Detecting directional forces in the evolution of grammar: A case study of the English perfect with intransitives across EEBO, COHA, and Google Books

TL;DR

This study investigates the evolution of the English perfect tense with intransitive verbs, providing evidence that natural selection influenced this grammatical change over time, based on large-scale historical language data.

Contribution

It combines multiple historical corpora with deep learning to analyze the directional forces behind the grammatical shift from be+PP to have+PP in English.

Findings

Most intransitive verbs shifted from be+PP to have+PP

Deep neural network classified the shift as driven by selection

Results support natural selection as a key factor in grammatical evolution

Abstract

Languages have diverse characteristics that have emerged through evolution. In modern English grammar, the perfect is formed with \textit{have}+PP (past participle), but in earlier English the \textit{be}+PP form also existed. It is widely recognised that the auxiliary verb BE was replaced by HAVE throughout evolution, except for some special cases. However, whether this evolution was caused by natural selection or random drift is still unclear. Here we examined directional forces in the evolution of the English perfect with intransitives by combining three large-scale data sources: EEBO (Early English Books Online), COHA (Corpus of Historical American English), and Google Books. We found that most intransitive verbs exhibited an apparent transition from \textit{be}+PP to \textit{have}+PP, most of which were classified as `selection' by a deep neural network-based model. These results suggest that the English perfect could have evolved through natural selection rather than random drift, and provide insights into the cultural evolution of grammar.