Determining Recurrent Sound Correspondences by Inducing Translation Models

Determining Recurrent Sound Correspondences by Inducing Translation Models Grzegorz Kondrak Department of Computer Science University of Toronto Toronto, Ontario, Canada M5S 3G4 Abstract I present a novel approach to the determination of recurrent sound correspondences in bilingual wordlists. The idea is to relate correspondences be- tween sounds in wordlists to translational equiva- lences between words in bitexts (bilingual corpora). My method induces models of sound correspon- dence that are similar to models developed for sta- tistical machine translation. The experiments show that the method is able to determine recurrent sound correspondences in bilingual wordlists in which less than 30% of the pairs are cognates. By employ- ing the discovered correspondences, the method can identify cognates with higher accuracy than the pre- viously reported algorithms. 1 Introduction Genetically related languages often exhibit recur- rent sound correspondences (henceforth referred to simply as correspondences) in words with similar meaning. For example, t:d, T:t, n:n, and other known correspondences between English and Latin are demonstrated by the word pairs in Table 1. Word pairs that contain such correspondences are called cognates, because they originate from the same protoform in the ancestor language. Correspon- dences in cognates are preserved over time thanks to the regularity of sound changes, which normally ap- ply to sounds in a given phonological context across all words in the language. The determination of correspondences is the prin- cipal step of the comparative method of language reconstruction. Not only does it provide evidence for the relatedness of languages, but it also makes it possible to distinguish cognates from loan words and chance resemblances. However, because man- ual determination of correspondences is an ex- tremely time-consuming process, it has yet to be ac- complished for many proposed language families. A system able to perform this task automatically English Latin t E n d e k e ‘ten’ t ¯u d u o ‘two’ ¯ı t e d ‘eat’ t ¯u T d e n t ‘tooth’ n E s t n i d ‘nest’ n ¯ı g e n ‘knee’ n E f j ¯u n e p o t ‘nephew’ f u t p e d ‘foot’ f ¯o m s p u m ‘foam’ w u l f l u p ‘wolf’ Table 1: Examples of English–Latin cognates exhibiting correspondences. The corresponding phonemes shown in boldface originate from a sin- gle proto-phoneme. from unprocessed bilingual wordlists could be of great assistance to historical linguists. The Recon- struction Engine (Lowe and Mazaudon, 1994), a set of programs designed to be an aid in language re- construction, requires a set of correspondences to be provided beforehand. The determination of correspondences is closely related to another task that has been much stud- ied in computational linguistics, the identification of cognates. Cognates have been employed for sentence and word alignment in bitexts (Simard et al., 1992; Melamed, 1999), improving statisti- cal machine translation models (Al-Onaizan et al., 1999), and inducing translation lexicons (Koehn and Knight, 2001). Some of the proposed cognate identification algorithms implicitly …

Content truncated to 500 words