Categorizing syntactic constructions in a corpus*
Nicholas D. Smith, Elena Seoane · Cambridge University Press eBooks · 2013
Introduction To arrive at a thorough description of the usage of a grammatical construction in a corpus involves a number of stages. Minimally it will require: (a) retrieval of a set of valid instances of the construction from the corpus; (b) categorization of those instances according to linguistic and/or extra-linguistic features; and (c) quantitative and/or qualitative analysis of the instances and their associated categories. Steps (a), (b) and (c) need not be linear – one may, for instance, wish to revisit the retrieval after looking at the quantitative results – but each step is nevertheless required. Our objective here is to demonstrate and discuss step (b) – categorization of a target structure. We take as a case study the English passive construction, more specifically the long passive, i.e. passives with an overt agent by -phrase, as in John was arrested by the police (Biber et al . 1999: 154). The kinds of coding we describe generally have to be inserted by hand, although like Sebastian Hoffmann (see Chapter 10, this volume) we recommend using a part-of-speech annotated version of the corpus to facilitate retrieval of the target data.