Using support vector machines and state-of-the-art algorithms for phonetic alignment to identify cognates in multi-lingual wordlists.