Testing the OED analysis of two phonesthemes with a corpus study of collocational behaviour of swand flwords in the OEC Abstract Phonesthemes (Firth 1930) are sublexical constructions that have an effect on the lexicogram matical continuum: they are recurring formmeaning associations that occur more often than by chance but not systematically (Abramova/Fernandez/Sangati 2013). Phonesthemes have been shown (Bergen 2004) to affect psycholinguistic language processing; they organise the mental lexicon. Phonesthemes appear over time to emerge as driven by language use as indexical rather than purely iconic constructions in the lexicon (Smith 2016; Bergen 2004; Flaksman 2020). Phonesthemes are acknowledged in construction morphology (Audring/Booij/Jackendoff 2017) as motivational schemas. Some phonesthemes also tend to have lexicographic acknowledgment, as shown by etymologist Liberman (2010), although this relevance and cohesion appears to be highly variable as we will show in this paper.

This paper seeks to compare two phonesthemes in a combined lexicographic and corpus study with a view to testing the results obtained. Firstly, following Smith (2016) which identified 11 semantic categories of fl words in the OED, we analyse the OED entries for 245 sw monomorphemes with a view to carrying out a key word analysis and a semantic trait analysis. The 245 monomorphemes have a total of 469 senses out of which 330 can be classified into 18 recurring semantic traits in Table 1.

Then, in a second step, the comparison between the OED analysis of fl- and sw- monomorphemes shows that sw- words appear less likely to undergo any semantic change and therefore appear to be less indexical. In the light of these differing lexicographic behaviours, we aim, in a third step, to analyse the collocational behaviour of some common phonesthemic verbs carrying fl- and sw-. Collocational behaviour via a collexeme analysis will enable us to identify combinatorial patterns of use. For the study, we use the very large contemporary (2 billion words) OEC corpus (2000–2005) using Sketch Engine (Kilgarriff et al. 2004). The results of the compared analysis allow us to discuss whether phonesthemes are actual (sub)lexical “chunks” deserving of a lexical status, or whether they belong to larger phraseological “chunks” or units. This question raises the issue of the architecture of the lexico-grammatical continuum, the “constructicon”: does the constructicon accommodate or require a sublexical layer?

What are the repercussions for lexicography and phraseology?