Konkani Integer Phonetic Transcription System

Abstract

This paper describes an ongoing work on the Phonetic Dictionary for Konkani language. In this work, we have build a resource that would phonetically transcribe Konkani Integers and generate their written form in the Devanagari script. The algorithm developed in this work takes an integer as an input and generates its written form in the Devanagari script, along with its phonetic transcription in the International Phonetic Alphabet (IPA). Our algorithm is a rule-based system in which phonetic transcriptions of numerals are created using rules from the available literature and for some cases we have proposed new forms for the numerals. The algorithm has been made robust enough to automatically give a written form of any Konkani numeral in the Devanagari script, along with its equivalent IPA transcription. This work is the first step towards providing an open-source phonetic dictionary for Konkani language. We have tried to keep the phonetic transcriptions as much as closer to their natural pronunciations. This is done for the purpose of capturing the general tendency of the language. So, for example, while the number '8' आठ [aúh] is written with an aspirated retroflex consonant ठ [úh], the final consonant [úh] is heard without aspiration in the actual speech. This loss of aspiration at word final position generally happens across all the consonants of the language, in the Konkani varieties spoken in Goa.