudinal study. Regarding the vowel development in her cross-sectional study on
normally-developing children, she found that they tended to reduce triphthongs and
diphthongs into diphthongs and single vowels respectively. The most frequent deleted
sequence of triphthong was [iɑo], simplified by 37% of the subjects, in which 29%
replaced it with [ia] and only 8% replaced it with [ɑo]. Another frequently reduced
sequence was [uei], simplified by 10% of the subjects, in which 7% of them were
reduced to [ei]. On the other hand, diphthongs were found to reduce to simple vowels
by dropping the onglide and offglide, retaining the most sonorant vowel within a
sequence. Aside from the cross-sectional study, the results found in longitudinal study
showed that [ei] and [iou] emerged first among all diphthongs and triphthongs, but the
[yɛ] combination was the last sound to appear.
The findings suggested by the previous research seem to imply that children
encounter difficulty in producing onglide and offglide in triphthongs and diphthongs,
producing the subtracted forms of vowels until 4 years of age. However, the study did
not discuss the onglide and offglide separately since they were regarded as a whole of
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
28
vowel combinations. Therefore, the performances of onglides and offglides were not
clearly discussed and remained for further inspection.
In relation to the performance on glides in Chinese-speaking children with
phonological disorder, Wu (1999) described two young children, aged 2;11 and 5.
From the data presented in their word inventory, the younger children performed
several errors containing glide deletion in both word-medial and word-final positions
and substitution in word-initial position. For example, the child deleted final
labiovelar [w] in [jaw] ‘want’ and palatal [j] in [pej] ‘cup’. The word [tɕʰjow tɕʰjow]
‘ball’ was produced as [tɔ tɔ], in which both the prenuclear glide and postnuclear
glide were reduced. Moreover, the initial glide in [ɥɛ] ‘moon’ was replaced with [j].
The data collected from the 5-year-old showed a similar process. Glides were reduced
in both medial and final positions, as in [kɔ pa] for [ʂow pʰa] ‘handkerchief’, [pi pɔ]
for [pʰiŋ kwɔ] ‘apple’, and [ka ja] for [tʰaj jɑŋ] ‘sun’. Zhu (2000) as well reported that
Mandarin-speaking children with phonological disorders demonstrated diphthongs
and triphthongs reductions and addition as well as vowel change. For example, [kwaj]
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
29
becomes [paj], showing the substitution of the initial consonants and the deletion
process of medial glide [w].
2.2 An introduction to Mandarin
This section presents an introduction to Mandarin phonology. In section 2.2.1,
both consonants and vowels are reviewed, followed by phonological accounts of
Mandarin glides in section 2.2.2. Finally, phonotactic of glide sequences are presented
in section 2.2.3.
2.2.1 Mandarin consonants and vowels
The possible consonant inventory in Taiwan Mandarin regarding both the place
of articulation (horizontal axis) and manner of articulation (vertical axis) based on
Lin’s (2007) study is presented below in Table 2.1 in International Phonetic Alphabet
symbols (IPA), a wildly used system for transcription of speech sounds for all
languages. The sounds appear on the left of the subdivided tables are unaspirated
voiceless, which share the same features of their corresponding counterparts on the
right except for the aspiration feature.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
30
Table 2.2 Mandarin consonants Bilabial Labio-
Table 2.2 Mandarin consonants Bilabial Labio-