Choice, chance and constraint: A multifactorial analysis of English permissive constructions in native and learner writing

This paper presents a quantitative, multifactorial analysis of variation among three English permissive constructions: let + V, allow + to V, and permit + to V. Using corpora of short argumentative essays and extended academic prose by native English speakers and Chinese EFL learners, we model the effects of nine semantic, syntactic, and stylistic predictors on the choice among permissive constructions with conditional inference trees and conditional random forests, revealing both shared and divergent patterns across the two groups. Results show that text type, causer animacy, and the semantics of the resulting‑event predicate are the strongest predictors of construction choice for both groups. While “let” functions as the default permissive across contexts, “allow” implies reduced agency and formal register, which is strongly associated with inanimate causers in both groups; “permit”, however, remains rare overall. Significant differences are observed at multiple levels of the conditioning factors. For native speakers, causer animacy is a robust predictor across genres; for Chinese learners, its effect is confined to extended academic prose, with negation conditioning choices in short argumentative essays. Additionally, in extended academic prose with animate causers, Chinese learners’ choice of permissive construction is conditioned by complement semantics, whereas native speakers are sensitive to the linear distance between the permissive verb and its infinitival complement. Furthermore, transitivity of the infinitival predicate significantly modulates native speakers’ choice between “let” and “allow”, whereas no corresponding effect is observed among Chinese EFL learners. Taken together, the findings, aligning with a view of language as choice and chance, advance constructional explanations of the no synonymy principle and statistical preemption.