1 Sonia Cristofaro (University of Pavia) The referential hierarchy: reviewing the evidence in diachronic perspective In D. Bakker and M. Haspelmath (Eds.) Languages Across Boundaries: Studies in the Memory of Anna Siewierska. Berlin and New York: Mouton de Gruyter: 69-93. Abstract: Typological hierarchies are usually accounted for on synchronic grounds. If the dis- tribution of some construction, as described by some hierarchy, is associated with some particular factor, then that factor is assumed to be responsible for the distribution, independently of the actual diachronic processes that gave rise to the construction in individual languages. These processes, however, may pose several challenges for current explanations of typological hierarchies. This is illustrated in the paper with regard to the diachronic development of several patterns described by the so-called referential hierarchy (also known as the animacy or topicality hierarchy), including split alignment in case marking, hierarchical alignment, and the presence of singular vs. plural dis- tinctions for different NP types. The available diachronic evidence shows that individual patterns originate from mechanisms different than those that can be postulated on synchronic grounds, that the various instances of a particular pattern need not always originate from the same mechanism, and that different patterns pertaining to the same grammatical domain are also a result of different mech- anisms. This suggests that, contrary to the traditional view, the patterns described by the referential hierarchy do not reflect any general principle. Rather, the hierarchy is best regarded as a descriptive schema that is general enough to capture the outputs of several independent diachronic processes. 1 Introduction Through her work, Anna Siewierska has crucially contributed to expanding and systematizing our understanding of a number of grammatical phenomena related to the so-called referential hierarchy reported in (1), including for example zero as opposed to overt bound person marking, hierarchical alignment, alignment splits, and voice (Siewierska (1998), (2004), (2010b), to mention but a few). (1) The referential hierarchy (Corbett (2000: 56), among others): 1st person pronouns > 2nd person pronouns > 3rd person pronouns > kin > human > animate > inanimate The referential hierarchy (also known as the topicality or animacy hierarchy) has been argued to reflect a variety of factors, including, for example, animacy, topical- ity, definiteness and natural attention flow (see Song (2001: chap. 3) for a review). These explanations, as is usually the case with typological explanations, have been proposed on synchronic grounds. If the cross-linguistic distribution of some con- struction is associated with some particular factor, and a plausible relation can be postulated between that factor and the properties of the construction, then the factor is assumed to be responsible for the distribution, independently of how the construc- tion actually originated in individual languages.If the cross-linguistic distribution of some construction is associated with some particular factor, and a plausible relation can be postulated between that factor and the properties of the construction, then