Papers by Samuel Yu-hsiang Pan
University System of Taiwan Working Papers in Linguistics (USTWPL), 2018
This paper focuses on monophthongization in Takibakha Bunun. According to the filedwork data coll... more This paper focuses on monophthongization in Takibakha Bunun. According to the filedwork data collected from the informant,/ai/,/ia/,/au/,/ua/in the underlying forms are monophthongized in the non-final syllable position. That is, we can observe vowel alternations, such as [aj~ e],[ja~ e],[aw~ o],[wa~ o], in the surface form of Takibakha Bunun. For example, ma-cuaz [matswáð]‘to plant (agent voice)’; coz-an [tsóðan]‘to plant (locative voice)’. As for the previous studies, Li (1988) and Lin (1996) mention similar alternation, vowel metathesis, in Isbukun Bunun. Li (1988) also observes the monophthongization of [aj] and [aw] across Bunun dialects. In addition, Jiang (2012), a sketch grammar of Takibakha Bunun, argues that [e] and [o] are actually phonemic, but she does not provide a complete description of the vowel alternation pattern. In this paper, I focus on the interaction between syllable structure and vowel alternation pattern, attempt to analyze this phenomenon in the positional faithfulness theory, proposed in Beckman (1997, 1998) and to explain the problem through the ranking of constraints in the OT framework.
Conference Presentations by Samuel Yu-hsiang Pan
16th International Conference on Austronesian Linguistics (16-ICAL), 2024
This paper discusses morphophonological alternations of stems with final /g/ in Tgdaya Seediq, an... more This paper discusses morphophonological alternations of stems with final /g/ in Tgdaya Seediq, and adopts the Optimality Theory (OT) to account for this phenomenon. Seediq is a Formosan language spoken in central Taiwan. The sound /g/ at the word final position in Tgdaya Seediq undergoes significant variations depending on the preceding vowels, as noted by Li (1981) and Yang (1976). As a result, the presence of /g/ can only be confirmed in the suffixed forms. There are four types of variations: [ug~uw], [eg~uw], [ag~o], and [ig~uj].
The 24th National Conference on Linguistics, 2023
本文針對清代平埔歌謠〈斗六門社娶妻自誦歌〉進行全面的語言學分析。該歌曲取自黃叔璥（1722）《臺海使槎錄》中的〈番俗六考〉一卷。此卷收錄了三十四首平埔原住民族歌謠，均以漢字記錄並附有文言文譯注。... more 本文針對清代平埔歌謠〈斗六門社娶妻自誦歌〉進行全面的語言學分析。該歌曲取自黃叔璥（1722）《臺海使槎錄》中的〈番俗六考〉一卷。此卷收錄了三十四首平埔原住民族歌謠，均以漢字記錄並附有文言文譯注。
利用土田滋（1982）的平埔族群語言詞彙表，本文探討了〈娶妻自誦歌〉的語言特徵，特別是其中的詞彙成分。透過詞彙比較分析，結果表明此曲很可能是以道卡斯語（Taokas）唱誦，例如：斗六門社 描下女 ba-ha-li、道卡斯語 mahali 「女人」。此外，其與黃叔璥（1722）中所記錄的道卡斯歌謠亦共享了一些近似詞彙，例如：斗六門社 㖿呼 ia-ho、竹塹社 耶呼 ia-ho、道卡斯語 yakau 「酒」。然而，根據伊能嘉矩及粟野傳之丞（1900）對於平埔族群分布的描述，雲林地區應為羅亞族（Lloa）的居住地，顯示此地區可能發生過語言轉移（language shift）。
本研究提供了〈斗六門社娶妻自誦歌〉的初步詞彙分析與解讀，填補了我們在 雲林地區平埔族群語言理解上所缺失的一塊。本研究之發現不僅增進我們對此歌謠語言特徵的認識，也對平埔族群語言在漢人移居之前的可能分布情形提出了新的見解。
本文探討賽夏語和布農語的 CCV(C) 詞根形式。綜觀南島語系而言，多數的南島語不允許輔音串(consonant cluster)的出現，或僅允許輔音串出現在詞中位置(Blust 2013:21... more 本文探討賽夏語和布農語的 CCV(C) 詞根形式。綜觀南島語系而言，多數的南島語不允許輔音串(consonant cluster)的出現，或僅允許輔音串出現在詞中位置(Blust 2013:215)。而由原始南島語的重建形式來看，多數詞彙亦符合雙音節的典型形式(disyllabic canonical shape) (Blust 2013:690)。因此，CCV(C) 這類單音節又帶輔音串的詞根在南島語中可說是十分罕見。
不過，我們在賽夏語和布農語中仍可發現一些 CCV(C) 詞根的痕跡，例如：賽夏 S<om>pan 「飼養(主事焦點)」~ ka-Span-an 「飼養(處所焦點)」、s<em>peng「衡量(主事焦點)」~ ka-speng-an 「衡量(處所焦點)」；卡社布農 mu-qna 「再」~ tu-qna 「再說」、mu-skun「一起」~ tu-skun「合唱」。其中，賽夏語的 |Span| 和卡社布農語的 |qna| 雖然都不能單獨出現，但從焦點變化或是詞綴的變化來看，我們仍可看出其原有的形式。而兩個語言在針對 CCV(C) 詞根的表層音韻調整和形態變化都有不同的區別。賽夏語中 CCV(C) 詞根形式的能產性也較高；相對而言，布農語中的 CCV(C) 詞根的附著性(boundedness)較強，只能與不同的詞彙性前綴(lexical prefix)結合。
若進一步擴大比較，我們還能發現在一些名詞性形式中也能看現CCV(C)詞根的蹤影，例如：賽夏 kae-hma’ 「舌頭」(< 原始南島語 *Sema)、ka-tboS「甘蔗」(< 原始南島語 *tebuS)；卡社布農 iskaan「魚」 (< 原始南島語 *Sikan，比較 mala-skan 「釣魚」)、iqtiiv ~ qatiiv「菇」。這些形式在兩個語言當中皆已不具能產性(productivity)，但仍可從音韻、形態上看出其詞根形式的遺留。
本文以前人文獻資料及線上詞典及為主要語料來源，針對上述包含 CCV(C) 詞根形式進行共時音韻限制及歷時發展分析，在相關討論時亦會附上個人田野筆記。希望能羅列賽夏語及布農語中的 CCV(C) 詞根形式，作為未來其他南島語言的比較研究參考。
The 32th Meeting of the Southeast Asian Linguistics Society (SEALS 32), 2023
Abstract:
The reconstruction of the proto-phoneme *S in Proto-Austronesian (PAn), with a reflex o... more Abstract:
The reconstruction of the proto-phoneme *S in Proto-Austronesian (PAn), with a reflex of *h in Proto-Malayo-Polynesian (PMP), is mainly based on evidence from Formosan languages (Blust 2013). Despite its significance, the irregular reflexes of *S in some languages have led to debate regarding the possibility of splitting *S into separate phonemes (Tsuchida 1976, Dahl 1981, Li 1985). Ross (2015) conducted a comprehensive examination of Proto-Austronesian coronals and proposed splitting *S into *S and *x, based on regularity in the cognate set.
In this study, I investigate the reflexes of *S in Formosan languages through a review of literature and available linguistic data. The results show that only about half of the 11 instances provided by Ross (2015) match the expected sound correspondence. For example, Amis heci ‘meat’ does not correspond to Ross’s *xeθi (expected **seci). Moreover, some reconstructed forms lack sufficient evidence, such as the missing reflex of *Cumex ‘clothes louse’ in the Southern Taiwan linguistic area.
Additionally, the non-typical reflexes of *S are mostly present in languages in or near the Northern or Southern areas, reflecting as /h/ and zero respectively. Based on this observation, I propose an early contact hypothesis, suggesting that the interaction of languages in these areas may have caused the irregularity of *S. A complete examination of the issue is provided through a discussion of linguistic areas across Formosan languages.
Historical Relationships of East and Southeast Asian Languages (HRESAL 2022), 2022
This study begins by mentioning previous scholars' hypotheses about the subgrouping of the "North... more This study begins by mentioning previous scholars' hypotheses about the subgrouping of the "Northwest Formosan languages" and further discusses the relationships between these languages. Using phonological correspondences as a basis, it was found that while there are some common words in the basic vocabulary, many lexical items also appear in neighboring languages, such as Atayal and Taokas. According to the results of this study, the common words with phonological correspondences can be explained by early contact, which also accounts for interactions with neighboring languages. Future research needs to further explore the contact situations between neighboring languages and the Austronesian languages of northwest Taiwan.
第十三屆台灣語言及其教學國際學術研討會(ISTLT-13), Oct 16, 2020
布農語詞彙後加成分 (suffixal component) 可分為後綴 (suffix) 與依附詞 (clitic) 兩類。儘管在過去研究有過不少討論，如李俐盈 (2010，郡社方言)、黃慧娟... more 布農語詞彙後加成分 (suffixal component) 可分為後綴 (suffix) 與依附詞 (clitic) 兩類。儘管在過去研究有過不少討論，如李俐盈 (2010，郡社方言)、黃慧娟與施朝凱 (2018)，但以往族語書寫或是學者研究有時並無特別區別這兩者的差異，例如鄭恒雄 (1977，巒社方言)、戴智偉 (de Busser 2009，丹社方言)、李俐盈 (2018，郡社方言)，在兩種後加成分的區辨上並未明確說明。然而，從音韻上，我們仍可區分兩者的不同，尤其是布農語中同音 (homophonous) 而分屬兩類的後加成分，例如：=ang (持續貌依附詞，durative clitic)、-ang (祈使式後綴，imperative suffix)，以卡社方言為例：mi-qúmic=ang「還活著」、mi-qumíc-ang「請好好活著，引伸為感謝之意」。由此可見，後綴與依附詞的差異，無論是在音韻結構或是語義呈現上，都須受到重視。
本文以布農語卡社方言為主，同時參考其他布農語方言，針對後綴與依附詞的差異進行探討。音韻方面的區別，大致有：1. 重音轉移 (stress shift) 發生與否 2. 喉塞音或軟顎音 /k/ 的加插 (insertion) 3. 元音換位 (metathesis) 及單元音化 (monophthongization) 發生與否 4. 長元音 (long vowel) 保留與否。除了上述形態─韻律 (morpho-prosodic) 差異之外，本文也會討論同音後綴與依附詞的語義區別，並提供一些在族語書寫及教學上的建議。
2020年第六屆全國原住民族教育及文化發展學術研討會, Nov 13, 2020
道卡斯語是台灣南島語言之一，雖然目前一般被認定為「死語」，但在文獻資料中(土田滋1982、湯慧敏1998)仍留有許多豐富的詞彙紀錄。而近代調查部分，仍能向耆老蒐集到部分道卡斯語詞彙(吳東南199... more 道卡斯語是台灣南島語言之一，雖然目前一般被認定為「死語」，但在文獻資料中(土田滋1982、湯慧敏1998)仍留有許多豐富的詞彙紀錄。而近代調查部分，仍能向耆老蒐集到部分道卡斯語詞彙(吳東南1990、李壬癸1993、劉秋雲2001)。在過去的研究中，通常將道卡斯語視為一個整體，不細究道卡斯語的內部的差異及其相關的音韻問題。然而，無論是文獻記錄(土田滋1982)或是實際調查訪談(劉秋雲2001)，我們仍可發現一些方言之間的語音轉變，甚或是一些記音上的缺誤。
本文以土田滋(1982)所匯整的詞彙資料為主，針對道卡斯語音韻系統進行探討，同時也將比較古南島語(Proto-Austronesian, PAn)以及與道卡斯語有親緣關係的法佛朗語(Farvorlang)並列參考。本文發現在音韻方面的對應大致有：1. h:x:k的對應關係 2.v:b的對應關係 3.r:l:d:j的對應關係 4.詞尾s:t的對應 5.詞尾t:n的對應 6.詞尾k:t的對應 7.詞首 gy:y的對應關係等。本文最後也會針對道卡斯語的語料轉寫方式提出一些初步的建議，希望在道卡斯語的研究或是語言復振上都能提供助益。
關鍵詞：文獻資料、道卡斯語、音韻、方言差異
2019, The 29th Meeting of the Southeast Asian Linguistics Society (SEALS 29), 2019
Takibakha Bunun, an Austronesian language spoken in Taiwan, has different types of glottal stop L... more Takibakha Bunun, an Austronesian language spoken in Taiwan, has different types of glottal stop Li (1988) makes a comparative study of Bunun dialects and mentions a few types of glottal stops in the Bunun language. Jiang (2012), a sketch grammar of Takibakha Bunun, also describes the distribution of glottal stop in the language. However, the relationship between prosodic/morphological structures and glottal stops is under-discussed.
From previous studies and collected data, we find that these glottal stops can be (i) underlyingly encoded in the lexical input, (ii) derived from underlying voice stops, (iii) inserted, or (iv) copied from the preceding glottal stop in a phonological phrase. In this paper, we investigate how prosodic and morphological structures make an influence on the appearance of the latter three types of glottal stops. In addition, understanding prosodic and morphological structures can also help us to identify lexically-encoded glottal stops in Takibakha Bunun. Issues about phonological status, phonetic realization and orthographic notation of glottal stop in Takibakha Bunun are also discussed in this paper.
The 29th Meeting of the Southeast Asian Linguistics Society (SEALS 29), 2019
Bunun is one of the Austronesian languages spoken in Taiwan. This study focuses on Takibakha dial... more Bunun is one of the Austronesian languages spoken in Taiwan. This study focuses on Takibakha dialect spoken in Dili village, Nantou. Previous studies, such asLi (1988) and Lin (1996), report that Bunun is a language with only three phonemicvowels /i, a, u/, and mid vowels [e, o] are actually diphthongs /ai, au/ undergoingvowel coalescence, while Jiang (2012) argues that the two mid vowels havedeveloped into independent phonemes.
In this study, I use both field work and secondary data. Excluding loanwords, these words can be classified into two groups: common words (non-reduplication) andreduplication words. In common words, we can find no mid vowels occurring in theword-final syllable position, and the alternation of mid vowels and diphthongs is alsoobserved, e.g. ma-tsuað ‘to plant (AV)’, tsoð-an ‘to plant (PV)’. There is no suchrestriction on the distribution of mid vowels in reduplication words, but the case hereis somewhat complicated: reduplication words with mid vowels are, mostly, ideophones, and the stem of these words appears in a prosodic foot; non-ideophonicwords follow the general phonological process.
Based on the observation, I suggest that mid vowels in Takibakha Bunun, at leastin non-reduplication words, can still be interpreted as diphthongs in the underlyingrepresentation. Vowel alternation can be generalized with an OT constraint ranking: UNIFORMITY/PROM ≫ NODIPH ≫ UNIFORMITY. The second case is related to lexicaland prosodic-morphological issues; Therefore, I set up an alignment constraint ALIGN-FOOT/IDEO to account for it. Through the constraint interaction, we can seehow mid vowels emerge and how prominence and prosody play a role in the language, while the phonemic status of mid vowels in Takibakha Bunun still awaits furtherdiscussion.
Drafts by Samuel Yu-hsiang Pan
教育部第八屆原住民族語文學獎作品集, Nov 1, 2023
Uploads
Papers by Samuel Yu-hsiang Pan
Conference Presentations by Samuel Yu-hsiang Pan
利用土田滋（1982）的平埔族群語言詞彙表，本文探討了〈娶妻自誦歌〉的語言特徵，特別是其中的詞彙成分。透過詞彙比較分析，結果表明此曲很可能是以道卡斯語（Taokas）唱誦，例如：斗六門社 描下女 ba-ha-li、道卡斯語 mahali 「女人」。此外，其與黃叔璥（1722）中所記錄的道卡斯歌謠亦共享了一些近似詞彙，例如：斗六門社 㖿呼 ia-ho、竹塹社 耶呼 ia-ho、道卡斯語 yakau 「酒」。然而，根據伊能嘉矩及粟野傳之丞（1900）對於平埔族群分布的描述，雲林地區應為羅亞族（Lloa）的居住地，顯示此地區可能發生過語言轉移（language shift）。
本研究提供了〈斗六門社娶妻自誦歌〉的初步詞彙分析與解讀，填補了我們在 雲林地區平埔族群語言理解上所缺失的一塊。本研究之發現不僅增進我們對此歌謠語言特徵的認識，也對平埔族群語言在漢人移居之前的可能分布情形提出了新的見解。
不過，我們在賽夏語和布農語中仍可發現一些 CCV(C) 詞根的痕跡，例如：賽夏 S<om>pan 「飼養(主事焦點)」~ ka-Span-an 「飼養(處所焦點)」、s<em>peng「衡量(主事焦點)」~ ka-speng-an 「衡量(處所焦點)」；卡社布農 mu-qna 「再」~ tu-qna 「再說」、mu-skun「一起」~ tu-skun「合唱」。其中，賽夏語的 |Span| 和卡社布農語的 |qna| 雖然都不能單獨出現，但從焦點變化或是詞綴的變化來看，我們仍可看出其原有的形式。而兩個語言在針對 CCV(C) 詞根的表層音韻調整和形態變化都有不同的區別。賽夏語中 CCV(C) 詞根形式的能產性也較高；相對而言，布農語中的 CCV(C) 詞根的附著性(boundedness)較強，只能與不同的詞彙性前綴(lexical prefix)結合。
若進一步擴大比較，我們還能發現在一些名詞性形式中也能看現CCV(C)詞根的蹤影，例如：賽夏 kae-hma’ 「舌頭」(< 原始南島語 *Sema)、ka-tboS「甘蔗」(< 原始南島語 *tebuS)；卡社布農 iskaan「魚」 (< 原始南島語 *Sikan，比較 mala-skan 「釣魚」)、iqtiiv ~ qatiiv「菇」。這些形式在兩個語言當中皆已不具能產性(productivity)，但仍可從音韻、形態上看出其詞根形式的遺留。
本文以前人文獻資料及線上詞典及為主要語料來源，針對上述包含 CCV(C) 詞根形式進行共時音韻限制及歷時發展分析，在相關討論時亦會附上個人田野筆記。希望能羅列賽夏語及布農語中的 CCV(C) 詞根形式，作為未來其他南島語言的比較研究參考。
The reconstruction of the proto-phoneme *S in Proto-Austronesian (PAn), with a reflex of *h in Proto-Malayo-Polynesian (PMP), is mainly based on evidence from Formosan languages (Blust 2013). Despite its significance, the irregular reflexes of *S in some languages have led to debate regarding the possibility of splitting *S into separate phonemes (Tsuchida 1976, Dahl 1981, Li 1985). Ross (2015) conducted a comprehensive examination of Proto-Austronesian coronals and proposed splitting *S into *S and *x, based on regularity in the cognate set.
In this study, I investigate the reflexes of *S in Formosan languages through a review of literature and available linguistic data. The results show that only about half of the 11 instances provided by Ross (2015) match the expected sound correspondence. For example, Amis heci ‘meat’ does not correspond to Ross’s *xeθi (expected **seci). Moreover, some reconstructed forms lack sufficient evidence, such as the missing reflex of *Cumex ‘clothes louse’ in the Southern Taiwan linguistic area.
Additionally, the non-typical reflexes of *S are mostly present in languages in or near the Northern or Southern areas, reflecting as /h/ and zero respectively. Based on this observation, I propose an early contact hypothesis, suggesting that the interaction of languages in these areas may have caused the irregularity of *S. A complete examination of the issue is provided through a discussion of linguistic areas across Formosan languages.
本文以布農語卡社方言為主，同時參考其他布農語方言，針對後綴與依附詞的差異進行探討。音韻方面的區別，大致有：1. 重音轉移 (stress shift) 發生與否 2. 喉塞音或軟顎音 /k/ 的加插 (insertion) 3. 元音換位 (metathesis) 及單元音化 (monophthongization) 發生與否 4. 長元音 (long vowel) 保留與否。除了上述形態─韻律 (morpho-prosodic) 差異之外，本文也會討論同音後綴與依附詞的語義區別，並提供一些在族語書寫及教學上的建議。
本文以土田滋(1982)所匯整的詞彙資料為主，針對道卡斯語音韻系統進行探討，同時也將比較古南島語(Proto-Austronesian, PAn)以及與道卡斯語有親緣關係的法佛朗語(Farvorlang)並列參考。本文發現在音韻方面的對應大致有：1. h:x:k的對應關係 2.v:b的對應關係 3.r:l:d:j的對應關係 4.詞尾s:t的對應 5.詞尾t:n的對應 6.詞尾k:t的對應 7.詞首 gy:y的對應關係等。本文最後也會針對道卡斯語的語料轉寫方式提出一些初步的建議，希望在道卡斯語的研究或是語言復振上都能提供助益。
關鍵詞：文獻資料、道卡斯語、音韻、方言差異
From previous studies and collected data, we find that these glottal stops can be (i) underlyingly encoded in the lexical input, (ii) derived from underlying voice stops, (iii) inserted, or (iv) copied from the preceding glottal stop in a phonological phrase. In this paper, we investigate how prosodic and morphological structures make an influence on the appearance of the latter three types of glottal stops. In addition, understanding prosodic and morphological structures can also help us to identify lexically-encoded glottal stops in Takibakha Bunun. Issues about phonological status, phonetic realization and orthographic notation of glottal stop in Takibakha Bunun are also discussed in this paper.
In this study, I use both field work and secondary data. Excluding loanwords, these words can be classified into two groups: common words (non-reduplication) andreduplication words. In common words, we can find no mid vowels occurring in theword-final syllable position, and the alternation of mid vowels and diphthongs is alsoobserved, e.g. ma-tsuað ‘to plant (AV)’, tsoð-an ‘to plant (PV)’. There is no suchrestriction on the distribution of mid vowels in reduplication words, but the case hereis somewhat complicated: reduplication words with mid vowels are, mostly, ideophones, and the stem of these words appears in a prosodic foot; non-ideophonicwords follow the general phonological process.
Based on the observation, I suggest that mid vowels in Takibakha Bunun, at leastin non-reduplication words, can still be interpreted as diphthongs in the underlyingrepresentation. Vowel alternation can be generalized with an OT constraint ranking: UNIFORMITY/PROM ≫ NODIPH ≫ UNIFORMITY. The second case is related to lexicaland prosodic-morphological issues; Therefore, I set up an alignment constraint ALIGN-FOOT/IDEO to account for it. Through the constraint interaction, we can seehow mid vowels emerge and how prominence and prosody play a role in the language, while the phonemic status of mid vowels in Takibakha Bunun still awaits furtherdiscussion.
Drafts by Samuel Yu-hsiang Pan
利用土田滋（1982）的平埔族群語言詞彙表，本文探討了〈娶妻自誦歌〉的語言特徵，特別是其中的詞彙成分。透過詞彙比較分析，結果表明此曲很可能是以道卡斯語（Taokas）唱誦，例如：斗六門社 描下女 ba-ha-li、道卡斯語 mahali 「女人」。此外，其與黃叔璥（1722）中所記錄的道卡斯歌謠亦共享了一些近似詞彙，例如：斗六門社 㖿呼 ia-ho、竹塹社 耶呼 ia-ho、道卡斯語 yakau 「酒」。然而，根據伊能嘉矩及粟野傳之丞（1900）對於平埔族群分布的描述，雲林地區應為羅亞族（Lloa）的居住地，顯示此地區可能發生過語言轉移（language shift）。
本研究提供了〈斗六門社娶妻自誦歌〉的初步詞彙分析與解讀，填補了我們在 雲林地區平埔族群語言理解上所缺失的一塊。本研究之發現不僅增進我們對此歌謠語言特徵的認識，也對平埔族群語言在漢人移居之前的可能分布情形提出了新的見解。
不過，我們在賽夏語和布農語中仍可發現一些 CCV(C) 詞根的痕跡，例如：賽夏 S<om>pan 「飼養(主事焦點)」~ ka-Span-an 「飼養(處所焦點)」、s<em>peng「衡量(主事焦點)」~ ka-speng-an 「衡量(處所焦點)」；卡社布農 mu-qna 「再」~ tu-qna 「再說」、mu-skun「一起」~ tu-skun「合唱」。其中，賽夏語的 |Span| 和卡社布農語的 |qna| 雖然都不能單獨出現，但從焦點變化或是詞綴的變化來看，我們仍可看出其原有的形式。而兩個語言在針對 CCV(C) 詞根的表層音韻調整和形態變化都有不同的區別。賽夏語中 CCV(C) 詞根形式的能產性也較高；相對而言，布農語中的 CCV(C) 詞根的附著性(boundedness)較強，只能與不同的詞彙性前綴(lexical prefix)結合。
若進一步擴大比較，我們還能發現在一些名詞性形式中也能看現CCV(C)詞根的蹤影，例如：賽夏 kae-hma’ 「舌頭」(< 原始南島語 *Sema)、ka-tboS「甘蔗」(< 原始南島語 *tebuS)；卡社布農 iskaan「魚」 (< 原始南島語 *Sikan，比較 mala-skan 「釣魚」)、iqtiiv ~ qatiiv「菇」。這些形式在兩個語言當中皆已不具能產性(productivity)，但仍可從音韻、形態上看出其詞根形式的遺留。
本文以前人文獻資料及線上詞典及為主要語料來源，針對上述包含 CCV(C) 詞根形式進行共時音韻限制及歷時發展分析，在相關討論時亦會附上個人田野筆記。希望能羅列賽夏語及布農語中的 CCV(C) 詞根形式，作為未來其他南島語言的比較研究參考。
The reconstruction of the proto-phoneme *S in Proto-Austronesian (PAn), with a reflex of *h in Proto-Malayo-Polynesian (PMP), is mainly based on evidence from Formosan languages (Blust 2013). Despite its significance, the irregular reflexes of *S in some languages have led to debate regarding the possibility of splitting *S into separate phonemes (Tsuchida 1976, Dahl 1981, Li 1985). Ross (2015) conducted a comprehensive examination of Proto-Austronesian coronals and proposed splitting *S into *S and *x, based on regularity in the cognate set.
In this study, I investigate the reflexes of *S in Formosan languages through a review of literature and available linguistic data. The results show that only about half of the 11 instances provided by Ross (2015) match the expected sound correspondence. For example, Amis heci ‘meat’ does not correspond to Ross’s *xeθi (expected **seci). Moreover, some reconstructed forms lack sufficient evidence, such as the missing reflex of *Cumex ‘clothes louse’ in the Southern Taiwan linguistic area.
Additionally, the non-typical reflexes of *S are mostly present in languages in or near the Northern or Southern areas, reflecting as /h/ and zero respectively. Based on this observation, I propose an early contact hypothesis, suggesting that the interaction of languages in these areas may have caused the irregularity of *S. A complete examination of the issue is provided through a discussion of linguistic areas across Formosan languages.
本文以布農語卡社方言為主，同時參考其他布農語方言，針對後綴與依附詞的差異進行探討。音韻方面的區別，大致有：1. 重音轉移 (stress shift) 發生與否 2. 喉塞音或軟顎音 /k/ 的加插 (insertion) 3. 元音換位 (metathesis) 及單元音化 (monophthongization) 發生與否 4. 長元音 (long vowel) 保留與否。除了上述形態─韻律 (morpho-prosodic) 差異之外，本文也會討論同音後綴與依附詞的語義區別，並提供一些在族語書寫及教學上的建議。
本文以土田滋(1982)所匯整的詞彙資料為主，針對道卡斯語音韻系統進行探討，同時也將比較古南島語(Proto-Austronesian, PAn)以及與道卡斯語有親緣關係的法佛朗語(Farvorlang)並列參考。本文發現在音韻方面的對應大致有：1. h:x:k的對應關係 2.v:b的對應關係 3.r:l:d:j的對應關係 4.詞尾s:t的對應 5.詞尾t:n的對應 6.詞尾k:t的對應 7.詞首 gy:y的對應關係等。本文最後也會針對道卡斯語的語料轉寫方式提出一些初步的建議，希望在道卡斯語的研究或是語言復振上都能提供助益。
關鍵詞：文獻資料、道卡斯語、音韻、方言差異
From previous studies and collected data, we find that these glottal stops can be (i) underlyingly encoded in the lexical input, (ii) derived from underlying voice stops, (iii) inserted, or (iv) copied from the preceding glottal stop in a phonological phrase. In this paper, we investigate how prosodic and morphological structures make an influence on the appearance of the latter three types of glottal stops. In addition, understanding prosodic and morphological structures can also help us to identify lexically-encoded glottal stops in Takibakha Bunun. Issues about phonological status, phonetic realization and orthographic notation of glottal stop in Takibakha Bunun are also discussed in this paper.
In this study, I use both field work and secondary data. Excluding loanwords, these words can be classified into two groups: common words (non-reduplication) andreduplication words. In common words, we can find no mid vowels occurring in theword-final syllable position, and the alternation of mid vowels and diphthongs is alsoobserved, e.g. ma-tsuað ‘to plant (AV)’, tsoð-an ‘to plant (PV)’. There is no suchrestriction on the distribution of mid vowels in reduplication words, but the case hereis somewhat complicated: reduplication words with mid vowels are, mostly, ideophones, and the stem of these words appears in a prosodic foot; non-ideophonicwords follow the general phonological process.
Based on the observation, I suggest that mid vowels in Takibakha Bunun, at leastin non-reduplication words, can still be interpreted as diphthongs in the underlyingrepresentation. Vowel alternation can be generalized with an OT constraint ranking: UNIFORMITY/PROM ≫ NODIPH ≫ UNIFORMITY. The second case is related to lexicaland prosodic-morphological issues; Therefore, I set up an alignment constraint ALIGN-FOOT/IDEO to account for it. Through the constraint interaction, we can seehow mid vowels emerge and how prominence and prosody play a role in the language, while the phonemic status of mid vowels in Takibakha Bunun still awaits furtherdiscussion.