|
PUBLICATIONS
|
 |
Bibliographies
|
|
0
|
|
|
Book chapters
|
|
0
|
|
|
Book reviews
|
|
0
|
|
|
Books
|
|
0
|
|
|
Books, edited
|
|
0
|
|
|
Conf. papers
|
|
0
|
|
|
Conf. presentations
|
|
0
|
|
|
Conferences
|
|
0
|
|
|
Discussion groups
|
|
0
|
|
|
Grants
|
|
0
|
|
|
Journal articles
|
|
0
|
|
|
Periodicals/series
|
|
0
|
|
|
Proceedings
|
|
0
|
|
|
Proceedings, papers
|
|
0
|
|
|
Reports
|
|
0
|
|
|
Special issues
|
|
0
|
|
|
Theses
|
|
0
|
|
|
Treaties
|
|
0
|
|
|
Working papers
|
|
0
|
|
|
|
STATISTICS
|
|
Citation Rank
|
|
|
|
|
Total Citations
|
|
0 |
|
|
Publications
|
|
0 |
|
|
Rank
|
|
325051 |
|
|
Viewers
|
|
3 |
|
|
Views
|
|
748 |
|
|
|
|
|
Dr. Greg Kondrak
(b. ----,
d. ----)
|
( Prev | Next )
|
|
|
|
|
POSITION(S) / JOB TITLE(S):
|
|
Assistant Professor |
|
|
The problem of language reconstruction can be defined as follows: given two related human languages, reconstruct their common ancestor. It has all the attributes of a driving problem -- a problem that is easy to state but very hard to solve, and which is a source of new methods and insights that are not restricted in application to a single field of study. Historical linguists, who have been studying the problem for decades, developed the comparative method of language reconstruction. A number of proto-languages have been proposed on the basis of the method. However, language reconstruction is an extremely time-consuming process that has yet to be accomplished for many language families. The amount of data and the complexity of the problem suggest that computers could be of great assistance in achieving this goal. Unfortunately, the comparative method appeals to intuitive criteria and is not sufficiently formalized for a straightforward computer implementation. The problem has attracted attention of prominent researchers in the computational linguistics community but no truly comprehensive and functional solution has been proposed so far.
In my doctoral dissertation I proposed three algorithms for implementing individual steps of the comparative method: identification of cognates, alignment of cognates, and the determination of recurrent correspondences. The algorithms, which incorporate some of the most recent techniques from bioinformatics and natural language processing, have been fully implemented and are publicly available. They represent significant steps towards the goal of automatic proto-language reconstruction from raw dictionary-type data.
|
|
|
Assistant Professor |
|
Computing Science |
|
University of Alberta |
|
|
Only Visible to Members of getCITED
|
|
|
Doctorate
|
|
|
University of Texas at Arlington
|
|
|
Male / English
|
|
|
Unknown |
|
|
1107-2050
(Last changed on
2003/04/27 19:35:44)
|
|
|
|
|
|
|