How to measure replacement: Auxiliary selection in Old Spanish bibles

Studies of the development of compound tense auxiliary selection in Spanish frequently analyse ser (‘to be’) + past participle (PtP) as an anterior construction, and its disappearance as a slow replacement process starting in Old Spanish, in which the new anterior auxiliary aver (‘to have’) replaced it. This article investigates and rejects the empirical basis for this claim on the basis of a comparative analysis of Old Spanish bible translations. It is argued that the majority of tokens of ser + PtP has a resultative function, as indicated by typical patterns of verbal mood, coordination and temporal-aspectual morphology. Old Spanish translators of the bible appear to have regarded aver + PtP as being more similar to simple imperfective preterit forms like cantaba (‘s/he sang’) than to ser + PtP. Comparing the types and rates of use of aver + PtP and ser + PtP in earlier and later bible versions with the help of generalised linear mixed-effects regression models shows that ser + PtP was more stable in Old Spanish than hitherto assumed. Rather than replacing ser + PtP in Old Spanish, aver + PtP expanded at the expense of simple preterit forms. In summary, this article provides empirical evidence against the replacement hypothesis for Old Spanish, while at the same time assessing ways to quantitatively identify replacement processes in diachronic linguistics.