Location is everything: A typological gap in person restrictions and its support for a positional analysis
In many languages the co-occurrence of pronouns within a clause is allowed or disallowed based on their person value (e.g. the equivalent of “They showed you him” is grammatical while “They showed him you” may not be). In this talk I present the findings of a large-scale crosslinguistic survey of such restrictions, spanning 106 languages from 26 families and 3 isolates, and discuss its implications for the analysis of person restrictions and syntactic theory more generally. I establish a number of new typological generalizations, including what defines the class of pronouns that are crosslinguistically subject to person restrictions and, more importantly, a previously unnoticed typological gap that concerns interactions between the strength of the restriction (how many pronoun combinations are disallowed: e.g. only “*They showed him you/me” or also “*They showed me you/you me”) and the arguments involved in the restriction: subject and object or two objects. Specifically, when person restrictions between subjects and objects and between objects both exist in a language, the former are never stronger than the latter.
I propose that what defines the pronouns affected by person restrictions is that they are unspecified for a person value at the start of a syntactic derivation, and that the restrictions arise because the sources of person values are scarce (restricted to phase heads; i.e. heads that delimit the syntactic cyclic domain) and person valuation is heavily constrained by locality considerations. I argue that the typological gap shows that variation in restriction strength results from differences in the positions where the pronouns are valued for person, rather than parameterization of syntactic dependencies between pronouns and agreement heads, which has been the predominant assumption in recent work. The typological gap then results directly from independently needed assumptions about syntactic cyclic domains and argument structure. This new approach also makes it possible to explain why some person restriction patterns are much less frequent than others as well as to derive some previously unnoticed correlations between the presence/strength of person restrictions and other morphosyntactic phenomena (richness of voice and object agreement morphology, alignment splits, extraction asymmetries, etc.). Lastly, I suggest that the proposed analysis offers a possible reconciliation between the way person restrictions are treated in formalist (minimalist syntax) approaches and functionalist approaches.