Foundations of Text Alignment: Statistical Machine Translation Models From Bitexts to Bigrammars

This book provides a systematic, foundational introduction to automatic alignment of parallel texts, a family of essential corpus analysis techniques for computing and learning the mappings between corresponding parts of the texts. Bitext alignment lies at the heart of all data-driven machine learning approaches to automatic translation, and the rapid research progress on alignment during the past two decades underlies the success of statistical machine translation approaches. Alignment is used across a wide range of resource acquisition applications including word sense disambiguation, terminology extraction, and grammar induction, as well as in translation memories and biconcordances for translators' assistants, bilingual lexicographers, and computer assisted language learners.