| This study investigates the variation of discourse markers and stance adverbials in spoken American English. Previous research shows a great deal of disagreement as to the identification of discourse markers, and lists of discourse markers often include lexical items which have also been described as stance adverbials. The aim of the present study is to identify the most common grammatical, semantic and functional properties of the lexical items well, actually, like, kind of, about, you know and see in order to (1) determine whether each item is a discourse marker or a stance adverbial; and (2) differentiate the class of discourse markers from the class of stance adverbials.; The study analyzes a large number of tokens of each target lexical item from the three-million-word Longman Corpus of Spoken American English. A combination of automatic and semi-automatic processes are used to produce detailed coding of each occurrence of a target lexical item for (1) its position in the utterance; (2) the grammatical structures which surround it; and (3) the age and sex of the speaker. Quantitative analyses of each item's preferred distributional patterns are combined with qualitative functional interpretation of extended corpus examples which illustrate those preferred patterns.; Findings indicate that the majority of the items investigated can not be classified into discrete categories defined by grammatical, semantic and functional properties. Rather, each item takes a different position along a semantic/functional continuum which ranges across three categories, from (1) highly restricted, adverb-like functions, through (2) propositional-level functions based on encoded lexical meaning, to (3) discourse/text/social-level functions of propositionally empty lexical items. Moreover, several of the target lexical items show evidence of grammaticalization in progress from one place along the continuum to another.; In conclusion, the implications of these corpus-based findings for the study of spoken discourse, language teaching and grammatical analysis are discussed, and suggestions of areas for future research are made. |
