Abstract
A hallmark of the Perfect/Preterit opposition is its instability; perfects tend to become something else, such as general pasts. In this article, I employ a sample of 40 Romance varieties to discuss this instability in the light of usage-based grammaticalization theory. The data-set illustrates that either Perfect or Preterit expansion is the likely outcome of a system in which both exist. Ultimately, I discuss these findings in the light of suggested motivations for morphosyntactic change, specifically the role processing plays in these. I briefly discuss the result of the change in the light of the distinction between source- and target-oriented explanations. The analysis uncovers a need for a more nuanced understanding of the assumed direction in the development of Perfects.