From complex prepositions to complex subordinators: challenging generalizations

ABSTRACTThis paper presents a first approximation of the relationships within and between complex prepositions and complex subordinators from a cognitive vantage point. Because of the similarities between the two categories at the syntactic and functional levels of analysis, we first hypothesize the existence of a common productive schema that encompasses members of both classes within the framework of Construction Grammar. We then put forward a methodology developed in Diachronic Construction Grammar to analyze the within- and between-category constructional links in complex prepositions and complex subordinators in American English. We use multivariate quantitative methods to highlight the similarities and dissimilarities between the categories and make them visually identifiable. The application of this methodology to historical corpus data reveals that contrary to our expectations, complex prepositions form a relatively stable schema over time, without converging towards a common schema with the loosely connected complex subordinators under scrutiny.KEYWORDS: Complex prepositionscomplex subordinatorscorpus-based diachronic construction grammarAmerican Englishmultivariate statistical analysis Disclosure statementNo potential conflict of interest was reported by the author(s).Notes1 We use the terms ‘productive/-ity’ to refer to the tendency for a pattern to be used frequently to produce new types in a grammatical category.2 In this context, we acknowledge that the preposition does not impose the same construal of the situation: the noun phrase the rain projects a static frame of the object ‘rain’, whereas the clause invites a more dynamic construal of a continuous event.3 As was pointed out by an anonymous reviewer, the noun fact can be used with a relative clause in a variety of contexts due to its very broad and opaque meaning, especially in oral usage.4 In the remainder of this paper, examples culled from COHA will be referred to as follows.(example number) linguistic example (genre, year)5 We refer the reader to the study by Brems and Davidse (Citation2010) for simple statistical frequencies on the co-occurrence of complex subordinators with the discussed syntactic features.6 Here we use small capitals to refer to construction names.7 for fear of/that/Ø is also known to express conditional relations.8 for the sake of can also be used to express a benefactive relation to nominal complements.9 instead of is not discussed in this paper because it varies too much in its complementation profile.10 Brems and Davidse (Citation2010, 106) highlight that the non-finite nature of the complement clause is a facilitating factor for complex subordinator readings.11 We also acknowledge the fact that there is, to date, not any corpus big enough to allow a greater time depth for the analysis of low-frequency linguistic items such as CPs and CSs.12 For our purpose, we used R version 3.4.4 with the function MCA from the package FactoMineR and the function apcluster from the package of the same name.13 Note that the percentage of variation is displayed along the axes in our analyses for the sake of completeness only.14 Dimension 5 was excluded from the analysis as it focused on a single feature in a binary manner, specifically highlighting the unique use of lest in fear-related constructions. This represents a distinct case of complementation. Additionally, Dimension 5 reveals a genre continuum between NEWS (negative coordinates) and NF (positive coordinates). However, since Dimension 4 already demonstrated a significant genre continuum and the inclusion of lest as a category reduced interpretability by compacting data points, Dimension 5 was omitted.15 The grey-shading of the cells serves to highlight that distinct clusters are identified when the features are considered in isolation versus when they are combined. This underscores the significance of interactions among the features and the necessity of examining their combined impact for discerning meaningful distinctions between the clusters.16 Admittedly, the procedural meaning of prepositions makes it hard to find groupings of prepositions which are semantically similar and to which we can interchangeably resort to in the same context.Additional informationFundingThis work was supported by the Fonds De La Recherche Scientifique - FNRS [32704391].