

Machine translation and type theory
pp. 281-311
in: P. Dybjer, Sten Lindström, Erik Palmgren, Göran Sundholm (eds), Epistemology versus ontology, Berlin, Springer, 2012Abstract
This paper gives an introduction to automatic translation via examples from the history of the field, where statistical and grammar-based methods alternate. Grammatical Frameword (GF) is introduced as an approach that uses type theory to provide high-quality translation between multiple languages. GF translation is fundamentally grammar-based but can be combined with statistical methods such as learning translation models from a corpus and ranking translation candidates by probabilities.