Developments Towards an Electronic Amharic Corpus.pdf

TALN 2005, Dourdan, 6-10 juin 2005 TALN 2005 Developments Towards an Electronic Amharic Corpus Daniel Yacob Ge’ez Frontier Foundation 7802 Solomon Seal Dr, Springfield, VA 22152, USA yacob@ Abstract The state of Amharic natural language processing was aptly assessed at TALN 2003 by Atelach, Asker and Mesfin. A public Amharic corpus and a comprehensive lexicon were two of the most needed items in absence for Amharic language researchers. Since the 2003 assessment some progress has been made in these two areas and researchers have begun informal collaboration to address the common goal of developing these public resources. In this same period Ethiopia’s legal system has changed to cloud the issue over what the legal status of an Amharic corpus would be. While a promising start is underway, corpus developers and researchers alike will have to familiarize themselves with the new legislature in Ethiopia and reexamine the status of their holding to avoid potential unintended violations. 1 Introduction Amharic is the most studied and best understood language of Ethiopia, it also serves as the country’s lingua franca. Researchers today, both inside and outside of Ethiopia, are increasingly interested in computational investigations of the Amharic language. The lack of a freely available electronic corpora, lexicon, and transcription standard, coupled with the complexities of Amharic orthography are a significant barrier to would be researchers. Amharic, along with its ten sibling Ethio-Semitic languages found in Ethiopia and neighboring Eritrea, is written in the Ethiopic syllabary. Amharic has been a written language for roughly 600 years and has as rich legacy of both typeset and calligraphic literature. Significant amounts of electronic corpora in