Acceso usuarios
José Luis Blas Arroyo 
Using the tools of Historical sociolinguistics, and based on a corpus made of by private correspondence and other ego-documents (diaries, memoirs, etc.), this article analyses the patterns of variation and change detected in two key phases of the history of Spanish (the Golden Age and the first modern Spanish) around the same phenomenon of syntactic variation: the presence / absence of que in dependent clauses of a doxastic predicate (creer). The two mixed-effect regression analyses performed, one for each period, show a significant decrease, albeit moderate and sustained over time, in the levels of elisions with respect to classical Spanish, a period in which the vernacular variant reached its greatest apogee. On the other hand, in the 18th century some structural forces that already acted in the past are observed, together with others that imply a restructuring of old constraints, if not a break with the Golden Age period. However, the greatest changes can be seen at the socio-stylistic level. In this sense, the variable ceases to be a sociolinguistic marker, as it was in the past, to be conditioned in the 18th century by other factor groups, such as sex, but especially the migratory context. Thus, an important gap between European and American Spanish is seen in the data, with much higher numbers in favour of elision in the latter, suggesting the possible conservation of a syntactic archaism in America. The article also discusses some theoretical implications, both linguistic and social, derived from these results.
