Automatic Learning of Grammatical Encoding
MetadataShow full item record
Avenue (Probst et al., 2002, Monson et al., 2004, Lavie et al., 2003, Font-Llitjoset al.,2005)1 is a machine translation system that automatically learns translation rules between two languages. In the Avenue scenario, one of the languages is a resource rich language like English or Spanish, for which there are many human and electronic resources (corpora,morphological analyzers, lexica, etc.).The other is a resource poor language with few human and electronic resources.For example, there might be no linguist available to write translation rules and there might not be large enough corpora for automatic machine learning of translation rules. This is true for the vast majority of languages. Within the current state of the art in commercial machine translation, it is not possible to build machine translation (MT) systems for resource poor languages.However,we have met with many indigenous communities (Mapuche, Quechua, and others), who want their languages to be used in jobs, education, government, and health care. Machine translation can be a tool for maintaining functionality in their languages, because it can help them access the content of the Internet and disseminate local culture and information without having to adopt a major national language like Spanish or English. The vision of the Avenue project is equal access to information for speakers of all languages.