Abstract Geeraerts & Speelman
On the importance of incorporating lectal factors in usage-based linguistics: A corpus analysis of make and let causatives in Dutch
There is a growing tendency in Cognitive Linguistics to define its type of linguistics as a form of usage-based linguistics (Barlow and Kemmer 2000, Tummers et al. 2005). The methodological consequence of taking a usage-based approach is to base the analyses on actual language data, as in corpus linguistics, and this in turn implies that sophisticated techniques of analysis are called for, viz. multivariate analysis. The present paper argues that such a multifactorial approach to grammar needs to incorporate lectal variation (i.e. variation associated with the difference between dialects, sociolects, registers, idiolects etc.) as a crucial factor.
The paper analyses the distribution of the Dutch causative verbs doen and laten in the Corpus of Spoken Dutch (CGN). The analysis takes the form of a logistic regression based on language internal features (the syntactic construction pattern, the animacy of the subject, and diverse measures of collocational strength) and on lectal features (the difference between Belgian Dutch and Netherlandic Dutch, and the situational context of utterance).
Two hypotheses receive specific attention. First, the existing literature on linguistic variation within Dutch points to a different distribution of doen and laten between Belgian Dutch and Netherlandic Dutch. This raises the question whether the difference between both national varieties of Dutch is one of degree or substance. Second, existing work on Dutch causatives by Verhagen & Kemmer (1997) and Stukker (2005) suggest that the difference between doen and laten may be described in terms of a difference between direct and indirect causation. This raises the question whether the factors that turn out to be significant according to the regression analysis, conform to the predictions of the direct/indirect causation model.
The study shows that the direct/indirect causation model does play a role in the distribution of the two causative verbs, but that it is much less important than studies focusing on a limited set of examples suggest. By contrast, lectal factors are generally more important. On a methodological level, it establishes the importance of advanced statistical analysis for corpus linguistics, more specifically, of multivariate techniques of analysis, and it draws the attention to the necessity of adding lectal variation to the inventory of relevant factors.
Barlow, Michael and Kemmer, Suzanne (eds.) (2000). Usage-based models of Language. Stanford: CSLI.
Tummers, José, Kris Heylen & Dirk Geeraerts. 2005. “Usage-based approaches in Cognitive Linguistics. A technical state of the art”. Corpus Linguistics and Linguistic Theory 1(2).
Verhagen, Arie & Susanne Kemmer (1997). "Interaction and causation. Causative constructions in modern standard Dutch". Journal of Pragmatics 27: 61-82.
Stukker, Ninke (2005). Causality marking across levels of language structure. Utrecht: LOT.