Syntactic changes in verbal clauses and noun phrases from 1500 onwards

Abstract

Can the promise of data-driven methods hold in historical linguistics? Can they detect salient syntactic changes and open new research avenues? I first use data-driven measures to detect patterns in the ARCHER corpus. Secondly, I qualitatively interpret the differences and build hypotheses. Thirdly, I validate these, with the Penn Corpora, investigating frequency and creativity. We observe several trends – among them: verbal (‘Doric’) style is decreasing, nominal (‘Attic’) style is increasing. A cascade from full to non-finite clause, and from paratactic to hypotactic style unfolds. Furthermore, constituent order is increasingly becoming fixed, strengthening the principle of dependency length minimisation. While data-driven approaches entail a complex interpretation step, their holistic perspective goes beyond well-trodden envelopes of variation to more global language models.