2009年1月7日 星期三

國語口語語篇韻律研究簡述

http://phslab.ling.sinica.edu.tw/
國語口語語篇韻律研究簡述

摘要:口語語流韻律是否只是字調和語調的連接和平滑?根據我們的研究結果,答案是否定的。我們透過大批語音資料,證明語流(連續口語)的韻律其實是字調(tone)、語調(intonation) 和語篇韻律(Discourse Prosody)疊加後的代數和,是由小到大各個韻律範疇互動的結果,其中又以範圍最大的語篇韻律扮演的角色最為重要。強調語篇韻律的特性,不是各個字調或短語句調的個別表徵的集合,而是人在說話時,如何透過韻律相關性所表達的語篇語意關聯與一致性。提出「階層式多短語韻律句群架構」假說,透過量化分析多短語口語語篇敘述,提取韻律句群中各短語間的跨短語韻律相關性聲學參數,並透過線性回歸的統計分析方式,得到與階層韻律架構相對應的分層及疊加貢獻度,證明語篇中存有跨短語語篇韻律的基型,各大小韻律範疇的互動具系統性關係:短語調導致字調變異;語篇韻律導致短語調變異。因此字調和語調在語流中的變化,其實是受到上層語篇韻律的制約作用,造成位於不同的語篇位置中的字調和語調「不到位」的結果,表現出有別於詞組或孤立句的字調和語調特性。因此我們進而論述,對國語而言,語篇韻律是字調和句調以上的韻律成分;對句調而言,是造成動態變化的原因。只要釐清下層韻律單位在多短語韻律句群中的位置,便可預測字調和句調在語流韻律中的動態變化。



關於口語韻律研究,我的研究假設是,在超音段的層面,口語的語流韻律除了字調與句調以外,在句子 (sentence)之上,還有表達句子之間語意的連接(cohesion)與連貫(coherence)的單位「多短語語段multiple-phrase speech paragraph)」,管轄制約句法句調,系統性的造成句調變化以表達語篇訊息,因此,字調與句調都只是語段的次級韻律單位,語段則是語篇的次級單位,語流韻律實為語篇韻律。從韻律的階層性來說,字調由詞義規範、短語句調由句法結構規範,因此孤立字、孤立詞和孤立短語都是靜態韻律;語流韻律由語篇規範,造成短語句調系統性的變化,因此語篇韻律是動態韻律。

我研究口語韻律的單位是以口語語篇中的多短語語段為主,承襲了中國語言學之父趙元任先生以『大波浪小波浪』描述聲調與句調之間相互疊加或抵消關係的看法,提出了階層式的多短語韻律模架構,強調各短語間,因受來自上層訊息的管轄,而構築了相互間的韻律關係,因此語段裡的短語,好像兄弟姊妹一樣有了親屬關係,而不是將各短語之間視為不相關、各自擁有本身的句調的研究觀點。從認知與生理角度出發,針對語流韻律中跨短句的抑揚頓挫、音樂性及節奏,層層找出證據,提出多短語句群的的韻律結構,並以這個韻律結構提出跨句模組式的聲學模型。如此由上到下(top-down)的研究取向深獲日本學者藤崎博也先生所讚揚,認為此一研究方式乃語音分析之先鋒。

從2003年開始,我進行一系列口語韻律假說的驗證工作,我逐步透過自行開發的語料庫語音學的研究方法,以計量研究方式,分析大批語料的超音段聲學參數及停頓與邊界效應,在2004年提出五層的「階層式多短語韻律句群Prosodic Phrase Grouping簡稱PG」假說,得到語流中跨短語節奏與邊界效應的統計證據,證明了口語語流韻律中的音長分佈是動態的,來自字調、韻律詞、韻律短語和語篇各韻律層級貢獻的總和,支持了階層式的PG假說。此一說不但解釋了語流韻律的規範制約,來自語篇語意而非僅句法結構,並進一步解釋了字調及句法句調為何無法規範口語韻律以及語流句調為何如此多變的原因。隨後我陸續得到音強響度與邊界停頓的證據,因而得以在2005年提出PG的基型,及相對應的模組式聲學語音數學模型,並在Speech Communication以 Quantitative Prosody Modeling for Natural Speech Description and Generation的主題特刊中發表重要論文:”Fluent speech prosody: Framework and modeling”;同年我將PG階層向上擴大至語篇共為六層;2006年,我繼續得到PG的跨短語調型基型在基頻軌跡方面的量化證據,並將研究結果收錄於我應邀為Advances in Chinese Spoken Language Processing一書所寫的「韻律分析(Prosody Analysis)」章節中,至此我在每一個聲學語音層面,都已得到PG一說的證據。

「多短語語流韻律模組架構」所包括的四個聲學模組:基頻框架、節奏樣版、能量分分佈趨勢以及停延與停頓結構,可說是現今數學性的韻律模型中,最完整的聲學語音模型。這一理論不單解釋了為什麼短語句調並非最大的韻律單位,以及句調在語篇中韻律相關性,而且還可以此模型加以延伸、應用,對當今語音科技裡TTS在語流韻律的力有未逮之處,提出兼具理論性及應用性的具體建議。
此外,基於學術資源共享之理念,此段期間所收集的語流韻律語音資料庫,以及所開發的以語音感知為基礎的語料分析平台,也已編輯成一套資料庫,定名為:中央研究院口語韻律語料庫暨工具平台(Sinica COSPRO and Toolkit),在2006年1月1日正式對外發行(詳見http://www.myet.com/COSPRO),希望藉此促進國內語音研究之蓬勃與全面性發展。

回顧近十年來研究走向的轉變,以台灣地區通行的國語語篇的語流韻律現象作為研究對象,由傳統的聲調研究走向全面性的語流韻律現象之探索。在研究方法和研究結果兩方面皆有重大的突破。不僅開發語料庫語音學的研究方法;也拓展了傳統語音學的研究視角──從宏觀的角度,提出了階層式口語敘述語流韻律的完整架構,強調聲學語音的證據與模組,說明了語流韻律的旋律、節拍、強弱、停頓的結構是如何由各個韻律層的分層貢獻度疊加整合為口語語流的韻律表面現象;同時建立數學模型,直接對語音科技開發做出貢獻。

2009年1月6日 星期二

alveolo-palatal fricative

The voiced alveolo-palatal fricative is a type of consonantal sound, used in some spoken languages. The symbol in the International Phonetic Alphabet that represents this sound is ʑ, and the equivalent X-SAMPA symbol is z\. The closest representation in English is a voiced postalveolar fricative as in the word Asia

Contents [hide]
1 Features
2 Occurrence
3 See also
4 References
5 Bibliography



[edit] Features
Features of the voiced alveolo-palatal fricative:

Its manner of articulation is sibilant fricative, which means it is produced by directing air flow through a groove in the tongue at the place of articulation and directing it over the sharp edge of the teeth, causing high-frequency turbulence.
Its place of articulation is alveolo-palatal, that is, palatalized laminal postalveolar, which means it is articulated with the blade of the tongue behind the alveolar ridge, and the body of the tongue raised toward the palate.
Its phonation type is voiced, which means the vocal cords are vibrating during the articulation.
It is an oral consonant, which means air is allowed to escape through the mouth.
It is a central consonant, which means it is produced by allowing the airstream to flow over the middle of the tongue, rather than the sides.
The airstream mechanism is pulmonic egressive, which means it is articulated by pushing air out of the lungs and through the vocal tract, rather than from the glottis or the mouth.

[edit] Occurrence
Language Word IPA Meaning Notes
Abkhaz ажьа [aˈʑa] 'the hare'
Catalan[1] ajut ? 'help' See Catalan phonology
Japanese 火事/kaji [kaʑi] 'fire' Found in free variation with [dʲʑ] between vowels. See Japanese phonology
Kabardian лIыжь [ɬʼɪʑ] 'old man'
Polish[2] źrebię [ˈʑrɛbʲɛ̃] (help·info) 'foal' Also denoted by the digraph . See Polish phonology
Russian езжу [ˈjeʑːu] 'I drive' Most speakers. Usually written <жж> or <зж>. See Russian phonology


[edit] See also
List of phonetics topics

[edit] References
^ Recasens, Daniel and Pallarès, Maria Dolores (2001). "Coarticulation, Assimilation and Blending in Catalan Consonant Clusters". Journal of Phonetics 29(3): 273–301. doi:10.1006/jpho.2001.0139.
^ Jassem (2003:103)

[edit] Bibliography
Jassem, Wiktor (2003), "Polish", Journal of the International Phonetic Association 33 (1): 103-107

voiceless labial-velar fricatives

From Wikipedia, the free encyclopedia
Jump to: navigation, search
IPA – number 175
IPA – text ɧ
Entity ɧ
X-SAMPA x\
Kirshenbaum x^ or S~
Sound sample (help·info)
The so-called voiceless palatal-velar fricative (also called a voiceless dorso-palatal velar fricative, voiceless postalveolar and velar fricative, or voiceless coarticulated velar and palatoalveolar fricative) covers a range of similar sounds used in most dialects of Swedish and transcribed as /ɧ/. In Swedish, they are commonly referred to as the sje-sound, based on one of the more common spellings. They have very similar acoustic properties, not dissimilar from the /ç/ sound of southern Dutch or German ch after e or i, and feature distinct labialization. The International Phonetic Association describes /ɧ/ as "simultaneous ʃ and x", but this claim is disputed among phoneticians, including at least one former president of the IPA.[1]

Contents [hide]
1 Features
2 Occurrence
3 References
4 See also



[edit] Features
Features of the voiceless palatal-velar fricative:

Its manner of articulation is fricative, which means it is produced by constricting air flow through a narrow channel at the place of articulation, causing turbulence.
Its place of articulation is disputed (see below).
Its phonation type is voiceless, which means it is produced without vibrations of the vocal cords.
It is an oral consonant, which means air is allowed to escape through the mouth.
It is a central consonant, which means it is produced by allowing the airstream to flow over the center of the tongue, rather than the sides.
The airstream mechanism is pulmonic egressive, which means it is articulated by pushing air out of the lungs and through the vocal tract, rather than from the glottis or the mouth.

[edit] Occurrence
This sound has been reported in certain dialects of Swedish, where it is most often known as the "sje"-sound.

Its place of articulation varies over Swedish regions, and is not agreed upon (see references). It has been variously found to be:
velar and postalveolar, meaning it is articulated simultaneously with the tongue dorsum (i.e. the back part of the tongue) approximating the velum (i.e. the soft palate) (like [x]) and just behind the teeth (like [ʃ]). However, doubly articulated fricatives are very difficult to pronounce or to hear, and many linguists doubt that they exist.[1]
Lindblad describes one of two common variants of Swedish /ɧ/ as labiodental with simultaneous velarization and protrusion of the upper lip, which would be transcribed as [fˠʷ]. (The English sigh of relief "phew!" is probably the closest approximation, resembling the voiceless labial-velar approximant that is used in some varieties of American English.) He does not use the symbol <ɧ> for this allophone.
Lindblad describes the second common variant of Swedish /ɧ/ as velar. The difference between it and [x] is not clear, but it may have less friction ([x̞]), or be further forward ([x̟]), or both.
A number of intermediate possibilities between these extremes.
Other articulations have been described as well, with no obvious standard emerging.
Consider the following comments by Peter Ladefoged and Ian Maddieson:

Some dialects of Swedish have a fricative that has been said to have two or even three articulatory constrictions (Abercrombie 1967). We do not, however, think it is correct for more than one of these constrictions to be considered a fricative articulation. There is good data available on the Swedish sibilant fricatives (Lindblad 1980) allowing us to consider these sounds in detail.
[…]
The […] Swedish fricative, usually symbolized by ɧ, is the most interesting. Lindblad describes two common variants of Swedish ɧ. The first, for which he uses a different symbol, he calls a highly rounded, labiodental, velar or velarized fricative. […] Lindblad suggests that the source of frication is between the lower lip and the upper teeth, and it certainly appears to be so from his x-ray. He also demonstrates that the upper lip is considerably protruded in comparison with its position with that in the gesture of i. In addition to these anterior gestures, Lindblad notes that the "tongue body is raised and retracted towards the velum to form a fairly narrow constriction. (The presence of this constriction is constant, but not its width or location, which vary considerably.)" The posterior constriction in this variety of ɧ is not great enough to be itself a source of turbulence, so that, although this sound may have three notable constrictions, one in the velar region, one labiodental, and a lesser one between the two lips, only the labiodental constriction is a source of friction.
The second common variant of Swedish ɧ […] is described by Lindblad as a "dorsovelar voiceless fricative" pronounced with the jaw more open and without the lip protrusion that occurs in the other variety. Lindblad suggests that the difference between this sound and the more usual velar fricative x is that the latter "is formed with low frequency irregular vibrations in the saliva at the constriction" (Lindblad 1980, our translation). We infer from his descriptions and diagrams that this variant of ɧ has less frication, and may be slightly further forward than the velar fricative x commonly found in other languages. Lindblad claims that between the extreme positions of the labiodental ɧ and the more velar ɧ, "there are a number of intermediate types with various jaw and lip positions, including some with both anterior and posterior sound sources." [W]e doubt that it is possible to produce turbulence at two points in mouth simultaneously for ordinary linguistic purposes.

The most well-known case [of a possible multiply-articulated fricative] is the Swedish segment that has been described as a doubly-articulated voiceless palato-alveolar-velar fricative, i.e., ʃ͡x. The IPA even goes so far as to provide a separate symbol for this sound on its chart, namely ɧ. The sound in question is one variant of the pronunciation of the phonological element ʃ, which is highly variable in Swedish dialects, receiving pronunciations ranging from a palatalized bilabial sound to a velarized palato-alveolar one to a fully velar one. [I]t is not clear that any of the variants is actually a doubly-articulated fricative.

– Ladefoged & Maddieson 1996:171–172, 330


[edit] References
^ a b Peter Ladefoged, The Sounds of the World's Languages.
Abercrombie, David. (1967). Elements of general phonetics. Chicago: Aldine.
Garlén, Claes (1988), Svenskans fonologi : [i kontrastiv och typologisk belysning] Lund : Studentlitteratur, 1988 [Ny utg.]
Ladefoged, Peter; & Maddieson, Ian. (1996). The sounds of the world's languages. Oxford: Blackwell Publishers.
Lindblad, Per. (1980). Svenskans sje- och tje-ljud i ett Allmänfonetisk Perspektiv. Travaux de l'Institut de Linguistique de Lund 16. Lund: C. W. K. Gleerup.

Voiced palatal implosives

From Wikipedia, the free encyclopedia
Jump to: navigation, search
IPA – number 164
IPA – text ʄ
Entity ʄ
X-SAMPA J\_<
Kirshenbaum J`
Sound sample (help·info)
The voiced palatal implosive is a type of consonantal sound, used in some spoken languages. The symbol in the International Phonetic Alphabet that represents this sound is ʄ, and the equivalent X-SAMPA symbol is J\_<. The IPA symbol can be considered to be a lowercase letter esh with a horizontal stroke, a turned lowercase letter f with a rightward hook protruding from the upper left, or a dotless lowercase letter j with a horizontal stroke (the symbol for the voiced palatal plosive) and a rightward hook (the diacritic for implosives).


[edit] Features
Features of the voiced palatal implosive:

Its manner of articulation is plosive or stop, which means it is produced by obstructing airflow in the vocal tract.
Its place of articulation is palatal which means it is articulated with the middle or back part of the tongue raised against the hard palate.
Its phonation type is voiced, which means the vocal cords are vibrating during the articulation.
It is an oral consonant, which means air is allowed to escape through the mouth.
It is a central consonant, which means it is produced by allowing the airstream to flow over the middle of the tongue, rather than the sides.
The airstream mechanism is glottalic ingressive, which means it is produced by pulling air downward with the glottis, rather than pushing it out.

[edit] Occurrence
Language Word IPA Meaning Notes
Swahili jana [ʄana] 'yesterday'
Sindhi ڄرو [ʄaro] 'cobweb'

Bilabial Clicks

From Wikipedia, the free encyclopedia
(Redirected from Bilabial click)
Jump to: navigation, search
IPA – number 176
IPA – text ʘ
Entity ʘ
X-SAMPA O\
Kirshenbaum p!
Sound sample (help·info)
The bilabial clicks are a family of click consonants found as phonemes only in the Tuu family, in the ǂHõã language of Botswana, in a single word in Hadza, and in the Damin ritual jargon of Australia.

The symbol in the International Phonetic Alphabet that represents the forward articulation of these sounds is ʘ. This must be combined with a symbol for the rear articulation to represent an actual speech sound. Attested bilabial clicks include but are not limited to:

[k͡ʘ] or [ʘ͡k] voiceless velar bilabial click (may also be aspirated, ejective, affricated, etc.)
[ɡ͡ʘ] or [ʘ͡ɡ] voiced velar bilabial click (may also be breathy voiced, affricated, etc.)
[ŋ͡ʘ] or [ʘ͡ŋ] nasal velar bilabial click (may also be voiceless, aspirated, etc.)
[q͡ʘ] or [ʘ͡q] voiceless uvular bilabial click
[ɢ͡ʘ] or [ʘ͡ɢ] voiced uvular bilabial click (commonly prenasalized)
[ɴ͡ʘ] or [ʘ͡ɴ] nasal uvular bilabial click
[ʘ͡ʔ] glottalized bilabial click
The last is what is heard in the sound sample at right, as non-native speakers tend to glottalize clicks to avoid nasalizing them.

Damin also had an egressive bilabial [k͡ʘ↑], the world's only attested egressive click

Features
Features of ingressive bilabial clicks:

The manner of articulation is a noisy, affricate-like release.
The rear closure may be voiced, nasal, ejective, or affricate, and have any of several phonations.
The forward place of articulation is bilabial, which means it is articulated with both lips.[1].
The rear place of articulation may be either velar or uvular.
Bilabial clicks may be either oral or nasal, which means air is allowed to escape either through the mouth or the nose.
They are central consonants, which means they are produced by allowing the airstream to flow over the middle of the tongue, rather than the sides.
The airstream mechanism is lingual ingressive (velaric ingressive), which means the pocket of air trapped between the two closures is rarefied by a "sucking" action of the tongue, rather than by the glottis or the lungs. The release of the forward closure produces the 'click' sound. (One of the two labial clicks in Damin is velaric/lingual egressive, which means the air spurts out into the mouth between the lips under the pressure of the tongue.)
The bilabial clicks are sometimes erroneously described as sounding like a kiss. However, they do not have the pursed lips of a kiss (that is, they're not rounded). Instead, they have an articulation more like like that of a [p], and sound more like a smack of the lips.

The egressive click differs from the above in that the trapped air pocket is compressed by the tongue until it is allowed to escape through the forward articulation.


[edit] Symbol
The bullseye or bull's eye (ʘ) symbol used in phonetic transcription of the phoneme was made an official part of the International Phonetic Alphabet in 1979, but had existed for at least 50 years earlier. It is encoded in Unicode as U+0298 LATIN LETTER BILABIAL CLICK.

Similar graphemes consisting of a circled dot encoded by Unicode are:

Gothic 𐍈 ƕair
astronomical symbol ☉ "Sun"
mathematical operators ⊙ "circled dot operator" and ⨀ and "n-ary circled dot operator"
geometrical symbol ◉ "fisheye"
A symbol created for the IPA,  (a turned b with a tail) was never widely used and was eventually dropped for ʘ.


[edit] Notes
^ Sometimes this may pass through a labiodental stage as the click is released, making it noisier (Ladefoged & Maddieson 1996:251).

[edit] References
Pullum, Geoffrey K.; Ladusaw, William A. (1996). Phonetic Symbol Guide. University of Chicago Press. pp. 132–133.
Ladefoged, Peter; Maddieson, Ian (1996). The Sounds of the World's Languages. Blackwell Publishers. pp. 246–280.