Processing typology is a research program that examines cross-linguistic variation from the perspective of language processing and performance. It is based on the observation that the patterns of preference one finds in performance in languages possessing several structures of a given type appear to be the same patterns found in the fixed conventions of grammars, in languages with fewer structures of the same type. E.g. relative clauses may exhibit a 'gap' or a 'resumptive pronoun' strategy (Hebrew structures corresponding to the students [that I teach (them)]), a variant with or without a relative pronoun (English, cf. the students [(whom) I teach]), etc. One of these strategies can be 'fixed' or conventionalized in certain environments, while there can be optionality and free variation in others. The selection from the variants in performance exhibits patterns: e.g. greater distance between the head noun and the position relativized on results in more resumptive pronouns (Hebrew) and in more relative pronouns (English). The distribution of the fixed variants across grammars also reveals patterns. The issue is: are these the same patterns, structured ultimately by the same principles? The 'Performance-Grammar Correspondence Hypothesis' (Hawkins 1994, 2004) suggests that they are, and that grammars have conventionalized the preferences of performance.

The PGCH can account for numerous exceptions to proposed grammatical universals, for universals that are not grammatically predicted, and for many universals that are correctly described by grammatical models, albeit with much stipulation. It also predicts patterns in performance and processing, derived from the observed universals of grammar.

In this talk I examine several questions raised by this approach for the grammar and processing of relative clauses. I will build on the discussion in Hawkins (2004:ch.7), but my goal is more general: to take stock of what we now know, and don't know, in this area and to draw attention to what we need to find out. E.g. we don't yet have enough performance data involving relative clauses from a sufficiently diverse range of languages, or cross-linguistic grammatical data involving certain types of relative clauses, to establish to what extent the PGCH is supported. This paper formulates some of the PGCH's predictions for performance and grammars, summarizes relevant data that we have, and points to areas requiring further testing.

Hawkins, J.A. (1994) A Performance Theory of Order and Constituency, CUP, Cambridge.
Hawkins, J.A. (2004) Efficiency and Complexity in Grammars, OUP, Oxford.