Managing Semantic Norms for Cognitive Linguistics, Corpus Linguistics, and Lexicon Studies
Bodo Winter · The MIT Press eBooks · 2022
examples where researchers commonly do not use norms to elucidate some of the problems that may arise for norm-less semantics with respect to the reproducibility of these studies (section 2); followed by an overview of some common norm data sets (section 3); and, finally, a discussion of methodological challenges of norm-based research and how norm-based linguistics fits within contemporary efforts to facilitate reproducible research (section 4). Norm-less semantics: Two examples Corpus linguistics: Semantic prosodyIt is widely known that word meaning depends on context.Take, for example, the verb to cause, which, when seen in isolation appears to be a rather neutral term.However, when using a corpus to look at the contexts this term tends to occur in (e.g., via a concordancer), it becomes apparent that by and large, only bad things get caused (Stubbs 2001), which is exemplified by the concordance lines from the Corpus of Contemporary American English (COCA; Davies 2009) shown in table 42.1.The idea that words consistently occur in certain types of attitudinal or emotional contexts has been dubbed semantic prosody in the British tradition of corpus linguistics (Hunston 2007;Louw 1993;Stewart 2010;Whitsitt 2005), and the fact that language users can so greatly mischaracterize the "connotation" of words when they introspect on them in isolation is a major argument for using corpus methods when looking at word meaning.In this field, other headwords that have been studied with respect to semantic prosody include to set in (Sinclair 1991) and utterly (Louw 1993), both of which tend to occur in negative contexts.As another example, consider the plural form days, for which Louw (1993) claims