カールゲル著ケート語初等読本およびドンネル収録音声資料試論

概要

論文の詳細を見る
In this paper we reviewed the first Latin alphabet primer of the Ket language, written by N.K. Karger in 1934, and also looked into the phonogrammic sound recordings of the languages of the Siberian indigenous minorities, made by Kai Donner in 1912-1913 and 1914, from the viewpoints of cultural anthropology and phonetics and proposed a talker identification method. Ket, genetically still unknown, is a minority language spoken by a population of approximately one thousand persons, who live in the areas along the Yenisey and its numerous tributaries. Before the Revolution Ket had no form of writing. In 1934 the first Latinized primer of Ket, which was replaced by a Cyrillic version in 1938, was published by the then Soviet authority and helped liquidate illiteracy among Ket people. The new writing system consisted of 28 Latin letters, a single Cyrillic one, and two kinds of diacritic symbols, each of which was designed to be easily learned by phonetically naïve Ket speakers. Most of the hand-drawn pictures and the texts portrayed their natural surroundings costumes, tent-houses, boat-houses, routine work, people, and shamanism. In short, they produced something like a biography of the indigenous people. In reviewing Ket, we encountered a question concerning how to interpret a glottal stop sound occurring as an allophone in this language. Our answer is that if the sound spectrographic pattern of energy seen in the interval between a glottal stop sound and its following consonant is proven to be created by an excessive glottalization, this interval can be interpreted as a trace of the syllable that once might have existed there. In looking into the phonogrammic sound recordings made by Kai Donner, we made a series of averaged spectral analysis on 28 speech samples from four linguistically different groups, I.e. Ket (Paleo-Yenissey), Samoyedic (Uralic), Kamassian (Uralic), Turkish (Altaic), Tataric (Altaic), and Russian (indo-European). It was discovered that all of the 28 speech samples showed a very similar acoustic energy pattern in which energy was attenuated around the five freqencies values of 0.2kHz,1kHz,2.6kHz,3.6kHz and 5.8kHz,irrespective of their different phonological environments. The dicovery prompted us to propose a method for a talker identification. The method focused on the local peaks in the frequency region between F0 and 0.2-1kHz for vowels Spectral envelopes of the frequency region between 3.6kHz and 8kHz were used for consonants. If the slopes of the local peaks of vowels and consonants are overlapped at several points in these frequency regions, the talkers of the samples under comparison will be regarded as the same. Using this method, we drew a figure in which Samoyedic speech samples were assorted into four groups and kanmassian ones made a single group.
北見工業大学の論文
2006-03-20

カールゲル著ケート語初等読本およびドンネル収録音声資料試論

スポンサーリンク

概要

著者

関連論文

スポンサーリンク