Detection and Correction of Phonetic Errors with a New Orthographic Dictionary
Sivaji Bandyopadhyay · Institutional Repositories DataBase (IRDB) · 2000
The Phonetic and the Homophone Error problem in a language have been characterized as a symbol substitution problem. Phonetically equivalent symbols or symbol combinations in the language are grouped together. Each group or a number of related groups give(s) rise to a dictionary or a number of dictionaries. A new design methodology for Orthographic Dictionaries in alphabetic languages has been described. The dictionaries include the root words. The meanings are stored only in case of Homophone words. Words are sorted on the basis of a Phonetic Ordering Scheme. The dictionaries are being used to detect and correct the Phonetic Error and the Homophone Error in isolated words of Bengali. The methodologies described in this paper can be used in developing spell-checkers not only for other Indian languages which have evolved from the ancient Brahmi script and have a common phonetic structure but also for other alphabetic languages with suitable modifications. The practical NLP systems must deal with erroneous or ill-formed inputs. The nature of errors basically depends on the source of input as well as on the user's exceeding the systems limited grammatical, conceptual or functional coverage. Automatic word correction research may be viewed [7] as focusing on