This PhD thesis describes the change in the system of auxiliary selection in Spanish. Until the end of the 17th century haber ('have') + participle replaced ser ('be') + participle in intransitive past contexts. In a first step I demonstrate that the two constructions do not have the same function in Old Spanish. A quantitative analysis of the distribution of the two constructions in a corpus of Spanish historiographical texts demonstrates that haber + PtcP is typically used as an anterior, whereas ser + PtcP is typically used as a copula construction with resultative function.
In a second step I describe the relevant factors for the development of Spanish auxiliary selection in historiographical texts after 1425. The directedness and speed of the process by which ser + PtcP was lost can be described by (a) the prototypicality of usage contexts of the replaced construction ser + PtcP and (b) the absolute frequency of use of specific ser + PtcP syntagms.
I demonstrate that prototypical usage contexts of ser + PtcP are affected at a later point in time by the replacement process because these contexts are dissimilar to the original transitive usage contexts of the expanding construction haber + PtcP. As a result, in these usage contexts ser + PtcP tokens can be found in late texts. The diachronic gradualness of the replacement process results in the synchronic gradience documented in the Auxiliary Selection Hierarchy.
Highly frequent ser + PtcP syntagms are subject to processes of conservation: they lose the paradigmatic connection to the original construction. Consequently, late ser + PtcP examples display paradigmatic atrophy and are often reanalysed as anterior constructions. These processes of reanalysis interact with persistence effects. The use of ser + PtcP in the preceding co-texts increases the probability of use of ser + PtcP in the following co-text. These persistence effects gain relevance over time: ser + PtcP is used more conservatively, if ser + PtcP is used in the preceding co-text.