Abstract
Data from corpora suggest that certain patterns in language are formed relatively frequently by the combination of various lexical items. These multiword units as prefabricated units serving different functions in discourse construction. In agglutinative languages, it has been argued that frequently occurring morpheme sequences may come to form n-grams akin to multiword units. In her seminal work, Tura (1986) points to discourse functions of morpheme sequences that are formed by -DIr (the generalizing modality particle) and a number of preceding tense, aspect and modality affixes. This paper presents the quantificational distribution of one such morpheme pattern from the corpus data. Particularly, the paper extracts collocations of the -mIş-person-DIr multimorpheme unit and analyzes its role as a stancemarker in discourse.
