A diachronic corpus-based multivariate analysis of “I think that” vs. “I think zero”

Research output: Chapter in Book/Report/Conference proceedingChapter

  • Christopher Shank
  • Koen Plevoets
    University College Ghent
  • Hubert Cuyckens
    University of Leuven
This corpus-driven study seeks to explain the choice between the zero complement and the that complement constructions, when occurring with the mental state predicate think. Previous studies have identified a range of factors that are argued to explain the alternation patterns. Such studies have also proposed that there is a diachronic drift towards zero complementation. Based on a sample of 9,720 think tokens, from both spoken and written corpora, from between 1560–2012, we test the hypothesis of diachronic change and the effect of eleven proposed factors on the constructional alternation. Using logistic regression, we demonstrate that, contrary to previous studies, there is in fact a diachronic decrease in zero complementation. Moreover, the study also demonstrates the importance of understanding the interaction of the various factors that explain the near-synonymous relation, including, especially, between the spoken and written modes.


  • complementation, logistic regression, mental state verb, near-synonymy, that/zero alternation
Original languageEnglish
Title of host publicationCorpus Methods for Semantics. Quantitative studies in polysemy and synonymy.
EditorsDylan Glynn, Justyna A. Robinson
Place of PublicationUK
PublisherJohn Benjamins Publishing Company
Number of pages24
ISBN (Electronic)978-9027223975
ISBN (Print)9027223971
Publication statusPublished - 6 Nov 2014
View graph of relations