This research aimed to describe endocentric and exocentric phrases as well as their references on news discourse in Kompas edition of February 2018 through syntactical and semantical approaches. This was adescriptive qualitative research. The research data were all the endocentric and exocentric phrases and theirreferences, whereas the data sources were 23 hottest news discourse of Kompas about corruption and bribery in February 2018. The data were collected through reading method of uninvolved conversation observation technique. This technique was conducted because the writer did not directly involve in the conversation, yet the writer played a role in reading the news in order to find out the endocentric and exocentric phrases as well as their references. Furthermore, the writer did note-taking and classified and analyzed the phrases . The data were
analyzed through descriptive qualitative of distributional and identity methods. The used basic techniques of distributional method were substitution, extension, and deletion; whereas the basic technique of identity method was determinant sorting technique of referential sorting. The results indicated that endocentric phrases were 286 data and the exocentric ones were 158 data. The number of references showed 141 data. Only some of the data were analyzed as they had represented classification.