This study intends to examine the allophonic characteristics of the Oriya retroflex nasal phoneme when it occurs in contrastive and complementary distribution in speech. The main thrust is to extract and analyze the spectrographic data of allophonic variants of retroflex nasal and their manifestation, which is significant by considering the acoustic values in different contexts. The variation values are measured in acoustic parameters such as duration of nasal bar, frequency of first, second, third and fourth formant, intensity and fundamental frequency, which are perceptually unique. This is necessary for the development of speech database. The variations of the speech units are classified into (1) co-articulatory variations, which to some extent may be negligible, and (2) allophonic variations, which are contextually significant. A Phoneme, which is an abstract unit comprising a bundle of features, is realized by the actual pronunciation of the phones/ allophones by the native speakers. The Speech synthesis system has to map the allophonic variations and their contextual rules in respect of each phoneme so as to transform it into phonetic conversion module.
The Oriya retroflex nasal phoneme and its different manifestations described by the Descriptivists in their non-instrumental study, therefore, need further classification, categorization and modification on the basis of acoustic features.
The previous studies show the allophonic variations of Oriya phonemes on the basis of their complementary distribution in word as well as in syllable initial, medial and final position. Complementary features of phonemes are positional specific, which have been realized by the changing of its degree of place and manner features. In Oriya, the substantial change of manner feature is noticed in the case of retroflex nasal when it precedes and follows the back category of vowels. In that context the retroflex nasal becomes retroflex nasal flap. A group of Oriya retroflex allophonic...