Schwa Deletion: Investigating Improved Approach for Text-to-IPA System for Shiri Guru Granth Sahib

Abstract: Punjabi (Omniglot) is an interesting language for more than one reasons. This is the only living Indo-Europen language which is a fully tonal language. Punjabi language is an abugida writing system, with each consonant having an inherent vowel, SCHWA sound. This sound is modifiable using vowel symbols attached to consonant bearing the vowel. Shri Guru Granth Sahib is a voluminous text of 1430 pages with 511,874 words, 1,720,345 characters, and 28,534 lines and contains hymns of 36 composers written in twenty-two languages in Gurmukhi script (Lal). In addition to text being in form of hymns and coming from so many composers belonging to different languages, what makes the language of Shri Guru Granth Sahib even more different from contemporary Punjabi. The task of developing an accurate Letter-to-Sound system is made difficult due to two further reasons: 1. Punjabi being the only tonal language 2. Historical and Cultural circumstance/period of writings in terms of historical and religious nature of text and use of words from multiple languages and non-native phonemes. The handling of schwa deletion is of great concern for development of accurate/ near perfect system, the presented work intend to report the state-of-the-art in terms of schwa deletion for Indian languages, in general and for Gurmukhi Punjabi, in particular.

Keywords: Gurmukhi Punjabi, IPA, Schwa deletion, Sikhism