Longobardi and Guardiano (2009) have argued that parametric grammar codes for history, showing that a comparison of the parameter values of the modern varieties may suffice to successfully reconstruct most European language families. For the significance of this method (PCM) it is also important to show that it may correctly capture the vertical relationship between mother and daughter languages. Here, we expect sister languages to progressively diverge through time from the common protolanguage and from each other. The grammars of modern Romance languages seem, however, measurably more similar to each other than to that of Latin, instantiating a significant convergence effect.
Convergence in the parametric values of separate languages can in principle be due to three sources: chance, contact (horizontal transmission), or the very structure of UG parameters. This article suggests that, within a structured theory of parameters and parametric phylogenies, Romance convergence can be explained through minimal appeal to areal contact, maximal reliance on the implicational structure of UG, and no need to invoke chance-based developments: hence it poses no particular problem to the validity of PCM.
