Abstract

Large-scale descriptions of academic English agree that modern scientific prose is increasingly built out of dense noun phrases rather than finite clausal elaboration. Yet existing diachronic accounts of this “nominal style” have focused primarily on the growth of phrasal modification, leaving open whether the semantic-syntactic type of nominal heads has also shifted in systematic ways. This study asks whether twentieth-century American informational writing increasingly favours eventive/process nominalisations (e.g., examination, development) over agentive/participant nominalisations (e.g., examiner, developer), a change that would align with the rhetorical pressures of abstraction, depersonalisation, and conceptual reification in scientific discourse.

Using decade-stratified samples from the Corpus of Historical American English (COHA; 1900–2000), we manually annotated 4,200 nominal tokens drawn from informational prose and a fiction control. Nominal heads were coded for eventive vs agentive type using formal diagnostics (argument-realisation cues, event-modification, and distributional properties), while noun-phrase context was coded for internal argument expression (especially of-PP complements, as in the analysis of the data), agent expression (by-PP/genitive, as in the analysis by Smith), and structural complexity (premodification, postmodification, and NP length). Mixed-effects modelling reveals a significant positive time trend for eventive nominalisations relative to agentive nominalisations in informational prose (β = 0.032, p < 0.001; odds +3.3% per decade, approximately +38% over the century). Eventive nominals also increasingly realise internal arguments via of-PP complements (β = 0.025, p = 0.002) while overt agent expression declines (p = 0.021). No comparable shift is attested in fiction.

These findings show that the historical move toward noun-phrase compression is not only a quantitative rise in modification but also a qualitative restructuring of nominalisation strategies: twentieth-century informational prose reorients from participant-reference toward process-packaging nominal heads that support argument embedding and hierarchical information packaging. The result is a grammar of scientific discourse in which processes are increasingly recast as manipulable “things”, enabling compact exposition while backgrounding human agency, a tendency that qualitative critics of academic style have also observed from rhetorical perspectives.

This work is licensed under a Creative Commons Attribution 4.0 License.