Mark C. Baker
Linguistic research has by now discovered many constructions which are quite rare crosslinguistically, but nevertheless appear in unrelated languages where they have consistent properties. This poses something of a dilemma for simple/standard views of Universal Grammar (UG): if UG does not apply to such constructions, then there is much that UG cannot explain about individual languages; but if UG does apply to such constructions, then it may need to attribute huge amounts of covert structure to all languages which is realized overtly only in a few. In this talk, I present an alternative vision, according to which rare linguistic features might be adaptations of the same UG-licensed infrastructure to different surface functions—much as the same basic forelimb structure can become a leg, an arm, a wing, or a flipper in different mammals. I illustrate the potential of this vision by considering five rare linguistic features: allocutive marking (addressee agreement), upward complementizer agreement, logophoric pronouns, indexical shift, and switch reference marking. The hypothesis is that all of these constructions are different adaptations of the possibility of having a null DP in the CP space of a finite clause. This DP can be agreed with, it can be controlled by a matrix argument, and it can bind an embedded pronoun. Different combinations of these formal ingredients give the five different constructions. I show that all five have substantive features in common (e.g. incompatibility with nominalization, subject sensitivity), and that superficial differences are not as big as has been thought (e.g. role of the embedded clause, object involvement). Furthermore, consideration of very rare languages that have two of the five constructions supports the claim that they are related, because there are strong correlations (positive or negative) between the constructions. Once one sees the five constructions as different surface manifestations of the same basic structural resources, the cluster of phenomena is not so rare after all, and shows itself to be a worthy topic of Universal Grammar. I conclude that better documentation and theoretical consideration of rare linguistic features should be a strategic area for growth in the next phase of comparative syntactic research.