The development of morphosyntactic structures in the process of diachronic semantic change involves both the grammaticalization of free lexical items to bound morphology (eg Latin cantare habemus> Spanish cantaremos ‘we will sing’) as well as the emergence and possible replacement of forms that exhibit degrees of semantic overlap with other forms in the grammar (eg the French compound perfect past or passé composé, which has replaced the simple past in spoken varieties). It has been noted that in the latter process increased frequency of the emergent form vis-à-vis another structure (or set of structures) can be taken as a benchmark for advanced stages of semantic change (Hopper and Traugott 1993: 126). Patterns of overall frequency, however, may obscure more specific trends in the evolution of grammatical structures that are subject to a variety of language-internal and external forces across languages and varieties. Thus, it seems that a more nuanced picture of semantic change requires a discussion of characteristics that go beyond simple frequency, taking into account different functional criteria related to the semantic distribution of these forms.

The current analysis questions the role of frequency in describing trends in the distribution of simple and compound forms of past reference across different varieties of Spanish. Various authors have observed that the (Present) Perfect (eg Diego ha terminado el trabajo ‘Diego has finished the job’) and the simple perfective past or Preterit (e)(eg Diego terminó el trabajo ‘Diego finished the job’) in Spanish can both be used to indicate reference to a past action in Spanish, though the Perfect is also said to carry the additional epistemic value of relevance to the moment of speech (see eg Brugger 2001). The distribution of these two forms varies across dialects, with some varieties demonstrating a greater preference (as measured by frequency) for one or the other form for indicating certain types of past reference (DeMello 1994, Penny 2000). In this analysis I develop a