The humanities scholar's perspective on rule based machine translation

Rule-based machine translation is an interesting application for natural language processing as well as digital humanities, for the reason that it spans over so many of the topics and concepts of NLP and DH. It can therefore be included in so many DH and NLP work-flows, producing necessary resources, such as dictionaries, digital grammars, as a side-product of research work in digital texts and corpora.

During this course we will create a simple rule-based machine translation system (based on Apertium) that is capable of translating one short text from one language to another. We will learn to write necessary dictionaries in an XML-based format, use version control software, and participate in open source development community.

Tentative schedules

Week one

• Intro to Machine Translation
• Installing the platform and tools
• Working with the tools
• XML basics
• Digital lexicography and morphology
• Parsing
• Word-based translations

Week two

• Phrase parsing / chunking
• Re-ordering and grammatical changes
• Evaluation and Quality Assurance
• Comparative grammars
• Conneting to large coverage dictionaries
• other systems (if there's time)