Construcción de un corrector ortográfico híbrido para el chabacano de Zamboanga

Abstract

Zamboanga Chavacano is a variety of Philippine Creole Spanish spoken mainly in Zamboanga City. Based mostly on Spanish with elements of Visayan, Tagalog and English origins, its mixed nature and peculiar etymology-based orthography forces speakers to deal with different writing systems in order to be able to correctly write the language. The diversity found in the (still in use) non-standard writing systems and the omnipresence of code-switching and code-mixing phenomena are a further challenge for automated spelling error detection and correction. This research aims to develop a scalable and interoperable spell checker, capable of handling the Zamboangueño orthographic problem. Our results show that it is possible to raise the de facto standard spell checker Hunspell performance to acceptable precision levels, by combining it with machine learning techniques and incorporating Tagalog and English data in its processing. Such a spell checker would also allow users to indirectly familiarize with the orthography.