The Integration of MILLION into the English System of Number Words : A Diachronic Study

Sammanfattning: This corpus-based variationist study traces the history of the English number word MILLION from its first attested use in the 14th century, focusing on the diachronic shift in function from one of heading its noun phrase (NP), as in the obsolescent (Obs) construction three millions of citizens, to one of a post-determiner of its NP head, as in Present-day English (PdE) three million citizens. The authentic historical materials used include both numerous electronic linguistic corpora and, more innovatively, very-large-scale general collections available online, especially historical newspaper collections from the US and UK comprising tens of billions of words. The latter sources necessitated the development of new methods for detailed processing of search results which enabled the documentation of the shift from Obs to PdE constructions in American (AmE) and British (BrE) English newspapers. The historical analysis of the ontological/semantic and morphosyntactic aspects of MILLION is based on universal constraints on number words. After extensive discussion of historical usage, including patterns of semantics and morphosyntax involving HUNDRED and THOUSAND in Old and Middle English, the study reaffirms the validity of these universals but restates them using updated terminology, seeing MILLION in the Obs construction above as exemplifying a condition of government in its NP and in the PdE construction one of dependency, the latter also being more economical and iconic and harmonizing better with semantics. The shift from Obs to PdE usage is shown to have occurred roughly three decades earlier in AmE newspapers (>50% ~mid-1880s) than BrE newspapers (>50% ~mid-1910s), and AmE is posited as easing the shift in BrE. Alternative numeral expressions (000,000) played a unique, dynamically related temporary role during the transition period, complicating statistical patterns. Finally, numerous factors, mostly linguistic, are documented in large-scale databases as respectively accelerating and retarding the usage shift.