華語幼兒音節縮減研究 - 政大學術集成

全文

(1)國立政治大學語言學研究所碩士論文. National Chengchi University Graduate Institute of Linguistics Master Thesis. 指導教授：萬依萍博士 Advisor：Dr. I-Ping Wan. 立. 政治大. ‧. ‧ 國. 學 y. Nat. 華語幼兒音節縮減研究. n. al. er. io. sit. A Study of Syllable Deletion in Mandarin Acquisition. Ch. engchi. i n U. 研究生：林欣瑩撰 Student: Shin-Ying Lin 中華民國一○五年六月 June, 2016. v.

(2) A Study of Syllable Deletion in Mandarin Acquisition. 立. BY 政治大 Shin-Ying Lin. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. i n U. v. A Thesis Submitted to the Graduate Institute of Linguistics in Partial Fulfillment of the Requirements for the Degree of Master of Arts. engchi. June, 2016 i.

(3) The members ofthe Committee approve the thesis of Shin-Ying Lin Defended on June 16th, 2016. 勺一. Z. I-PingWan($依萍 ) Advisor. 立. 政治大. ‧ 國. 學 Committee Member. ‧. n. er. io. sit. y. Nat. al. Ch. engchi. ""么. iv n Li-Hsin U Ning ()f俐馨) Committee Member. Amoved: CM^f{^^J@ Hsun-Huei Chang (jk^P M ), Director, Graduate Institute of Linguistics. 7乙.

(4) 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. Copyright © 2016 Shin-Ying Lin All Rights Reserved iii. v.

(5) 致謝其實到現在對於「通過口試」還是有種不真實感，連現在要打致謝詞還是很惶恐，因為念研究所的這段期間，我受到的關愛和幫助實在是太多了。首先要感謝我的指導教授萬依萍老師，從最初模糊的論文主題和方向，到確定題目和開始撰寫，老師在過程中不斷幫我釐清許多觀念和修正細節，老師在我徬徨焦躁時，不斷地給予鼓勵，讓我有信心地完成這本論文。非常感謝我的口試委員：甯俐馨老師、黃瓊之老師以及謝佳玲老師，三位老師百忙之中來參加我的口試，謝謝甯老師在統計分析及論文架構上給我很多寶貴的意見，黃老師針對邏輯與論文結果分析上給我許多建議，謝老師則是給予我很多論文撰寫與編排上的想法，真的非常感謝老師們。謝謝語言所的老師們給予的教導與提攜，讓我在這三年的研究生生涯，不僅獲得許多珍貴的學術知識，也讓我更加熱愛語言學，能夠在政大語言所完成碩士論文，這中間的成就感是無可比擬的；謝謝助教學姊，從入學修課到口試準備畢業，以及在所辦工作等等，都有助教學姊給予的協助；另外，也要感謝劉澤學長幫忙我設計程式，處理語料；謝謝統計系劉惠美老師與謝明哲學長熱心地協助我處理統計問題。一定要提到語言所 102 級的朋友們，跟你們一起學習、一起辦活動、一起野餐、一起聚會、一起為彼此加油打氣，因為有你們的陪伴，研究所的生活過得好充實快樂！很謝謝語音心理實驗室的大家，謝謝心怡學姐、冠霆學姊和彥棻學姊在工作和論文方面，都給予我好多的幫助；謝謝庭瑄和馨云，有你們兩位一起在實驗室奮鬥(瘋狂趕工作進度和論文進度)、一起聊天，之後畢業一定會很想念這段時光；也要謝謝品宇和雯淇的配合，實驗室多了你們的加入，變得更加歡樂了啊！謝謝我的好朋友郁倩、敬諭、曉婷，在我焦躁不安時，耐心地聽我抱怨，幫我加油，讓我又充滿鬥志打拼論文。最感謝我的家人，謝謝爸爸和媽媽支持我完成碩士學業，放手讓我做自己想做的事，在我碰到瓶頸時，儘管會小小地向家人抱怨，但是爸爸、媽媽和妹妹們總是做我最佳加油團為我打氣；還要謝謝我的男朋友孝宇，放假時總會帶我出門透透氣，舒緩緊繃的情緒，或是在我趕論文時，陪我上圖書館和咖啡館認真，用溫暖開朗的心，陪在一旁。最後，感謝每一位曾跟我說過加油的朋友，謝謝你們！現在想想，我能夠完成這本碩士論文，也是很激動、興奮、充滿無數的感恩，謝謝自己很努力，也謝謝收到許許多多的加油鼓勵，我好幸運能擁有許多人的幫助，我想我不會忘記這段寫論文的時光。終於，我可以開心地說：「我要畢業了！」。非常感謝大家！. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 2016/6/22 iv. i n U. v.

(6) Table of Contents Chapter 1 Introduction ................................................................................................... 1 1.1 The research background ................................................................................. 1 1.2 Cross-linguistic studies on syllable deletion .................................................... 5 1.3 Research questions ........................................................................................... 7 1.4 Organization of the thesis ................................................................................ 9 Chapter 2 Literature review ......................................................................................... 10 2.1 Introduction to Mandarin phonology ............................................................. 10 2.1.1 Syllable types in Mandarin ................................................................. 10 2.1.2 Syllable structures ............................................................................... 14 2.2 Syllable acquisition and syllable deletion in cross-linguistic studies ............ 17 2.2.1 English ................................................................................................ 18 2.2.2 Spanish ................................................................................................ 21 2.2.3 French ................................................................................................. 23 2.3 General theories on syllable deletion ............................................................. 25 2.3.1 Markedness theory .............................................................................. 26 2.3.2 Frequency effect .................................................................................. 29 2.3.3 Perceptual account .............................................................................. 31 2.3.4 Production account.............................................................................. 32 2.4 Syllable deletion in Mandarin, Cantonese and Taiwan Southern Min ........... 34 Chapter 3 Methodology ............................................................................................... 38 3.1 Data collection ............................................................................................... 39. 立. 政治大. ‧. ‧ 國. 學. sit. y. Nat. n. al. er. io. 3.1.1 Participants .......................................................................................... 40 3.1.2 Procedures ........................................................................................... 41 3.1.3 Recording equipment .......................................................................... 43 3.2 Data Analysis ................................................................................................. 44 3.2.1 Transcription and coding..................................................................... 44 3.2.2 Criteria for target word ....................................................................... 46 3.2.3 Frequency of overall syllable type ...................................................... 49 3.2.4 Frequency of syllable deletion ............................................................ 51 3.2.5 The pattern of syllable deletion........................................................... 52 Chapter 4 Result and analysis ...................................................................................... 54 4.1 Data Background Information ....................................................................... 54 4.2 Frequency of syllable types ........................................................................... 56 4.2.1 Frequency of overall syllable types .................................................... 57 4.2.1.1 Overall monosyllabic words .................................................... 57 4.2.1.2 Overall disyllabic words .......................................................... 61 4.2.1.3 Overall multisyllabic words ..................................................... 64. Ch. engchi. v. i n U. v.

(7) 4.2.2 Frequency of syllable deletion ............................................................ 68 4.2.2.1 Deleted monosyllabic words .................................................... 68 4.2.2.2 Deleted disyllabic words .......................................................... 72 4.2.2.3 Deleted multisyllabic words .................................................... 76 4.3 Patterns of syllable deletion in syllable positions .......................................... 79 4.4 Syllable relationship between children and caretakers .................................. 84 Chapter 5 Discussion and conclusion .......................................................................... 89 5.1 Summary of the findings ................................................................................ 89 5.2 Discussion on syllable analysis ...................................................................... 92 5.2.1 The overall syllable types ................................................................... 92 5.2.2 The syllable deletion types .................................................................. 94 5.2.3 The syllable deletion patterns ............................................................. 96 5.2.4 The syllable relationship between children and caretakers ................. 97 5.3 Conclusion ..................................................................................................... 99 References .................................................................................................................. 102. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. vi. i n U. v.

(8) List of Tables Table 2.1 Mandarin consonants (Lin, 2007)…………………………………………11 Table 2.2 Vowel phones in Taiwan Mandarin (Wan & Jaeger, 2003)………………12 Table 2.3 Possible Mandarin in Mandarin (Wan, 1999)…………………………….13 Table 2.4 Acquisition of syllable (Ingram, 1978)…………………………………...18 Table 3.1 Participants’ age range and recording duration……………………………40 Table 3.2 Participants’ gender and relation with children…………………………....41 Table 3.3 The sample of coding………………………………………………...........46. 政治大 Table 4.2 Background information regarding the caretakers………………………...56 立 Table 4.1 Background information regarding the children…………………………..55. Table 4.3 Duncan's Multiple Range Test for overall monosyllables………………...58. ‧ 國. 學. Table 4.4 Tokens and percentages of overall syllable types in monosyllabic words..59. ‧. Table 4.5 Percentages of overall syllable types of each child. sit. y. Nat. in monosyllabic words…………………………………………………….60. io. er. Table 4.6 Duncan's Multiple Range Test for overall disyllables…………………….61 Table 4.7 Tokens and percentages of overall syllable types in disyllabic words……62. al. n. v i n C hsyllable types of U Table 4.8 Percentages of overall each child in disyllabic words…..63 engchi. Table 4.9 Duncan's Multiple Range Test for overall multisyllables…………………65 Table 4.10 Percentages of overall syllable types of each child in multisyllabic words……………………………………………………67 Table 4.11 Duncan's Multiple Range Test for deleted monosyllables………………69 Table 4.12 Tokens and percentages of deleted syllable types in monosyllabic words………………………………………………….70 Table 4.13 Percentages of syllable deletion types of each child in monosyllabic words………………………………………………….71 Table 4.14 Duncan's Multiple Range Test for deleted disyllables…………………..73 vii.

(9) Table 4.15 Tokens and percentages of deleted syllable types in disyllabic words…..73 Table 4.16 Percentages of syllable deletion types of each child in disyllabic words..75 Table 4.17 Duncan's Multiple Range Test for deleted multisyllables……………….76 Table 4.18 Tokens and percentages of deleted syllable types in multisyllabic words…………………………………………………..77 Table 4.19 Percentages of syllable deletion types of each child in multisyllabic words…………………………………………………..78 Table 5.1 Ranking of used syllable types patterns of each child…………………….89. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. viii. i n U. v.

(10) List of Figures Figure 4.1 Percentages of overall syllable types in monosyllabic words……………59 Figure 4.2 Percentages of overall syllable types in disyllabic words………………..62 Figure 4.3 Percentages of overall syllable types in multisyllabic words……………66 Figure 4.4 Percentages of deleted syllable types in monosyllabic words…………...70 Figure 4.5 Percentages of deleted syllable types in disyllabic words……………….74 Figure 4.6 Percentages of deleted syllable types in multisyllabic words……………77 Figure 4.7 Percentages of syllable deletion in different positions…………………...81. 政治大 Figure 4.9 Percentages of syllable types of child #1 and caretaker #1………………84 立 Figure 4.8 Percentages of syllable deletion patterns of each child…………………..82. Figure 4.10 Percentages of syllable types of child #2 and caretaker #2……………..85. ‧ 國. 學. Figure 4.11 Percentages of syllable types of child #3 and caretaker #3……………..86. ‧. Figure 4.12 Percentages of syllable types of child #4 and caretaker #4……………..87. n. er. io. sit. y. Nat. al. Ch. engchi. ix. i n U. v.

(11) 國. 立. 政. 治. 大. 學. 研. 究. 所. 碩. 士. 論. 文. 提. 要. 研究所別：語言學研究所論文名稱：華語幼兒音節縮減研究指導教授：萬依萍研究生：林欣瑩論文提要內容 (共一冊，25,397 字，分五章)：本篇論文是針對以台灣華語為母語之幼兒，採長期觀察與收錄自發性語料之方式，研究台灣華語幼兒音節習得與音節縮減(syllable deletion)，並詳加描述單音節、雙音節和多音節之中音節類型出現的頻率、音節縮減模式，以及幼兒與母親的音節比較。本研究同時以 Jakobson (1968)的音節標記理論與頻率使用理論來檢驗幼兒的音節表現。本研究觀察四位年齡在十個月至兩歲六個月的幼兒，以及四位幼兒的照顧者 (母親)的語料。以兩個星期一次的頻率收集幼兒和母親之間的自然對話，並利用錄製回來的影音檔做譯寫分析。研究結果顯示幼兒的音節習得的表現是有規則可循的，在所有音節產出以及音節縮減的模式中，無標記音節類型 CV 的頻率最高；在幼兒初階的語言產製中，越是無標記的音節類型，發生頻率則會越高。另一方面，幼兒在早期的音節表現中容易產生音節縮減的模式，常使用韻尾省略(coda-dropping)的方式產出詞彙，以省略韻尾介音[w]和韻尾鼻音為主。此外，為了檢驗每一幼兒與其照顧者的音. 立. 政治大. ‧. ‧ 國. 學. sit. y. Nat. n. al. er. io. 節類型之間有無影響，透過統計分析，研究結果顯示幼兒與母親的音節類型比較呈現正相關，亦即幼兒的音韻產出或許會受到音韻環境的影響，進而產生與語言環境(母親語言)相似的音節類型與模式。. Ch. engchi. i n U. v. 關鍵詞：音節縮減、音節習得、音節縮減模式、頻率、台灣華語. x.

(12) Abstract The aim of this study is to discuss the issue concerning syllable deletion of the young children acquiring Taiwan Mandarin. The two main aspects are including in the following: syllable acquisition and syllable deletion pattern. The frequency of overall syllable types, the patterns of syllable deletion, and the syllable relationship between children and caretakers were examined. The findings can be explained by markedness theory proposed by Jakobson (1968) and frequency effects. The four children, aged between 0;10 to 2;6, and the four caretakers are investigated in this study. The spontaneous speech between the child and the mother every other week was recorded and analyzed. A longitudinal observation study is conducted by the researcher and the research team. Based on the analyses, the young children show a general syllable tendency; that is, the unmarked syllable type, CV, has the highest frequency in the analyses of overall syllable types and syllable deletion types respectively. The more unmarked syllable types occur more frequently in the children’s early production. Since the young children’s phonological systems are unstable and immature, they tend to produce the reduced syllable forms, especially the deletion of the postnuclear glide [w] and the deletion of the final nasal. The possible reason may due to the fact that the children use coda-dropping as a simplification strategy in their production. Furthermore, according to statistical findings, the syllable types of the four child-caretaker pairs are positively correlated. Young children’s phonological productions may be influenced by the overall productivity and the phonetic content of the ambient language.. 立. 政治大. ‧. ‧ 國. 學. er. io. sit. y. Nat. al. n. v i n Keywords: syllable deletion,Csyllable acquisition, U h e n g c h i deletion pattern, frequency, Taiwan Mandarin. xi.

(13) Chapter 1. Introduction 1.1 The research background Although different languages have their own distinct syllable types, all syllables in all languages have the basic form CV based on cross-linguistic reports (Spencer, 1996). In terms of syllable acquisition, numerous studies working on the acquisition. 治政 of phonology have focused on the order and pattern 大 of syllables. Researchers are 立 ‧ 國. 學. curious on whether language-universal syllable types exist in first-language acquisition (e.g., Ingram, 1974, 1978; Lleo & Prinz, 1996; Stoel-Gammon, 1998;. ‧. Levelt, Schiller & Levelt, 2000). Furthermore, the developmental facts about the. sit. y. Nat. io. n. al. er. syllable acquisition of phonology have been reasonably assessed in many places (e.g.,. i n U. v. Ingram, 1989; Stoel-Gammon, 1998, 2011). Children acquire CV type first because it. Ch. engchi. is the core syllable, which is the most structurally unmarked (Ingram, 1978; Stark, 1980). The studies of Ingram (1978) and Stark (1980) both proposed that children acquire CV and CVCV syllable forms first, and are then followed by V, VC or CVC forms. Moreover, the children aged around 2;0 are reported to exhibit more consistent patterns, characterized by more simpler words and syllable structures such as CV, CVCV and CVC (Fenson, Marchman, Thal, Dale, Reznick, & Bates, 2007). The problem stems from the fact that there is inconsistence of acquisition of 1.

(14) 2. syllable order in cross-linguistic studies. Some studies have revealed that syllable-final consonants are mastered earlier than syllable-initial consonants (So & Dodd, 1995; Zhu & Dodd, 2000), and word-final clusters are acquired several months earlier than word-initial clusters (Lleo & Prinz, 1996). However, some children are reported to acquire word-initial clusters (CCVC) earlier than word-final clusters (CVCC) (Levelt et al., 2000). In order to solve this problem, several previous studies. 政治大. aim at the effects of frequency in acquisition syllables. Researchers found that. 立. children are sensitive to the high-frequency linguistic structures of the language.. ‧ 國. 學. Bernhardt and Stemberger (1998) proposed that children have a tendency to produce. ‧. the less complex and more natural syllable structures in the speech; the simpler. Nat. io. sit. y. structures are acquired earlier. Roark and Demuch (2000) also showed that the earlier. al. er. acquired structures are much higher in frequency regarding syllable structures and. n. v i n words in a language. H. Y. C Wang h e(2014) h i U that the ranking of syllable types n g cdescribed. which are most frequently used is CV, and is followed by CGV, GV or V. Results also showed that syllable types with higher frequency and lower variability rates are unmarked syllable types. Moreover, researchers in the field of child acquisition also examined the language interaction between the caretaker and the child. They have wondered whether children are sensitive to the ambient language and the input frequency at the.

(15) 3. early stage. Ambient effects can be observed from children’s final consonant production over the period of 9 to 18 months (de Boysson-Bardies & Vihman, 1991) and children’s use of syllabic organization studies (de Boysson-Bardies, 1993). In addition, Albin and Echols (1996) compared the speech performances between children and caretakers in terms of labelling objects; the caretakers produced prosodically highlighted syllables more frequently to the children; children were. 政治大. sensitive to these prosodic input characteristics. Zamuner, Gerken and Hammond. 立. (2005) also discussed that high frequency inputs in the ambient language are more. ‧ 國. 學. likely to be produced by children than are low frequency inputs. Thus, studies have. ‧. provided persuasive evidence of ambient language influence on children’s linguistic. io. sit. y. Nat. development.. al. v i n C h in acquisition,Uchildren important engchi. n ambient language is. er. Although the syllable shapes that occur most frequently are acquired first and the may exhibit individual. differences in the learning path when the frequencies of two syllable structures are the same (Levelt et al., 2000). For instance, some children prefer the complexity of the syllable types in the syllable-initial first (i.e. CCVC) while others favor the complexity of the syllable type in the syllable-final first (i.e. CVCC). Data showed that the first earlier acquired and the highest frequency syllable structure is the most unmarked syllable CV, but the last acquired and the least frequent syllable structure is.

(16) 4. the most marked syllable CCVCC. Thus, it is suggested that the frequency of these syllable types coincided with markedness. Stites, Demuth and Kirk (2004) claimed that some children may show more frequency-based while others may show more markedness-based according to the investigation of word-final consonant acquisition in the corpus. In addition, the study of Tsay (2006) presented that CVC is the second most frequent syllable types, following the CV type, occurring in children acquiring. 政治大. Taiwan Southern Min. However, the error rate of producing CVC syllable types is up. 立. to 98% in the children’s production.. ‧ 國. 學. Since the relationship between frequency and the acquisition order of syllables. ‧. has been a subject of debate, many researchers attempted to discuss the acquisition of. Nat. io. sit. y. syllables of different languages. For instance, English allows at most three consonant. al. er. clusters in the syllable onset but Mandarin permits only two glides. Besides, the. n. v i n C h onset can be a nasal second segment in a two-segment e n g c h i U or liquid, like snack, train, and. black in English, whereas the second segment in a onset must be a glide (i.e. [j], [w], or [ɥ]) in Mandarin. Jakobson (1968) has shown that English-speaking children were acquiring open syllables earlier than closed syllables. Although English allows a large number of syllable types, CV syllable types are the most frequent in English words (Zamuner et al., 2005). Moreover, several researchers observed that young children, including English speaking children and French-speaking children, use augmented.

(17) 5. (CVC→CVCV) or truncated words (CVC→ CV) in their production (Allen & Hawkins, 1980; Demuth & Johnson, 2003). 1.2 Cross-linguistic studies on syllable deletion This section presents the research background of the types of syllable deletions 1 based on cross-linguistic studies. In order to investigate the area of syllable deletion, the first issue that needs to be considered is to define syllable deletion in children’s. 政治大. acquisition. In phonological development, syllable deletion can be regarded as a. 立. number of articulatory processes. Young children, aged 18 months to 32 months,. ‧ 國. 學. frequently omit certain syllables while producing polysyllabic utterances (Carter &. ‧. Gerken, 2003). The examples of children’s syllable omissions are presented in (1a) –. n. al. Ch. (1b) [dʒə'ræf] → ['ræf] giraffe. engchi. er. io. (1a) [bə'nænə] → ['nænə] banana. sit. y. Nat. (1b).. i n U. v. In phonetic studies, the articulatory weakening in the production of vowels particularly means “reduction.” In the study of Laver (1994), vowel-reduction can be seen as the process of shortening, pitch-lowering, and intensity-lowering. Along with syllable re-organization processes, these processes formed the compression or reduction of syllables in fluent speech (Hilton, Schüppert & Gooskens, 2011).. 1. In this study, I use syllable deletion throughout the whole paper. This term does not involve any specific process in relation to syllable contraction or expansion..

(18) 6. The second issue is the relationship between child acquisition and types of omitted syllables. There have been a number of analyses from cross-linguistic perspectives to examine syllable deletion (e.g., Allen & Hawkins, 1980; Wijnen, Krikhaar & Den-Os, 1994; Tsay, 2006). In terms of stress languages, studies showed that the children’s syllable deletion is strongly related to the stress assignment. For example, English-speaking children frequently produce the stressed portion of. 政治大. bisyllabic words (e.g., nana rather than banana) while omitting or producing with. 立. reduction of unstressed syllables. A similar pattern also occurs in children’s early. ‧ 國. 學. productions of Dutch and Spanish. They frequently omit the first syllable of words. ‧. Nat. io. sit. (Macken, 1992; Wijnen et al., 1994).. y. with a weak-strong stress structure and words with a weak-strong-weak structure. al. er. Other than stress languages, several linguists are interested in young children’s. n. v i n omission of tone languagesCand that stress is far less important. Tse h ethey n gobserve chi U (1991) proposed that Cantonese-speaking children in the phonological processes show structural simplification, such as syllable-final consonant deletion and cluster reduction. Furthermore, Zhu and Dodd (2000) explicitly put forward a study of phonological acquisition of 134 children, aged 1;6 to 4;6 in Beijing Mandarin. Based on speech error patterns, children on the one hand are likely to omit syllable-initial consonants before the vowels [i], [u], and [y] all the time; on the other hand, they are.

(19) 7. inclined to delete syllable-final nasals (i.e. /n/ deletion and /ŋ/ deletion). Tsay (2006) explored the relationship between prosodic models and syllable deletion produced by children acquiring Taiwan Southern Min. Results showed that word-initial syllables are omitted more frequently than word-final syllables in the corpus. It was suggested that children use simplification strategies, such as syllable omission, to shorten long and complex utterances. Consequently, by reviewing the studies of syllable. 政治大. acquisition, there are many different syllable development and types of syllable. 立. deletion from other languages. This study aims to investigate this issue by looking at. ‧ 國. 學. the patterns of syllable deletion in Taiwan Mandarin-speaking children.. ‧. There are only a few studies concerning the issue of acquisition of syllable in. Nat. io. sit. y. Mandarin and less is known about children’s types of syllable deletion in Mandarin. It. al. er. is still an issue how Mandarin children use the syllable deletion in the acquisitional. n. v i n C hsuch a process hasUa cross-linguistic tendency. This process, and whether or not engchi present study hopes to examine the syllable types and syllable deletion patterns from young children’s natural production. 1.3 Research questions This study examines syllable acquisition and syllable deletion patterns produced. by the four children acquiring Taiwan Mandarin, aged between 0;10 and 2;6, in a longitudinal observation method. The children’s pre-meaningful speech and.

(20) 8. meaningless production are excluded in the thesis. The frequency of each deleted syllable patterns will also be the focus in the study. The three main research questions to be addressed are described as follows: (1) Regarding overall syllable patterns: what is the general syllable type used by the four children? What is the most frequent syllable type used by the children? What is the least frequent syllable type used by the children? What is the rank-order of. 政治大. frequency in different syllable types? Is there obvious individual variation. 立. between four children’s overall syllable types?. ‧ 國. 學. (2) Regarding syllable deletion: What are those types of syllable deletion produced. ‧. by young children during their early phonological stage? What are the rates of. Nat. io. sit. y. different syllable deletion types? What is the most frequent syllable deletion type?. n. al. differences between. er. What is the least frequent syllable deletion pattern? Are there obvious individual. v i n C hchildren’s syllableUdeletion two engchi. patterns? What kind of. strategy do the children use to produce syllable deletion in lexicon? (3) Regarding caretaker-child language interaction: what is the general syllable type used by the four caretakers? What is the most frequent syllable type used by the caretakers? What is the least frequent syllable type? Is there relationship between caretaker’s frequent used syllable types and the child’s syllable types? Is the child’s syllable pattern influenced by the caretaker’s overall productivity?.

(21) 9. 1.4 Organization of the thesis The study is organized as follows. In chapter 1, I have set out the introduction to the research background of this study and research questions concerning data analysis. In chapter 2, firstly I will review the literature of Mandarin consonants, vowels, glides and syllable types in 2.1, and I will introduce the issue of syllable acquisition in cross-language studies and syllable deletion in 2.2. Theories on syllable deletion will. 政治大. be thoroughly shown in 2.3. The deletion in Mandarin and Taiwan Southern Min will. 立. be discussed in 2.4. Chapter 3 includes the methodology of this study. Section 3.1. ‧ 國. 學. presents the data collection methods. Section 3.2 is the data analysis explaining how. ‧. the data were arranged and analyzed. Chapter 4 will present the results in tables and. Nat. io. sit. y. figures. Section 4.1 will reveal the overview of the overall data. Section 4.2 will. al. er. demonstrate the results of syllable analysis, including the frequent syllable types that. n. v i n C hthe frequency of syllable children tend to produce and e n g c h i U deletion in the production. Section 4.3 will laid out the different patterns of syllable deletion of the four children’s productions. Section 4.4 will show the results of four child-caretaker’s syllable relationship. The discussion and explanation will be provided in chapter 5. Section 5.1 summarizes the findings in chapter 4. Section 5.2 presents the discussion on syllable analysis. The conclusion for the finding is provided in section 5.3..

(22) Chapter 2. Literature Review In this section, a review of literature of Mandarin phonology regarding Mandarin consonants, vowels and glides will be presented in 2.1. Secondly, I will introduce the issue of syllable acquisition in cross-linguistic studies in 2.2. I will illustrate the relationship between acquisition and syllable omission in 2.3. Theories on syllable. 治政 omission will be thoroughly shown in 2.4. Finally,大 I will summarize the omissions in 立 ‧ 國. 學. Mandarin and Taiwan Southern Min in section 2.5. 2.1 Introduction to Mandarin phonology. ‧. There are two parts in this section. The possible syllable types and the syllable. sit. y. Nat. io. n. al. er. structures in Mandarin will be introduced. 2.1.1 Syllable types in Mandarin. Ch. engchi. i n U. v. According to Lin (2007), consonants are articulated with obstruction of the airstream in the vocal tract to different degrees based on the manner of articulation. Consonants can also form constrictions in different locations by the place of articulation. The inventories of Mandarin consonants including glides and nasals are presented in IPA symbols in the following Table 2.1. Throughout this study, the Mandarin phones are transcribed by IPA forms.. 10.

(23) 11. Table 2.1 Mandarin consonants (Lin, 2007) Bilabial Stop. Labiodental. pʰ. p. Fricative. t f. Affricate Nasal. m. Central approximant. w ɥ. Dental. PostAlveolo Palatal alveolar -palatal. tʰ. k. s. ʂ. ts. tsʰ tʂ. ɕ tʂʰ. tɕ. tɕʰ ŋ. ɹ. 立. kʰ. x. n. Lateral (Approximant). Velar. j ɥ. 政l 治大. w. As shown in Table 2.1, symbols are arranged by the place of articulation and the. ‧ 國. 學. manner of articulation. Symbols that are under the same manner of articulation share. ‧. the every feature while symbols that are under the same place of articulation share. y. Nat. al. er. io. sit. every feature other than aspiration. The shaded one on the left is voiceless unaspirated. v. n. while the unshaded one on the right is voiceless aspirated. Nasals and approximants are all voiced.. Ch. engchi. i n U. Based on the studies of Duanmu (2000, 2007) and Lin (2007), the followings are some phonetic properties regarding Mandarin consonants. Firstly, aspiration is a distinctive feature. That is, aspirated and unaspirated stops are separate phonemes in Mandarin. The change of aspiration can change the meaning of the word; for example, the meaning of the word [pa]51 father differs from the word [pʰa]51 to fear. Secondly, the lateral and the nasal [m] can appear only in syllable initial position; all consonants.

(24) 12. except for the nasal [ŋ] in Mandarin can occur in the syllable-onset position, and only nasals [n] and [ŋ] can occur in the syllable-final positions. Finally, alveolo-palatals are not independent phonemes but allophones. They occur only before high front vowels or glides; that is, [ɕ], [tɕ] and [tɕʰ] appear only before [i]/ [j] and [y]/[ɥ]. Regarding the vowels of Taiwan Mandarin, they are classified according to degree of openness (vowel high), location of the active part of the tongue (vowel. 政治大. backness), and lip position (vowel rounding). There are twelve surface vowels in. 立. Mandarin, as shown in Table 2.2.. ‧ 國. 學. Table 2.2 Vowel phones in Taiwan Mandarin (Wan & Jaeger, 2003). e. Lower-mid. ɛ. io. al. a. n. Low. ɨ ə. Ch. engchi U. Back. Unrounded ɤ. y. Mid. y. Unrounded. sit. i. Nat. High. Rounded. er. Unrounded. Central. ‧. Front. Rounded u o ɔ. vɑ i n. Among the inventory of Taiwan Mandarin vowels, only the five vowels [i], [y], [u], [ɤ] and [a] can occur alone as complete syllables. The vowels [i], [ɨ], [y], [u], [ɛ], [ɔ], [ɤ] and [a] can appear in the open syllables. The vowels [i], [ɨ], [y], [ɛ], [ə] and [a] can occur in syllables closed with the nasal [n]; the vowels [i], [o] and [ɑ] can appear in syllable closed with the nasal [ŋ]. In addition, the vowels of [o] and [ɑ] can occur in syllables closed with the glide [w] while the vowels [a] and [e] can occur in syllables closed with the glide [j]..

(25) 13. Glides can be treated as phonetic variants of high front vowels rather than phonemes in Mandarin because prenuclear glides do not contrast with corresponding high vowels (e.g., Wan, 1999, 2002; Duanmu, 2007; Lin, 2007). When a high vocalic segment alternating with a glide is adjacent to a nonhigh nucleus vowel, there is an alternation of [i] with [j], [u] with [w], or [y] with [ɥ]; however, there is no glide alternation of the high central phone [ɨ] in Mandarin (Wan & Jaeger, 2003).. 政治大. Finally, since we have introduced consonants and vowels, we then look at the. 立. possible syllable types of Taiwan Mandarin, as in Table 2.3, based on the study of. ‧. ‧ 國. 學. Wan (1999).. Table 2.3 Possible Mandarin in Mandarin (Wan, 1999). Gloss. V. i55. dependent. VG VN. al. ja55. Caj51 hengchi U an55. sit. mother. er. ma55. n. GV. io. CV. y. Phonetic Transcription. Nat. Syllable Type. v ni. push love safe. CVG. maj21. buy. CVN. tjŋ214. top. CGV. ɕjɛ35. shoes. GVG. jɑw35. shake. GVN. wan51. ten thousand. CGVG. tjɑw51. drop. CGVN. tʰjan55. sky. As can be seen from Table 2.3, Taiwan Mandarin allows at most four segments in a raw for a syllable and it is analyzed as having twelve syllable types: V, CV, VG, GV,.

(26) 14. VN, CVG, CVN, CGV, GVG, GVN, CGVG, and CGVN. A syllable that has no coda is called an open syllable, such as a CV, GV or CGV syllable, whereas a syllable that has a coda is called a closed syllable, like CVN, CGVG or CGVN. 2.1.2 Syllable structures In terms of traditional analysis, the syllable can be divided into two main parts: the “initial” and the “final”. The initial means the syllable initial non-glide consonant,. 政治大. which can be the consonant or the nasal, while the rest of the syllable following the. 立. initial consonant is the final, which could be separated into the “medial” and the. ‧ 國. 學. “rime.” The medial is the glide before the nuclear vowel, and the rime consists of the. ‧. nucleus and the ending. The nucleus is obligatory in every syllable whereas the initial,. Nat. io. sit. y. the medial and the ending are all optional in Mandarin. The maximal syllable is. n. al. er. CGVX, where C is a consonant, G is a glide, V is a vowel, and X can be a glide or a nasal (cf. Lin, 1989, 2007).. Ch. engchi. i n U. v. Although the traditional analysis of Mandarin has been adopted for several researchers, the increasing studies reanalyze the syllable structure of Chinese with contemporary view (Duanmu, 1990, 2000; 2007; Bao, 1990, 1996; Lin, 2007). Regarding the status of prenuclear glide, the traditional analysis considers prenuclear glide to be structurally part of the rime, whereas the contemporary view shows that the prenuclear glide can serve as either the onset of the syllable, or the secondary.

(27) 15. articulation of the onset position (cf. Bao, 1990; Duanmu, 1990; Z. Wang, 1996). Bao (1990) proposed that the prenuclear glide formed a cluster with an initial consonant. The prenuclear glide was argued to be part of the onset rather than the rime segment. Based on the acoustic evidence, dialectal evidence, historical evidence and poetry rhyming patterns, Duanmu (1990) further elucidated that the prenuclear glide is the part of the onset and acts as the secondary articulations.. 政治大. Since the structural status of prenuclear glides in Mandarin has been a subject of. 立. debate, more researchers attempted to resolve this problem by experimental evidence. ‧ 國. 學. (H. S. Wang & Chang, 2001), or psycholinguistic evidence (Wan, 2002). H. S. Wang. ‧. and Chang (2001) asked participants to blend two syllables into one in the first. Nat. io. sit. y. experiment, and then choose from two alternatives (i.e. prenuclear glides clustering. al. er. with the onset or the rime) to break up a syllable in the second experiment. The results. n. v i n showed that the participantsC preferred the glide with the rime. Therefore, H. h e ntogclassify chi U S. Wang and Chang (2001) treated prenuclear glides as part of the final in support of traditional analysis. However, Wan (2002) proposed that prenuclear glides are part of initial or final depending on the place of articulation of the preceding consonant. Wan (2002) used psycholinguistic and acoustic evidence to reexamine the status of prenuclear glides from speech errors. She claimed that the glide could be syllabified in two ways in.

(28) 16. surface phonological representation. As the glide shared the same place of articulation with the onset, the glide formed a unit with the onset as the consonant cluster. Nevertheless, while the onset and the glide were not the same place of articulation, the glide stayed away from the onset. Wan (2002) thus concluded that whether the prenuclear formed with an onset or a rhyme largely depended on Mandarin phonotactic constrains and articulatory gestures.. 政治大. The syllable-structure status of glides in the postvocalic position is faced with. 立. different analyses owing to treatments of postnuclear glides and coda nasals. As. ‧ 國. 學. previously noted, the traditional analysis divides the syllable structure into the initial,. ‧. which is the onset consonant, and the final, which consists of prenuclear glides,. Nat. io. sit. y. nuclear vowels, and either postnuclear glides or nasals. It is suggested that the. n. al. er. postnuclear glide and the final nasal not only were in the same structure position, but. Ch. engchi. were treated as codas (Cheng, 1973).. i n U. v. However, investigators (Lin, 1989; Bao, 1990; Chiang, 1992) described that the postnuclear glides should be considered part of the nucleus instead of the coda based on linguistic data from Taiwanese language games. Lin (1989) proposed that the vowel and ending nasal were treated as different phonetic units in the rhyme position, whereas the vowel and the postnuclear glides were treated as the same unit (i.e. a diphthong) in the nucleus position. Chung (1989) used Hakka data to claim that.

(29) 17. postnuclear glides and postnuclear nasal consonants did not have an equal syllable-structure status. Therefore, Lin (1989) and Chung (1989) both suggested that a structure of postnuclear glides and ending nasals showed an asymmetrical behavior because the postnuclear glides were seen as part of the nucleus instead of the coda. Furthermore, Wan (2006) attempted to examine the status of postnuclear glides and coda nasals based on a corpus of speech errors in Taiwan Mandarin. She validated. 政治大. speech-error data as external evidence by psycholinguistic studies, which have been. 立. used for decades in English (Fromkin, 1973; Stemberger, 1983). Wan (2006) proposed. ‧ 國. 學. an asymmetry in the syllable structure between glides and nasals in the postvocalic. ‧. position. Because postnuclear glides were derived from vowels and associated with. Nat. io. sit. y. the nuclear vowel, postvocalic glides were not as firmly affiliated with the coda. al. er. structure as nasals. Besides, Chien (2011) also claimed that postnuclear glides should. n. v i n be treated differently from C final nasals. Based onUexperimental elicitation of speech he ngchi. errors, results showed that the error frequency of the interaction between postnuclear glides and final nasals was fewer than that between final nasals and final nasals. Consequently, these studies generally addressed that syllable structure of postnuclear glides and coda nasals did not behave in a parallel fashion. 2.2 Syllable acquisition and syllable deletion in cross-linguistic studies In this section, I will review studies on syllable acquisition and types of syllable.

(30) 18. omission on different languages, including English, Spanish and French. Linguists are interested in whether there are language universal in first language acquisition and children’s preferred types in syllable omission. 2.2.1 English The early study of Ingram (1974) found that there were general phonological processes operating in the child’s acquisition, including the deletion of consonant. 政治大. clusters and the deletion of unstressed syllables. He proposed that identifying general. 立. rules could account for the children’s simplification strategies. Ingram (1978) then. ‧ 國. 學. focused on his English-speaking daughter’s first-word stage in order to carefully. ‧. examine the acquisition of syllable types. Data showed that the child acquired CV and. Nat. io. sit. y. CVCV first, and were followed by CVC form. By the age of two, the child produced. n. al. er. most words containing closed syllables in the following Table 2.4.. Ch. engchi. Table 2.4 Acquisition of syllable (Ingram, 1978). i n U. v. Monosyllabic Words. Disyllabic Words. 1;3. 89% CV. 87% CVCV. 1;6. Mostly CVC. 47% CVCV. 2;3. Most of the words contained closed syllable. Ingram (1978) analyzed the monosyllabic and disyllabic token separately. When the child was at 1;3, 89% of words were CV form in monosyllabic words; however, when she was aged 1;6, most forms in monosyllabic words was CVC. The child was.

(31) 19. acquiring open CV syllables earlier than closed CVC forms. Moreover, before infants begin to produce their first words, specific and orderly changes that occur in the vocalizations can be observed. Stark (1980) found that in the “canonical babbling” stage, English-speaking infants in the 6-month-old age started to produce sequences of identical CV syllables (e.g. [mama]). At around 12 or 13 months, syllable strings, with varying consonants and vowels, emerged as the more. 政治大. frequent type in this stage. In terms of syllable patterns, when infants were. 立. 10-month-old, syllables like V, VC, and CVC started to appear at the babbling stage.. ‧ 國. 學. Ingram (1978) and Stark (1980) put forward an ordering of syllable development:. ‧. children acquired CV and CVCV forms first, and then V, VC, and CVC forms are the. Nat. earlier than coda consonants.. al. er. io. sit. y. next steps. Both studies showed that children mastered syllable onset consonants. n. v i n Stoel-Gammon (1998)Chowever that some English-speaking children’s h e n gfound chi U. most common target syllable forms were CVC, far more exceeding the frequency of CV and CVCV forms. Kehoe and Stoel-Gammon (2001) therefore claimed that codas were produced early by some English-speaking children because of lexical frequency. Onsets and codas were therefore presented in some English-speaking children’s first words. Hence, syllable patterns observed in child language offer a main source of evidence for investigating and understanding of phonology. Rose and Inkelas (2011).

(32) 20. proposed that every study on child phonology should be carefully examined, including all properties of the child’s language that occurred in the development. Although some studies focused on children’s general syllable types and occurrences of onsets or codas in syllables, other studies were related to the issue where young children acquiring English tended to truncate forms and delete syllables in production (e.g., Ingram, 1974; Allen & Hawkins, 1980; Gerken, 1994; Kehoe &. 政治大. Stoel-Gammon, 1997; Carter & Gerken, 2003, 2004). Young English-speaking. 立. children frequently omitted the initial weak syllables when producing polysyllabic. ‧ 國. 學. utterances. For example, when producing a weak-strong 2 (iambic) syllable structure. ‧. word giRAFFE, children tended to produce RAFFE, but rarely omitted the weak. Nat. io. sit. y. syllable of a strong-weak (trochaic) word such as MOKkey, producing MON (Allen &. al. er. Hawkins, 1980). The study of Gerken (1994) documented that the 2-year-old. n. v i n C h a preference forUwords with the primary stress on English-speaking children showed engchi the first syllable or for trochaic word structure, especially strong-weak syllables.. Recent studies have been concerned with the frequency of use of unstressed syllable omission by age (e.g., Dodd, Holm, Hua, & Crosbie, 2003; James, van Doorn, & McLeod, 2007). James et al. (2007) focused on 283 English-speaking children’s. 2. Throughout the study, S will refer to a strong or stressed syllable, and W will refer to a weak or unstressed syllable. Target lexical items will appear in italics with primary and secondary stressed syllables denoted by uppercase letters..

(33) 21. weak syllable deletion in terms of different age groups. Results showed that about two thirds of the participants used nonfinal weak syllable deletion, whereas final weak syllable deletion and deletion of stressed syllables could be negligible. Researchers explained that children did not acquire the final weak syllables and stressed syllables until the age of three, whereas they did not master the nonfinal weak syllable in polysyllabic words until the age of seven. Consequently, based on observation of. 政治大. different age groups, it is clearly noted that English-speaking young children’s. 立. omitted syllables largely depend on the stress factors.. ‧ 國. 學. 2.2.2 Spanish. ‧. The early study of Macken (1978, 1992) focused on the phonological. Nat. io. sit. y. development of syllable structures and co-occurring consonants by analyzing a. al. er. Spanish-speaking child, aged between 1;9 and 2;6. This child showed a gradual. n. v i n C hof the syllable numbers, increase in complexity in terms e n g c h i U syllable types and ordering when producing final-nasal words and fricative words. For numbers, monosyllable productions (i.e. CV forms) and the simplest of two-syllable productions (i.e. VCV forms) preceded disyllable productions of the CVCV forms. In terms of syllable types and ordering, CV was the most common, followed by CVC in this child’s production; no other syllable types were produced until this child was at the age of 2;6 when the first rudimentary CCV syllables were produced. In general, the child’s preferred.

(34) 22. syllable structure was CV form. Macken (1978) explained that this simple structure CV can be seen in the production processes, adapting adult’s CVCV and VCV both to the CV form. In addition to the phonological relationship between syllable types and consonants, Lleo and Prinz (1996) investigated the early stage of acquisition of consonant clusters of four Spanish-speaking children, aged from 0;9 to 2;1. Data. 政治大. revealed the following acquisition order: CV > CVC > CVCC > CCVCC. Based on. 立. the syllable ordering, they found that word-final clusters were mastered several. ‧ 國. 學. months earlier than word-initial clusters in the children’s production. Furthermore,. ‧. Nat. io. sit. explained by strategies of different syllabification.. y. they tended to reduce target clusters to a single consonantal position, which could be. al. er. Studies of Macken (1978, 1992) and Lleo and Prinz (1996) showed that children. n. v i n acquiring Spanish tended toC omit syllables during phonological acquisition. h ethentarget gchi U Macken (1978) proposed three major processes which combined to achieve the optimal CV syllable type: (1) syllable deletion (2) final-consonant deletion (3) consonant cluster reduction. Specifically, Macken (1992) in the following study revealed that the reason why Spanish-speaking children tended to omit the syllable structures because of trochees. That is, the first syllable of words with WSW structure would be omitted, such as manZAna [mən’zanə] “apple” omitted as ZAna [‘zanə]..

(35) 23. In order to account for the omission of initial unstressed syllables, the empirical basis for a longitudinal corpus in CHILDES (MacWhinney, 2000) has been used by a number of researchers (e.g., Prieto, Bosch-Baliarda, & Saceda-Ulloa, 2005; Saceda-Ulloa, 2005; Prieto, 2006). Saceda-Ulloa (2005) compared the initial unstressed syllable deletion in WS and WSW words of two Spanish children from the corpus CHILDES. Two children produced the initial unstressed syllables of bisyllabic. 政治大. WS words from the start, but there were almost no iambic (i.e. WS structure). 立. truncation. However, at the age of 1;2 to 1;8, children would omit the initial. ‧ 國. 學. unstressed syllable of trisyllabic WSW words. This omission of WSW syllable. ‧. structures of young Spanish-speaking children was consistent with previous studies. Nat. io. sit. y. (e.g., Prieto et al., 2005). Besides, the findings of Prieto (2006) accorded with. n. al. er. previous studies, showing that the omitted patterns were closely associated with. Ch. engchi. trochee model in children acquiring Spanish.. i n U. v. 2.2.3 French Other than English and Spanish, researchers found that children acquiring French presented a slightly different pattern of syllable acquisition. The early proposal by Fee and Ingram (1982) has noted that 24 French-speaking children, aged between 1;1 and 2;8, who frequently used reduplication would use multiysllables and show limited ability at final consonant production. Findings suggested that reduplication was.

(36) 24. regarded as a general pattern during the earlier stages of phonological development, and final consonant in CVC form seemed to be difficult for French-speaking children. In addition, Levitt and Aydelott Utman (1992) worked on cross-linguistical studies in order to explore the general and language-specific effects on child acquisition. They focused on the relationship between sound systems and syllable types of one French-speaking child and one English-speaking child. The utterances of. 政治大. both infants at 0;5, 0;8, 0;11 and 1;2 were acoustically analyzed for syllable types. In. 立. terms of syllable characteristics, the French-speaking infant less frequently produced. ‧ 國. 學. closed syllables than the English-speaking infant. Besides, results showed that the. ‧. occurrences of closed syllables for the French-speaking infant remained stable during. Nat. io. sit. y. observations, whereas the English-speaking infant showed a dramatic increase over. al. er. time because English has more closed syllables than French.. n. v i n C hwith an iambic prosodic It is reported that French, word structure that differs engchi U. from English and Spanish, shows another interesting perspective for examining young children’s syllable omission. According to the acquisition studies, young children were likely to omit disyllabic or trisyllabic targets to monosyllabic CV and CVC forms observed in the study of three French-speaking children, aged between 1;3 and 1;10 (Archibald & Carson, 2000). Moreover, Rose (2000) investigated many predicted syllable forms (e.g., CV, CVC and CVCV reduplicated form) in the.

(37) 25. longitudinal study of two children learning Quebec French. Data showed that not only the child would truncate CVC targets to CV forms, but also the trisyllabic targets were frequently reduced to disyllables. It is suggested that a binary foot was the maximal form for early prosodic words. Consequently, Rose (2000) claimed that syllable omission in French is expected to occur frequently. Demuth and Johnson (2003) further re-examined a longitudinal data collected by. 政治大. Deville (1891) of his daughter, acquiring Parisian French. The data gathered from the. 立. onset of the child’s first words until the age of two. The finding revealed that in the. ‧ 國. 學. beginning this child was faithful to the target form, correctly producing CV and. ‧. disyllabic reduplicated words. Then she went through a short period during which she. Nat. io. sit. y. produced CVC targets as CVCV reduplicated forms. During this period, she also. al. v i n foot. C Thereafter began to drop h e n gshe chi U. n maximally a binary. er. truncated trisyllabic words to disyllabic words, showing that her prosodic words were word-final consonants,. producing CVC targets as subminimal CV words, and truncated more disyllabic and trisyllabic targets. 2.3 General theories on syllable deletion Researchers have been particularly interested in explaining why children omit certain syllables in certain word positions more than others. Some researchers have proposed that children’s syllable deletion can be accounted by markedness theory.

(38) 26. (e.g., McCarthy & Prince, 1994; Pater, 1997; Roark & Demuth, 2000); other studies showed that frequency effect can explain the patterns of syllable deletion produced by the young children (e.g., Demuth & Johnson, 2003; Ota, 2003, 2006); other investigators have claimed that children’ syllable deletion are closely related to perceptual biases (e.g., Echols & Newport, 1992; Echols, 1993); still other researchers have observed the production account on young children’s truncations (e.g., Menn,. 政治大. 1983; MacNeilage & Davis, 1990, 1993). In order to investigate young children’s. 立. syllable deletion, four approaches, including markedness theory, frequency effect,. ‧ 國. 學. perceptual account and production account will be laid out in this section.. ‧. 2.3.1 Markedness theory. Nat. io. sit. y. Jakobson’s early work (1968) proposed a universal order of acquisition of. n. al. utterances would exhibit. er. syllable structure based on markedness theory. He proposed that young children’s. v i n Ch “unmarked” structure. e n glinguistic chi U. Children acquired CV. unmarked syllable, and gradually followed by marked syllable types. That is, they began the processes of phonological acquisition with CV or CV reduplicated syllable forms, and then followed by complex syllables, such as CVC and CVCV. In order to account for young children’s phonological development, the claim of optimality theory (McCarthy and Prince, 1994) posited of constraints on children’s outputs of the syllable acquisition in terms of markedness, where “unmarked”.

(39) 27. structures play an important role. Children appeared to start with “unmarked” structures in their early grammars (i.e. the syllable CV as the only possible output of the grammar), and gradually moved to more “marked” types of structures during the course of development. Gnanadesikan (1996) noted that young children’s first syllables would be the unmarked form of core syllables, or CV, showing a preference for the least marked onsets. Levelt et al. (2000) conducted a longitudinal study of. 政治大. twelve Dutch-speaking children, aged from 1;0 to 1;11. Results showed that there was. 立. a shared developmental order: children were acquired the order CV, CVC, V, and VC. ‧ 國. 學. subsequently. Tsay (2007) also examined the issue of the interactions between. ‧. markedness and syllables types of children acquiring Taiwan Southern Min, showing. Nat. io. sit. y. that CV was the most frequently used syllable, and was followed by CVC, CVV, and. n. al. er. V. According to above studies, CV syllable is considered as the most unmarked syllable across languages.. Ch. engchi. i n U. v. While discussing young children output data, Ingram (1986) proposed an active role for the children acquiring their phonological system. He viewed phonological processes as “a universal set of hierarchically ordered procedures used by children to simplify speech” (Ingram, 1986). Besides, Matthei (1989) found that the children learning English and Dutch initially omitted the use of coda consonants, producing early words with unmarked “core” CV syllables. They frequently omitted the pretonic.

(40) 28. syllable, such as [nænə] in baNAna, which conforming to trochaic foot. Pater (1997) further examined the truncation patterns in children’s productions. Although some children were likely to truncate banana to [nænə], he found that the children truncated early words containing an initial unstressed syllable, selecting the word-initial stop to fill the onset of the truncated form, such as [bænə] banana. As a result, Pater (1997) explained that young children had a tendency to preserve the consonant that was unmarked in the onset position.. 立. 政治大. With the development of OT theory, children’s early prosodic structures also can. ‧ 國. 學. be understood in terms of a series of hierarchically linguistic constraints, which. ‧. involve structural constraints and faithfulness constraints (cf. Pater, 1997; Roark &. Nat. io. sit. y. Demuth, 2000). Among these constraints, the lack of early codas indicates the relative. al. er. higher ranking of a constraint “No-Coda”, which is the prime key of the study. CVC. n. v i n target forms are realized asCCV, the relative ranking of faithfulness h edepending n g c honi U. constraints on the realization of segments. The early syllable deletion is closely related to the re-ranking of markedness constraints. On the other hand, in order to examine the learnability preferences in terms of markedness and frequency, Stites et al. (2004) proposed that children exhibited the different learning strategies in coda consonants owing to different input frequency. Results showed that one child acquired the more marked but more frequent stop codas first, while the other child acquired.

(41) 29. less marked but less frequent nasal and fricative codas first. Thus, although children’s early stages of development could be determined by markedness constraint and frequency effect, the researchers suggested that the unmarked structures might be also the most frequent, resulting in early and stable acquisition patterns cross-linguistically (e.g., Demuth, 1996; Stites et al., 2004). 2.3.2 Frequency effect. 政治大. A number of studies have discussed the frequency effects in the course of child. 立. language development, including the input frequency of phonemes, phoneme. ‧ 國. 學. sequences and syllable structures (e.g., Ingram, 1988; Demuth & Johnson, 2003;. ‧. Zamuner et al, 2005; Ota, 2003, 2006). The likelihood that a young child produced a. Nat. io. sit. y. target phonological structure was closely related to the input frequency (Ota, 2006).. al. er. For instance, Ingram (1988) found that the interaction between the accuracy and. n. v i n C hcould show the frequency acquisition timing of phonemes in the language. Chen and engchi U. Kent (2005) analyzed 24 Mandarin-speaking children’s CV affiliation patterns in early spontaneous vocalizations and examined the caretakers’ CV combinations as well. The findings showed the positive correlation between number of CV productions of children and their caretakers. That is, the young children’s phonological productions may influenced by the overall productivity and the phonetic content of the ambient language. Besides, in a cross-linguistic study, Roark and.

(42) 30. Demuth (2000) found that English-speaking children and Spanish-speaking children would acquire structures that were much higher in frequency relative to syllable structures and word shapes in each language. It is reported that syllable deletion also appeared in French-speaking children’s production. Many disyllabic and trisyllabic words frequently truncated to monosyllabic words (Demuth & Johnson, 2003); however, this deletion pattern. 政治大. occurred less frequently in the production of Japanese-speaking children (Ota, 2003).. 立. Demuth and Johnson (2003) therefore attributed these cross-language differences to. ‧ 國. 學. the high frequency of CV syllable words by French-speaking children. In addition,. ‧. Ota (2006) further examined whether the frequency effects of prosodic word. Nat. io. sit. y. structures and lexical item impacted on Japanese-speaking children’s patterns of early. al. er. word truncation based on CHILDES database. The findings revealed that young. n. v i n children’s target words wereCless h elikely h i Uwhen they were lexical items that n gtoctruncate were more frequent in the inputs. Ota (2006) explained that perhaps mothers knew some words that their children enable to produce in a target-like fashion, so they would repeat those words more frequently. Accordingly, it was concluded that the importance of taking input frequency account when discussing about young children’s phonological systems..

(43) 31. 2.3.3 Perceptual account The perceptual salience proposes that stressed syllables are acoustically more prominent than unstressed syllables. Moreover, word-final syllables are more prominent than nonfinal syllables because they may receive acoustic cues related phrase boundaries such as lengthening and accent effects. The recent articulation of a perceptual model has been provided by Echols and. 政治大. colleagues (Echols & Newport, 1992; Echols, 1993). They gathered data from. 立. English-speaking children during the one-word stage of language development. They. ‧ 國. 學. observed that stressed and final syllables with more salient acoustic information were. ‧. retained, whereas others were omitted. Results therefore showed that stressed. Nat. than syllables that were unstressed and nonfinal.. al. er. io. sit. y. syllables or final syllables were frequently preserved and more accurately produced. n. v i n C h further discussedUthe relationship between saliency In addition, Echols (1993) engchi. and stressed/unstressed syllables. Because unstressed nonfinal syllables were acoustically less salient than stressed syllables, they were less specified in underlying representations and they were vulnerable to omission. Moreover, Kehoe and Stoel-Gammon (1997) also indicated the importance of the phonetic content of syllables, which caused them to be less salient and less represented than other syllables. Because of under such salience proposals, weak syllables were omitted and.

(44) 32. they were less likely to be fully encoded in the first place. As a result, banana would be encoded as [nænə] because [bə] is a weak syllable and stress is the salience in English. Although a mechanism focusing on the perceptual properties of syllables is consistent with the pattern of syllable deletions observed, this account still faces problems. On the one hand, it is unclear how to explain the truncations such as [bænə]. 政治大. banana in terms of perceptual account. The children must have perceived the onset to. 立. the unstressed syllable if it is mapped into the word form. On the other hand,. ‧ 國. 學. individual variation is frequently attested. For example, children frequently alternate. ‧. between truncated and non-truncated forms in their production. That is, banana. Nat. io. sit. y. [bənænə] and banana [nænə] occur at similar point, suggesting that their lexical. al. er. representations are not constrained (Kehoe & Stoel-Gammon, 1997). Thus, the. n. v i n C hunanswered questions perceptual account still leaves e n g c h i U regarding why children delete certain syllables from their early words. 2.3.4 Production account Other studies attempt to discuss syllable deletion on the basis of articulatory framework mainly proposed by Davis and MacNeilage (1990). This articulatory-base account is the frame-content theory (Menn, 1983; MacNeilage & Davis, 1993). In the beginning, investigators introduced the concept of articulatory modulation in terms of.

(45) 33. the basic consonant-vowel (CV) syllable type from the closing and the opening mandibular movements. The closing is for consonant production, and the opening is for vowel. They argued that young children’s syllable production process is restricted by cycles of mandibular oscillation. Besides, mastication bears a motor resemblance to the opening and closing phases of articulatory cycle, associated with syllabic elements of spoken language. Hence, this account notes that articulatory regularities. 政治大. in the sound patterns of babbling and early speech could be attributed to properties of. 立. basic mandibular cycle (MacNeilage & Davis, 1993).. ‧ 國. 學. However, this articulatory account runs into similar problems with perceptual. ‧. account. If the young children are articulatorily restricted in either syllable complexity. Nat. io. sit. y. or the number of syllables per word, researchers expected these maturational. al. er. limitations to be found cross-linguistically. However, according to acquisition studies,. n. v i n it is reported that children C acquiring produced WSW words like manZAna h e nSpanish gchi U. [mən'zanə] several months before their English-speaking children peer group produced WSW trisyllablic words like baNAna [bə'nænə] (Demuth, 2001). Spanish-speaking children also produced trisyllabic words with an initial unstressed syllable also earlier than German-speaking children (Lleo, 2001). Therefore, it seemed that English- and German-speaking children’s syllable omission of initial unstressed syllables must be due to non-articulatory factors (Roark & Demuth, 2000)..

(46) 34. On the whole, most of the cross-linguistic studies showed a strong relationship between stress and syllable in children’s phonological acquisition. However, in Mandarin, the salient status of stress is lower. Duanmu (2000) suggested that it is difficult to detect stress in Mandarin since the most important phonetic cue for stress is F0, but F0 is already for tone use in order to differentiate word meanings, and thus cannot be used freely for stress. Moreover, Zhu and Dodd (2000) showed that weak. 政治大. stressed syllables have a very short duration and a much reduced pitch range. The. 立. pitch of a weak syllable is primarily determined by the preceding tone. Compared. ‧ 國. 學. with the saliency of tones, consonants and vowels, the saliency of weak stress is far. ‧. less important because of the low value in differentiating word meaning (Zhu & Dodd,. Nat. io. sit. y. 2000). Consequently, because this study attempts to investigate the syllable. n. al. er. acquisition in Mandarin and types of syllable deletion, the following section will be. i n C laid out the relative phonological elements in Mandarin. he ngchi U. v. 2.4 Syllable deletion in Mandarin, Cantonese and Taiwan Southern Min As previously stated, studies have indicated that young children’s types of syllable omission will be easily influenced by the factor of stress because stress is salient in stress languages. However, children’s syllable deletion in tonal languages may not be related to stress. According to Zhu and Dodd (2000), they worked on the phonological acquisition of Beijing Mandarin, and they found that children acquiring.

(47) 35. Beijing Mandarin acquired syllable elements in the following order: tone was acquired first; and followed by syllable-final consonants and vowels; syllable-initial consonants were mastered last. Results showed that tone errors were rare, even in the youngest group of children. Tone has the highest saliency in Mandarin: it is compulsory for every syllable structure; change of tone will change the lexical meaning (Zhu & Dodd, 2000; Yip, 2002). It is suggested that the saliency of the. 政治大. components in a language system determines the order of acquisition. Consequently,. 立. based on the comparison of phonological processes used by children acquiring. ‧ 國. 學. different languages, phonological patterns showed both universal tendencies and. ‧. language-specific constraints on acquisition.. Nat. io. sit. y. In terms of Cantonese acquisition, Tse (1991) investigated the process of the. al. er. acquisition of the phonological system based on a longitudinal study of one. n. v i n C h to the analysis Cantonese-speaking child. According e n g c h i U of child’s speech data, the most used structure in the order of frequency was CV, and was then followed by CVG and CVC. However, CVG syllable remained a relatively unstable structure with omission of the final glide. Furthermore, while analyzing the structural simplification, Tse (1991) observed that the child was likely to delete the consonant cluster [kw] to be either [k] or [w]. In addition to initial consonant, final consonant deletion was seen as a fairly active process, with 17.6% of the final consonant deletion. The most.

(48) 36. frequently affected type was postnuclear glides with 24.6%, followed by nasal finals with 10.6% and stop finals with 1.4%. Hence, Tse (1991) found the evidence for universal trends in the sequence of phonological acquisition, and concluded that these phonological patterns, simplification strategies and salient phonological distributions were accounted for Cantonese children. In order to further attest phonological patterns, So and Dodd (1995) enrolled 268 Cantonese-speaking children to analyze their. 政治大. speech errors in a picture-naming test. Based on the comparison of phonological. 立. structures of different languages, the younger Cantonese-speaking children made. ‧ 國. 學. more errors in the syllable-initial position than in the syllable-final position. Results. ‧. revealed that only a relatively small proportion (10.7%) of the total number errors. Nat. io. sit. y. were made on final syllable positions, including fronting, backing and final consonant. al. er. deletion. However, for syllable-initial consonants, So and Dodd (1995) indicated that. n. v i n the acquisition patterns of C children Cantonese exhibited a interaction of h e nacquiring gchi U linguistic universal tendency with language-specific effects. On the other hand, Tsay (2006) found that the patterns of syllable deletion occurred differently in terms of the word-position and the number of the syllables. Tsay (2006) explored the relationship between prosodic model and the syllable omission patterns produced by children acquiring Taiwan Southern Min, aged from 1;6 to 3;0. The children used syllable omission as simplification strategies to shorten.