High Vowels Devoicing and Elision in Japanese: a Diachronic Approach
Total Page:16
File Type:pdf, Size:1020Kb
High vowels devoicing and elision in Japanese: a diachronic approach Francesca Pinto Università degli studi di Roma “La Sapienza” - Fondazione Ugo Bordoni [email protected] ABSTRACT partly due to the ease in pronouncing certain consonantal clusters instead of others – a cluster The present study will focus on occurrence of fully (resulting from the elision of an interconsonantal voiced, devoiced and elided high vowels in vowel) composed of a palatal fricative plus an Japanese. High vowels elision in Japanese is alveolar occlusive is easier to pronounce than one supported by clear acoustic evidences, even though composed of two occlusives. Other than that, it is not globally accepted by scholars in the field. words frequency plays an important role in Elided vowels, hence, are considered here to have determining vowel elision regularity: words like their own status. desu ‘to be’ or shita ‘(verbal morpheme for past Moreover, this study is conceived in order to tense)’ have a high degree of frequency in Japanese state differences and analogies among four dialects and their realization with high vowel elision has and between five age groups. In fact, the main gradually become a stable habit. interest of this research is to prove that high vowel Notwithstanding clear and neat phonetic elision may be considered as an innovative evidences that prove vowel elision, traditionally phenomenon. Consequently, it is expected that a only vowel devoicing is accepted by scholars in diatopically and diachronically based study may this field, especially by the native ones. It is suggest the drift of Japanese language about the possible to assume that the reason for that is said phenomenon. Therefore, it may be possible to mainly due to the influence exerted by the hypothesize changes on the strict syllable structure Japanese writing system. Apart from Chinese deriving from Japanese writing system, mainly ideograms and Latin characters, Japanese language composed of syllabic morae, which conditions is conveyed by kana, a syllabic writing system mother tongue perception of Japanese phonology where each symbol stands for a mora composed of and phonotactics. a vowel, or a consonant plus a vowel, with the exception of the symbol <n>, the only consonant Keywords: High vowels devoicing, elision, that can occur alone, but not in word initial Japanese dialects, diachrony. position. In other words, Japanese kana has influenced the perception of Japanese phonology 1. INTRODUCTION and phonotactics so much that even scholars are lead to think that only open syllables are possible – Japanese high vowels devoicing is a well known with the said exception – and hardly acknowledge and studied phenomenon, examined by many vowel elision and consonantal clusters that derive native and non-native scholars [e.g. 8, 11, 15] since from this process. the early stages of literature on Japanese In the present paper it is claimed that vowel linguistics. elision has its own specific phonological status. High vowel devoicing is a phenomenon for Furthermore it can be considered as an innovative which vowels are pronounced without the vibration feature in Japanese language. The main proof to of the vocal folds and where oral organs support this hypothesis relies upon dialect configuration is more similar to that of whispering. differences. It is well known [5, 11] that Kansai They are hence, voiceless. Vowel devoicing is dialects, in the South-centre of Japan, mainly more likely to occur in specific phonotactic represented by Ōsaka and Kyōto, are highly contexts, namely between two voiceless conservative and tend to keep archaic features, as consonants or between a voiceless consonant and a compared to Standard Japanese. In this area a fully pause. Factors like speech rate may also have a voiced realization of /i, ɯ/ is more frequent, strong influence. instead of vowel devoicing or elision and it is Under the same circumstances high vowel partly because here rounded variant of the high elision may occur as well. Depending on a higher back vowel is used, whereas in Standard Japanese speech rate, on the quality of the consonantal its unrounded counterpart is used, making the environment and on the word recurrence, this latter phone more audible and neater. Moreover, younger phenomenon may have an even higher degree of speakers are more inclined to vowel reduction than occurrence as compared to vowel devoicing. In adults or elderly. fact, it is quite evident that some specific contexts For this reason, an analysis of this phenomenon require vowel elision rather than fully voiced or occurrence through a diachronic study may lead to devoiced vowels. The reason for this distribution is a new perspective on this topic and make a conjecture about Japanese language future devoicable position, for instance: (1) the case development. were /ɯ/ stands between /s/ and a pause; (2) when In this paper, then, emphasis on diachrony is high vowels are between two voiceless occlusive; pursued through both analyses on different age (3) between two voiceless fricatives; (4) between a groups and regions. It is expected that a voiceless occlusive and a fricative; (5) between a comparison among four diatopic and five voiceless fricative and an occlusive. diachronic variants may allow hypothesis on two About 200 words for each of the above main topics. It would be possible to propose some mentioned contexts – 966 in total – have been accurate assumption about a description of dialects selected, extracted and acoustically analyzed. distribution from a diachronic point of view. Other It is well known that the definition of the than this, it is hoped that analysis from the present concept of ‘word’ is quite difficult and it is even paper will help in outlining the possible harder to define what a word is in Japanese, due to phonological drift that Japanese is following, at the large varieties of more or less bound affixes, least with regards to high vowels treatment and, particles and other kind of nearly grammaticalized eventually, to phonotactic implications. constructions. Therefore, for the aim of the present study constructions like nouns plus a particle, verb 2. METHODOLOGY plus inflectional suffixes, adverbs or conjunctions in isolations are considered as single units. 2.1. Corpus Construction where prosodic and syntactic boundaries might not coincide, like nouns followed The corpus used for the aim of the present study is by a copula, have been excluded by the selection, The Speech Corpus and Database of Japanese as well as nouns followed by three or more Dialects (henceforth SCDJD), supervised by Miy- particles [3]. Moreover, only units with no pitch oko Sugito during the 90’s [14]. This corpus is accent on the studied segments have been chosen. composed of interviews on Japanese native speak- ers divided into five age groups, coming from 14 2.4. Data analysis Japanese prefectures. Audio files have been sampled at 48kHz with 16bit accuracy and are The focus of the analysis was to discriminate the based on interviews dealing with isolated words quality of the pronounced vowel and to state and short sentences, stories, weather forecast, whether it is fully voiced, devoiced or elided. A greetings, questions and answers, short conversa- voiceless vowel can be easily distinguished from a tions, Japanese syllabaries reading and numerals. fully voiced vowel for a series of acoustic Speech samples from four regions, namely one peculiarities: it does not display any trace of the in the North, Hirosaki, one in the South, Kagoshi- voice bar at the bottom of the spectrogram; shows ma, and two in the central part of Honshū island, a sensitive increase of both F1 and F2; formants Nagoya and Ōsaka – the latter being representative look less neat and sharp; pitch is absent; signal is of the Kansai dialect – have been used here. aperiodic; overall, the phone length is shorter. On the other hand, when a vowel is elided, there is the 2.2. Speakers complete absence of the phone with all its characteristics. In the SCDJD five age groups are present, namely: Based on the above mentioned parameters, high elderly, aged over 60 years old; adults, aged 40 to vowels in devoicable position of each word have 59; youth, aged 20 to 39; middle school students, been analyzed and, then, divided into three aged 13 and 14; elementary school students, aged categories: fully voiced, devoiced and elided, in 10 to 12 years old. Even though a scrupulous dia- accordance with their acoustic nature. chronic analysis would need a wider time frame, it Resulting data have been statistically analyzed is possible to state that five generations of speakers and subcategorized according to the four chosen may anyhow contribute significantly to a dia- Japanese dialects and to the five age groups chrony based study. represented in the corpus. For the aim of the present research speech Acoustic analyses have been conducted using samples of twenty speakers per each region have Praat [4]. been used. In particular, speech samples are related to at least two speakers per age group and sex. 3. RESULTS A total amount of 80 speakers have been ana- lyzed here. Data results are shown in the following table. Main parameters taken into account to compose this 2.3. Collected data table are diatopic variance, diachronic variance and high vowel articulatory manner. Other parameters Starting from the corpus transcription, selection like vowel quality, consonant quality surrounding have been made on words representative of five vowels or phonotactic context are not taken into phonotactic contexts, involving high vowels in account here. In this phase, then, results obtained from the youth have stronger bounds and deal with more above mentioned acoustic analysis are combined frequently. and normalized. The displayed percentages are Looking at the table from a diatopic point of cumulative and subcategorized into groups view it will be clear that Nagoya and Kagoshima according to age and dialectal area.