Skažem tak : Discourse markers originating from verbs of speech in Russian

Based on data from the Russian National Corpus, first and foremost, from its multimedia corpus, we address two members of the family of discourse markers developed from the finite form of the verb of speech skažem (the first person plural present tense form of the verb ‘say’): isolated skažem and its combinations with an adverb or a pronominal adverb such as skažem prjamo ‘let’s put it bluntly’, skažem tak ‘let’s put it this way’, etc. We show that these markers demonstrate a typical pragmaticalization trajectory: unlike the original speech verb, they do not refer to a speech event, their main function is to express a fuzzy nomination, or hedging, i.e., these markers are used when the speaker disclaims responsibility for the possible inaccuracy of the propositional content or of the nomination used in relation to the fixed propositional content. We demonstrate the main shifts in the distribution of these markers, which occur due to the fact that they fall out of the rigid clausal structure and acquire the status of an operator which scopes over discourse segments of various (potentially any) length. We study the main prosodic properties of these markers instrumentally and perceptually and, in addition, discuss the use of markers of this family together with other means of signaling that the speaker needs additional efforts in choosing an adequate expression. These can be other lexical markers of fuzzy nomination, lexical and prosodic indicators of an open list, self-corrections, signals of hesitation (filled pauses, non-phonological lengthening of sounds), etc. The results obtained convincingly show: (1) that even within the same family of markers, the shift from a unit with propositional meaning to a discourse marker can follow different trajectories and lead to larger or smaller shifts in meaning and form; and (2) that the study of discourse markers that uses data on prosody and such specific phenomena of oral discourse as speech disfl uencies makes it possible to get more comprehensive lexicographic portraits of single units and their families.