本研究旨在探討漢語失語症病人於字長及句子基頻兩方面的掌控能力，並試圖檢驗以音律(prosody)表現完好與否來區分Broca氏及Wernicke氏兩種失語症病人的有效性，以期對漢語失語症病人的音律表現有進一步認識。 　　我們針對六個音律特性(prosodic features)進行聲學分析。實驗結果顯示不同音律特性的損壞其彼此間是可分離的(dissociable)。此外，就病人的音律表現而言，Broca氏病人的音律並非全面性的損壞，他們仍能掌控某些音律特性的表現（如句尾基頻的下降趨勢）；而另一個方面，Wernicke氏病人未如一般預期保有完整的音律表現，其語言優勢半球的操作連帶影響了這些病人於字長與語句基頻的掌握能力。因此，若僅以音律一詞來界定，而不釐清是以音律的那些特性作比較，未免失之籠統，以致無法作出有效的失語症分類判斷。 / According to clinical observations, Broca's aphasics are characterized by dysprosody, while wernicke's aphasics are regarded as having normal prosody. From the crosslinguistic viewpoint, the present study aims to examine chinese aphasics' ability in f₀ programming and timing control and to assess if the prosodic dichotomy between dysprosody of Broca's aphasics and normal prosody of wernicke's aphasics is valid for aphasics of tone languages, i.e., chinese aphasics. 　　By means of acoustic analysis, six prosodic features are investigated: sentence-final lengthening, F₀ range, sentence-final peak-to-valley F₀ fall, the breath-group pattern, the declination pattern, and the P1 effect. The results indicate that Broca's subjects exhibit normal tendency of sentence-final peak-to-valley F₀ fall, though they have deficits in producing other prosodic features. On the other hand, wernicke's aphasics have disturbances in controlling F₀ range, and producing the declination and the breath-group patterns and sentence-final lengthening, though they remain intact in producing other prosodic features. The information obtained here points out the separate disruption of prosodic features in aphasic speech and disputes the validity of differentiating Broca's from wernicke's aphasia in terms of "dysprosody" or "normal prosody". The problem with this prosodic dichotomy is that the concept about prosody is used in a too general way. It is suggested that when we use prosody as the defining symptom, we should specify which aspect of it is involved to make the classification viable.

人聲大致上可分為語音和非語音兩部分。傳統上對於聲音分類的研究大多強調語音、音樂和環境聲的分類。在本論文中，我們採取不同的觀點，著重於人聲中非語音部份的研究，聲音種類為笑聲、尖叫聲、打噴嚏聲和鼾聲。為了達到此目標，我們調查常用的幾種聲學特徵，並以多元適應性雲形迴歸和支持向量機進行特徵值篩選，找出對於非語音人聲分類具有代表性的聲學特徵。此外我們也進行多方面的模擬，以觀察雜訊對辨識率的影響。 本論文第二部份為鼾聲研究，我們比較一般普通麥克風和目前醫療用鼾聲麥克風(snoring microphone)、壓電感應器(piezo sensor)三者在偵測鼾聲上的表現。此外，並以KL divergence 和EMD兩種計算差異度的方法進行普通鼾聲與阻塞型鼾聲的分群。同樣地，我們加入不同程度雜訊至鼾聲訊號中，以測試兩方法抗雜訊的穩健度，結果顯示此兩種方法均有不錯的表現，其中EMD在大多數情況下有較佳的結果。 / Human sounds can be roughly divided into two categories: speech and non-speech. Traditional audio scene analysis research puts more emphasis on the classification of audio signals into human speech, music, and environmental sounds. We take a different perspective in this thesis. We are mainly interested in the analysis of non-speech human sounds, including laugh, scream, sneeze, and snore. Toward this goal, we investigate many commonly used acoustic features and select useful ones for classification using multivariate adaptive regression splines (MARS) and support vector machine (SVM). To evaluate the robustness of the selected features, we also perform extensive simulations to observe the effect of noise on the accuracy of the classification. / The second part of this thesis is concerned with the analysis snoring signals. We use ordinary microphone as our snoring recorder and compare its sensitivity with snoring microphone and piezo sensor, which are often utilized in clinical settings. In addition, we classify simple snores and obstructive snores using two distance measures: KL divergence and earth mover's distance (EMD). Similarly, we add noises to the snoring signals to examine the robustness of these two measures. It turns out that both methods perform satisfactorily, although EMD generates slightly better results in most cases.

本論文為以共振峰頻率資料來研究海陸腔客語母音音質之聲學研究。 語言學的文獻中，以共振峰頻率資料來研究世界上各種語言的母音的音質已經有相當豐富的文獻，在過去五十年來，前人在客語的語音及音韻系統上已經有豐富的研究成果。但以聲學資料來對海陸客語做描述的文獻相當缺乏。 本研究的受試者為居住在新竹縣新埔鎮，六位以海陸腔客家話為母語的人士，其中有三名男性及三名女性。本研究使用了32個測試字，包含了單元音、雙合元音及三合元音。語音資料是利用KAY CSL 4100 (KAY Electronics)來做分析。本研究測試字包含的音節結構有:CV, CVV, CVVC, CVC, 以及CVVV。測試字在實驗的過程中，被受試者單獨發音，或是放在句子中間。在實驗資料完成分析後，我們以Origin 6.0軟體繪出海陸腔客語母音的聲學空間圖。 根據供振峰研究資料，本研究得到以下有關海陸腔客語母音音質的表現。首先關於單元音，海陸腔客語共有六個單元音: [i], [e], [ɨ], [a], [o], 以及 [u]。其次，海陸腔客語共有十一個雙元音: [ie], [ia], [io], [iu], [eu], [ai], [au], [oi], [ui], [ue], 以及 [ua]。我們比較單元音及雙元音的聲學空間圖(vowel space)後，發現相同的母音在雙母音的環境下，由於受到鄰近母音發音位置的影響，在聲學空間上，比起在單母音的時候，變的更高低，或更前或後。第三，三母音[iai], [iau]和[uai]在聲譜圖上的表現，第一個母音[i]和[u]都比其後的兩個母音，長度來的相對的短一些，表現很像是滑音[j]和[w]。 最後，本研究以共振峰聲學頻率的形式描述及記錄了當代海陸腔客語母音的音質，前人的研究中，以楊時逢(1952)最能說明本研究所提供的聲學資料。希望這些資料能對海陸腔客語母音的研究有所貢獻。 / This thesis is an acoustic study that investigates the vowel quality of Hai-lu Hakka vowels with formant frequency data. The acoustic method has been widely applied to the study of vowel quality of languages worldwide. In the past 50 years, Hakka researchers have yielded rich results in phonological system and phonetic description of Hakka, but there are relatively fewer research focus on the acoustic properties and characteristics of vowel phones in Hai-lu Hakka. The subjects of this study included three male and three female native Hai-lu Hakka speakers. Testing items used in this study were 32 syllables that involve monophthongs, diphthongs, and triphthongs. The speech data were analyzed by using KAY CSL 4100 (KAY Electronics). The data discussed in this study were testing items produced in citation form and sentence form, including the following syllable structures: CV, CVC, CVV, CVVC, and CVVV. The vowel qualities of Hai-lu Hakka vowels were measured and analyzed, and the acoustic vowel space of Hai-lu Hakka is plotted by the software Origin 6.0. Several results concerning the vowel quality of Hai-lu Hakka vowels were reported based on the formant frequency data. Firstly, there are six monophthongs in the vowel system of Hailu Hakka: [i], [e], [ɨ], [a], [o], and [u]. Secondly, there are eleven diphthongs in the vowel system of Hai-lu Hakka : [ie], [ia], [io], [iu], [eu], [ai], [au], [oi], [ui], [ue], and [ua]. By comparing the relative position of the vowel in a diphthong and the corresponding vowel as in a monophthong, we found that the second vowel in a diphthong tend to be higher or lower, more fontal or back, and it is possibly due to the coarticulatory influence of the adjacent phones.the Thirdly, as shown in the spectrogram of the three tripthongs [iai], [iau] and [uai], the duration of [i] in [iai55], [i] in [tʰiau55] and [u] in [kuai55] are relatively shorter than the rest two vowels in the same syllable, as [ai] in [iai55], [au] in [tʰiau55] and [ai] in [kuai55]. In these CVVV syllables, [u] and [i] are similar to glides or so-called semi-vowels, or approximants [j] and [w] as shown in the spectrogram. Finally, the vowel system in Yang (1957) is more suitable for accounting for the data in this study. Hopefully, the vowel formant data presented in this study will contribute to the study of vowels in Hakka.

在語言學文獻上，對於音段在連併音節上的研究已非常眾多且詳細。而目前也有不少學者開始處理音節連併的超音段(聲調)部分，但是大多數研究是從音韻學的觀點。本文試圖從聲學語音學的角度，檢驗連併音節的產生與說話者的速度、說話者對內容熟悉度之間的關係；及分析連併音節的基頻與其音節長度。實驗統計結果證實說話者的速度與內容熟悉度的確與連併音節成正相關。三名以台灣地區的國語為母語的受試者，被要求分別錄製五段短文，各短文裡分別有數十個可能產生雙音節連併的詞組。在連併音節的基頻分析結果顯示，呈現出些微降調及升調兩種調型。我們並且發現，連併前的聲調起始值(onset)為產生連併聲調的最佳基本單位，我們也歸納出兩條規則來加以解釋。另外，在連併音節的長度分析結果顯示，連併音節所需的時間單位比起一般音節更短。綜合以上結果，我們暸解除了可經由規則推論出連併音節的調型之外，在語音的面向上，這些連併聲調皆呈現出較窄的調域、居中的調值和較短時間等的特徵。因此從本研究的結果顯示，我們較傾向支持連併音節是類似於輕重音表現的說法，而不再局限於本身的調值的衍算與否。另外，對於在我們分析過程所發現伴隨著連併音節的鼻音現象，證明了一般認為在快速語流中，共同發音所產生的自然現象，也給予了認為鼻音是另屬一個自主音段的說法更進一步的實證支持。 / Segment deletion or merger in contracted syllables has been well understood in the literature. Several researchers have also been investigating the tones in contracted syllables. However, most of their researches are based on phonology. The present study analyzes the tones in disyllabic contraction by examining its relationship with speech rate and familiarity. It also analyzes the fundamental frequency (F0) and duration of the contracted tones. An experiment was conducted to explore the idea that familiarity with the test materials and the rate of speech correlate positively with the occurrences of tone contraction. Three native speakers of Mandarin Chinese (in Taiwan region) recorded five paragraphs, each of which contains dozens of possible disyllabic contractions. Analysis of F0 reveals that there are basically two types of tonal shapes of contracted syllables—slightly falling and rising; and the onsets of the original syllable tonal units are the most adequate portions of forming a contracted tone. Two governing rules were generalized from various the findings of the experiments. Results of duration indicate less time is taken by a contracted syllable than an uncontracted one. The findings of this study provide support for the view that the contours of the contracted tones are predicted by rules. They also show that much of the variability of F0 contours in contracted syllables exhibits a fixed tonal pattern (narrower pith range, more middle tonal height, and shorter length) and that contracted tones has an accentual character in essence. In addition, the nasality accompanying contracted syllables in this study provides testimony to the natural co-occurrence of nasal and oral features in fast speech and at the same time also give support to the hypothesis that nasality acts on a separate autosegmental tier.

本論文為以共振峰頻率資料來研究台灣華語母音音質之聲學研究。 本研究的受試者為六名生長在台灣台北市的男性華語母語使用者。本研究使用的測試字為70個包含所有韻母型態的單音節字。語音資料是利用KAY CSL 4100 (KAY Electronics)來錄製及分析。本研究所採用的資料為獨立發音且無聲母的測試字。 根據共振峰頻率資料本研究得到以下有關台灣華語母音音質的發現。首先研究的部分為音標所隱含的語音對比。在齒槽（alveolar）及後齒槽（post-alveolar）組的舌尖母音（apical vowels）間並沒有發現音質上的顯著差別，但統計及其它證據支持低母音裡的前後對比及中母音裡的半開（open-mid）及半閉（closed-mid）對比的存在。 其次，研究的結果也包括語音環境中的音段對於母音音質的影響。結果發現韻母中的元素，包含母音前介音、母音後介音、及韻尾鼻音，都對母音有不同的語音制約（conditioning）。另外，本研究發現順同化（progressive assimilation）比逆同化（regressive assimilation）對於低母音音質有較大的影響。 第三，本研究藉由與之前聲學研究的結果作比較來討論台灣母音音質的改變。結果發現硬顎及後硬顎組的舌尖母音間的音質差異隨著時間縮小終至於消失。此外，研究發現半閉母音及半開母音在母音高度及母音前後相對位置出現反轉。共振峰頻率的比較顯示這種反轉的原因為半閉母音[e]在音質上的改變。然而，我們需要進一步的研究才能對這些音質的改變有合理的解釋。 最後，本研究以共振峰頻率的形式描述及紀錄了當代台灣華語表層母音的音質。希望這些資料能對台灣華語母音的研究有所貢獻。 / This thesis is an acoustic study that investigates the vowel quality of Taiwan Mandarin vowels with formant frequency data. The subjects of this study included six male native Mandarin speakers born and raised in Taipei city, Taiwan. Testing items used in this study were 70 syllables that involve all combinations of segment in the syllable final. The speech data were recorded and analyzed with KAY CSL 4100 (KAY Electronics). The data discussed in this study were testing items produced in citation form without initial consonant. The vowel qualities of Taiwan Mandarin surface vowel were measured and analyzed. Several results concerning the vowel quality of Taiwan Mandarin surface vowel phones were reported based on the formant frequency data. Firstly, the distinction between vowel phones implied by their transcription was examined. The difference in vowel quality was not observed in the alveolar and post-alveolar groups of apical vowels, while the front-back distinction in low vowels and the open-mid and closed-mid distinction in mid vowels were found with statistic and other evidence. Secondly, the conditioning of segments in the neighboring environments on the vowel quality was investigated. It is found that elements in the syllable final, including pre-nuclear, post-nuclear glide and coda nasals, generally have different types of conditioning on the vowel. In addition, the progressive assimilation was found have greater influence on the vowel quality of low vowel phones than the regressive assimilation. Thirdly, the change in vowel quality was discussed through the comparison with results in previous acoustic studies. The difference in vowel backness between the alveolar and post-alveolar groups of apical vowels were found to be decreasing and eventually disappeared. Besides, the relative vowel height and backness of closed-mid and open-mid vowel were found to be in reverse compared with that in the literature. Comparison in formant frequencies showed that this reverse lay in the change of vowel quality in the close-mid [e]. However, further investigations were needed to offer reasonable explanations for the change. Finally, the vowel qualities of present Taiwan Mandarin surface vowel phones were described and recorded as the formant frequency data presented in this study. Hopefully, the data could contribute to the study of vowels in Taiwan Mandarin.

本文旨在比較台灣閩南語常人、弛緩型(flaccid dysarthrics)、與痙攣型(spastic dysarthrics)神經性構音障礙病患子音絕對及相對音長之差異，並探討此病態語音在了解說話運動系統之特徵以及了解語言產製過程上可能的暗示。 結果顯示神經性構音障礙病患的子音絕對音長顯著長於常人受試者，但兩組病患延長的型態並不相同。首先，弛緩型病患子音音長顯著長於痙攣型病患；再者，以子音的帶音與否、發音位置、發音方法這三個向度來看，弛緩型病患之子音音長展現了全面性的延長，而痙攣型病患則產生了選擇性的延長，其子音之延長，大多集中在有聲子音上，這些差異乃因兩組病人病理狀況不同所致。 另一個結果則顯示，雖然神經性構音障礙病患，在子音音長呈現了延長的狀況，但在子音間相對音長次序上，仍保持到某種程度的完好，可是，子音雖保持一定的次序，但彼此間音長的差值及比值卻也發生了改變。而子音在音節中的所佔比例卻又和正常人無異。 因此，本文推斷，在常人子音的產製上，是要求精確的絕對音長，而且子音間音長亦要保持一定的相對關係；但若因病變致使說話者無法同時兼顧二者時，說話者會選擇犧牲絕對音長，而仍保持子音音長間的相互關係。這可能是語言產製與感知的一些經濟原則(principles of economy)互動所導致的結果。此外，構音障礙病患表就子音在音節中的相對時長這個向度上，和常人無異，這可能是語言文法與生理限制二項因素互動所致。 / This study aims (1) to investigate the absolute and relative timing of consonant duration in Taiwanese neurogenic dysarthric speech by means of acoustic measurements, (2) and to explore their implication on the characteristics of motor system and on the status of timing control in speech production. The results show that absolute consonant duration in flaccid and spastic dysarthric speech are significantly longer than that in normal controls. However, the "lengthening" phenomena display different patterns in flaccid and spastic groups. First, for the patients with the same degree of muscular dysfunction, the flaccid dysarthrics lengthen the consonant duration significantly greater than the spastic ones do. Second, whereas absolute consonant duration in flaccid group shows overall lengthening regardless of voicing state, stricture type, and place of articulation, that in spastic group displays selective lengthening. Besides, though the absolute consonant duration is lengthened in dysarthric speech, the durational opposition between consonants is more or less distinct. However, only the durational ranking between consonants (external timing relation) is maintained in dysarthric, but the durational distance or ratio between consonants (internal timing relation) is modified. Relative timing of consonant duration, that is, the consonant-to-syllable ratio (cs-ratio) within a syllable remains intact in dysarthric speech even though the segmental duration is lengthened. Lengthened absolute consonant duration in dysarthric speech is accounted for by the neurological deficits and compensatory effects. Longer absolute consonant duration in flaccid than in spastic speech is attributed to slowness and weakness without spasticity of speech musculature in flaccid dysarthria. Overall lengthening of consonant duration in flaccid group results from the entirely-impaired cranial nerves innervating speech musculature, whereas selective lengthening of consonant in spastic group is ascribed to spasticity and the selective pattern as well as directionality of neuromuscular impairments in spastic dysarthria. The reserved durational ranking between consonants in dysarthric speech may result from the compromise of principles of economy in speech production and perception. The normal-like CS-ratios in dysarthric speech stem from the interaction between biological constraints and the grammar of a language. Based on the data collected from dysarthric speech, this study discusses the temporal variance as well as invariance in speech motor system and the status of timing in the grammar of a language. The importance of principles of economy in speech production and perception is also indicated.

國語二聲在文獻上多被描述為一調值35的高升調。然而近來不少學者發現台灣地區的國語二聲在閩南語/國語雙語語者族群中有變異的現象：一是323的降升調，二是接近31的低降調。文獻顯示這可能與語言背景(Lo, 2004)和地區(H. J. Hsu,2004)有關。然而對於國語單語語者的研究並不多。此外，性別、母音、聲調環境和句中位置也有可能影響聲調的變異。因此，本研究旨在調查台灣地區的聲調變異現象，藉由語音學實驗的方式，探究國語二聲的變異和語言背景（閩語/國語雙語語者或國語單語語者）的關係。並進一步探討其與社會因素：地區（北、中、南）、性別和語言學因素：母音、聲調環境和句中位置的關係。受試者以17-27歲的年輕族群為主。結果發現台灣地區的年輕人普遍以平調（中平調或低平調）(level)作為二聲變異的主要變體，另一較少數的變體則為低升調(low-rising)，低降調在本研究中並不明顯。這些二聲變體有可能是受到整體調域向下縮窄的影響。而閩語/國語雙語語者的確發出較多的聲調變異，特別是低升調，這可能是受到閩語聲調系統的影響。地區、性別，以及母音、聲調環境和句中位置都會影響二聲變異。本研究進一步證實二聲的聲調變異在台灣年輕族群中的發展，支持了台灣地區國語的調域縮窄的現況。 / Mandarin Tone 2 is canonically described as a high rising tone. However, recently, T2 in Taiwan was found to have other contour types: the dipping contour (323) and falling (31) variants (Fon, 1997; H. J. Hsu, 2004), with the former being considered as a normal rising tone, but the later totally losing the rising feature. Those were all found in Taiwanese (Tw)-Taiwan Mandarin(TM) bilinguals. The reasons causing this tonal variation is supposed to be relating to the language backgrounds and other social and linguistic factors from previous studies. However, no study has confirmed those correlated factors, especially the comparison between Tw-TM bilinguals and TM monolinguals. This study thus aims to discover what the T2 variation is and where it occurs. By sampling 24 subjects, aged 17-27, with half Tw-TM bilinguals and half TM monolinguals, we conducted three experiments to test the T2 in different contexts. Results showed that instead of the falling variant, more speakers produced a level variant and a small amount of low-rising variant. Besides, the bilinguals did produce more T2 variations than the monolinguals. The results indicated that language background does affect the T2 variation among young speakers in Taiwan. Moreover, the high percentage of the Level variant support the finding of a narrower tonal range of TM, compared to the standard Mandarin. Other factors, such as genders, regions, vowels, tonal environments and sentence positions were also found to have influence on T2 variations.