The Acoustic Characteristics of Hakka Consonants and Vowels

• Main Authors: Liang Chiou-wen, 梁秋文
• Other Authors: Chung Ruang-fu
• Format: Others
• Language: en_US
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/05942158409960269577

碩士 === 國立高雄師範大學 === 台灣語言及教學研究所 === 93 === ABSTRACT: The researches on Hakka in the past 50 years have yielded rich results in phonetic description, phonological characterization, morphological structure, syntactic components and so on. However, the researches based on acoustic or experimental approaches are not very rich and hence there is a gap waiting for further studies. This study aims to describe the acoustic characteristics and set up a models of acoustic nature of Hakka vowels and consonants. The findings can offer the resources for the speech recognition, production, perception, and synthesis, or provide references for Hakka language teaching. There are two major Hakka sub-dialects spoken in Taiwan: Hai-lu and Si-xian. This study focuses on the Si-xian Hakka spoken in Meinung. 20 subjects (10 males and 10 females) are involved in this study. All of them are fluent native speakers of Si-xian Hakka from Meinung. We collected the speech data in the fieldwork and fetched the information of the acoustic cues such as the first and second formant frequencies, Voice Onset Time (VOT), release burst, turbulence noise, nasal murmur and formant transition by the speech analyzing program: PRAAT. And the data and information are demonstrated graphically by the professional graphing software: Microcal Origin 6.0. There are six inventories of the test stimuli. The first inventory of stimuli focuses on the F1 and F2 frequencies of vowels. The second inventory of stimuli aims to investigate the VOT, release burst, turbulence noise of consonants and the vowel-consonant transitions. The third inventory of stimuli aims to examine the friction preceded the zero-initial syllables of /i/ and /u/. The fourth inventory of stimuli aims to probe the syllabic nasals of Hakka. And the fifth and sixth inventories of stimuli focus on the transitions of Hakka diphthongs and triphthongs. The results of this study show that the Hakka vowel /i/ is the most front and highest vowel (even higher than /u/), and the vowel /ii/ is a central and high vowel (approximate to the height of /u/), and the vowel /a/ is a central and low vowel. And the higher vowel has higher F0 frequency. In addition, compare to the unchecked tones, the tongue positions of six vowels on checked tones seem to have the tendency of neutralization. In general, the Hakka vowels /i, e, a, o, u/ are similar to the English vowels /i, e, a, o, u/ (Ladeforged & Maddieson 1996). Only Hakka /a/ is more front than English /a/. The stop /p/ (12 ms) has the shortest VOT, and /t/ (14.7 ms) is a little longer than /p/, and /k/ (27.8 ms) is longer than /t/. And the affricate /ts/ (64.7 ms) is shorter than /tsi/ (73.3 ms). The duration of aspiration of ph-p is 61.4 ms, th-t is 59.6 ms, kh-k is 59.4 ms, tsh-ts is 70.5 ms, and tsih-tsiis 73.5 ms. It seems the duration of aspiration of ph, th, kh is about 60 ms, and duration of aspiration of tsh, tsih is about 70 ms. Both the major region of noise energy and the high energy peak for the alveolar fricative /s/ are higher than the palatal fricative /si/. The fricative /f/ is voiceless, and /v/ is voiced. The F2 frequency increases during the transitions from labial consonants to vowels, and increases from coronal consonants to /i, ii, e/ but decreases from coronals to /a, o, u/, and decreases from velar consonants to vowels. The transitions are smooth and level between palatalized consonants and /i/. The F2 frequencies decrease during the transitions from vowels to /m/ and /p/ (inversely to the transitions from /m/ and /p/ to vowels). The F2 frequencies decrease during the transitions from /i, ii, e/ to /n, t/, but they increase for the /a, o, u/-/n, t/ transitions. The F2 frequencies increase during the transitions from vowels to /ŋ, k/.