Akshara-to-sound rules for Hindi

Abstract

The present paper brings to focus the main issues of regularity and inconsistency in orthography-phonology relation in Hindi arising out of an attempt at developing a program of Akshara-to-Sound rules for a pronunciation lexicon of Hindi for use in language technology for voice browser applications. The program of rules is based on interdisciplinary insights from research on writing systems, phonology and rule writing for a computer program. The paper first discusses the motivations for the Akshara-to-Sound rule set formulated to generate two levels of output—one of phoneme level and the other of phonemes and the prosodic structures of words in terms of strong and weak syllables. The irregularities in the correspondence between the Akshara and the sound are taken up next for an account. The paper ends with a discussion of the implications of the study for research on the acquisition and processing of written words in Hindi.

Keywords:

• Akshara-to-sound rules/ Grapheme-to-phoneme rules
• Pronunciation lexicon
• Devnagari script
• Syllable
• Prosodic structure

I wish to thank T. Meiraba and M. Mahesh at the early stage and to Dhirendra and Somnath Roy at the later stage of the work reported here for their assistance. Somnath's assistance has been crucial in writing out the program. I also wish to thank Sonali Nag for very useful comments and suggestions on an earlier draft of the paper and to two anonymous reviewers for their detailed comments and suggestions on the submitted draft.

The work has been carried out as a part of a project on Pronunciation Lexicons for Hindi and Bangla in collaboration with the Department of Information Technology, IIT Kharagpur, with financial grant from the TDIL programme of the Department of Information Technology, Govt of India for the period 2011–13. The program for generating the prosodic structure of Hindi words has been developed independently for Hindi.

I wish to thank T. Meiraba and M. Mahesh at the early stage and to Dhirendra and Somnath Roy at the later stage of the work reported here for their assistance. Somnath's assistance has been crucial in writing out the program. I also wish to thank Sonali Nag for very useful comments and suggestions on an earlier draft of the paper and to two anonymous reviewers for their detailed comments and suggestions on the submitted draft.

The work has been carried out as a part of a project on Pronunciation Lexicons for Hindi and Bangla in collaboration with the Department of Information Technology, IIT Kharagpur, with financial grant from the TDIL programme of the Department of Information Technology, Govt of India for the period 2011–13. The program for generating the prosodic structure of Hindi words has been developed independently for Hindi.

Notes

¹ The slash in </PS> has the function of indicating closure of the element <PS> in the document, following a general convention.

² The discussion on Devnāgarī below draws from a recent document entitled, ‘Devanagari Script Behaviour for Hindi’ prepared by the Expert Committee on Linguistic Resources and Language Technology Standards for the TDIL programme of the Department of Information Technology, Govt of India. The present author has been a member of the Committee.

³ Listed as a vowel in Sanskrit grammar, it has lost its vocalic quality in all the present-day pronunciations of Sanskrit, alternatively pronounced as /ru/ in languages of south India.

⁴ Although traditionally श and ष are in the abstract distinguished as palate-alveolar and retroflex voiceless fricatives, in isolation the distinction between them is neutralised in all varieties of standard Hindi. The grapheme, however, is retained in the orthography; the retroflex allophone is pronounced consonants clusters preceding retroflex plosives, e.g कष्ट/kəʂʈ/ ‘pain’, पृष्ठ /priʂʈ^h/ ‘page’, कृष्ण /kriʂɳ/ ‘Krishna’.

⁵ The grammatical used in the glosses have their full forms as follows: CAUS= CAUSATIVE, IMP= IPERATIVE, NH= NON-HONORIFIC, VTR= VERB TRANSITIVE, VINTR= VERB INTRANSITIVE, 3PS= THIRD PERSON SINGULAR, PST= PAST