Semantics of Token Positions in Sentences

This publication doesn't include Faculty of Arts. It includes Faculty of Informatics. Official publication website can be found on



Type Article in Proceedings
Conference Slavonic Natural Language Processing in the 21st Century
MU Faculty or unit

Faculty of Informatics

Keywords token position; histogram; clustering; semantics; corpus linguistics; distributional semantics
Description This paper analyses absolute positions of Czech and English words in sentences. It shows that even this simple statistics provides some insight into semantics of the words. When the absolute position distribu- tions are represented as matrices and clustered the resulting hierarchy reveals semantic relations between the words.