Comparative correlatives and parameters

The comparative correlative construction, exemplified by The more I read, the more I understand, has a number of idiosyncrasies, highlighted in Culicover and Jackendoff (1999), which pose a problem for the view that grammatical systems are the result of setting a relatively small number of parameters. Den Dikken (2005) suggests that the construction is no problem for a principles-and-parameters approach to language. However, he provides no analysis and ignores the idiosyncrasies. He proposes that the first clause is a kind of free relative. This seems to explain some of its properties, but the same properties in the second clause are unexplained. Den Dikken shows that a number of languages have broadly similar CC constructions. However, data from French suggests that he seriously underestimates the extent of cross-linguistic variation. The first clause of the French construction does not resemble a free relative and for some speakers is not subordinate clause. In addition the two clauses may be connected by the conjunction et ‘and’. Den Dikken argues that the cross-linguistic variation is lexical and not syntactic. Within his general approach this is true. However, it does not make the variation any less problematic for a principles-and-parameters approach.