Genre-specific properties of adverbial clauses in English academic writing and newspaper texts Elma Kerz & Daniel Wiechmann A constructionist perspective on patterns Schematicity Complex sentence constructions [SUBJ [V DO IO]] [SUBJ [V POSS way ] OBL]] i i jog <someone’s> memory kick the bucket table Complexity Complex sentences (minimally) bi-clausal construction •  comprises main clause & … … complement clause relative clause adverbial clause … … temporal causal concessive … Complex sentences (minimally) bi-clausal construction •  comprises main clause & … … complement clause relative clause adverbial clause … … temporal causal concessive … Genre-specific patterns •  Killgarriff (2003: 54): •  “General English” is a theoretically difficult notion, and a language can be seen as a modest core of lexis and constructions, plus a wide array of different sublanguages, as used in each of a myriad of human activities.” •  “any patterns generalized for all of English are not likely to be valid for any actual text or register – rather, generalized patterns would merely level the important patterns of use found across registers” (Biber et al. 1998: 83) •  Case in point: Academic writing* (A growing number of studies looking at recurring multiword expressions (e.g. Biber, Conrad & Cortes 2004, Biber 2006, Hyland 2008, Simpson-Vlach & Ellis 2010) Two views of the constructicon Constructicon A Constructicon B Genre 1   Genre k   Genre 2   Genre …   Genre … Two views of the constructicon Constructicon A Constructicon B Genre 1   Genre k   Genre 2   Genre …   Each construction C Genre (e.g. ditransitive) …   has a global/uniform entrenchment value in speaker’s mind Two views of the constructicon Constructicon A Constructicon B Genre 1   Genre k   Genre 2   Genre …   Each construction C Genre (e.g. ditransitive) …   A construction C has a global/uniform (e.g. ditransitive) entrenchment value in has a local/genre- speaker’s mind specific entrenchment value in speaker’s mind Aim of the study •  Assumptions: Grammar can be conceived of as a set of sub- repositories encompassing genre-specific constructions •  Goal : Detect conventionalized & cognitively entrenched (~strongly represented) types of concessive and reason constructions in the domain of academic writing Corpus & data used in the study   •  Academic writing: –  ~700,000 words extracted from 100 peer-reviewed research articles covering a wide range of academic disciplines •  Newspaper: –  same-sized sample was extracted from the newspaper sub-component of the BNC •  Query for subordinators –  reason clauses {as, because, since} –  concessive clauses {although, even though, though, while, even if, whereas} •  A total of 1911 data points coded in terms of 15 variables –  Academic writing (N = 575; N = 471) Concessive Reason –  News (N = 517; N = 348) Concessive Reason