Predicate initiality in space and time

Abstract

This dissertation consolidates data from available areal studies and hundreds of grammars on languages from across the globe to compile the most comprehensive atlas of predicate initiality in the world’s languages. It approaches predicate initiality (P1) from different perspectives: from a functional-theoretical perspective, from an areal-typological perspective, and from a historical perspective. This study identifies 811 languages that allow the placement of the predicate before the two core arguments of a transitive clause in pragmatically neutral statements, which represents around 11% of the world’s languages. The resulting typological picture is that P1 is not only relatively rare, but its global distribution is highly uneven. 88% of P1 languages are found in one of the five major P1 hotspots: on Pacific Islands (including Taiwan and the Philippines), in the Pacific Northwest coast of America, Mesoamerica, the Maghreb, or the East African Rift. In these areas, P1 order is the norm, either within a single language family (e.g. Austronesian in Pacific Islands), or across various neighboring language families in a linguistic area (e.g. in Mesoamerica). By contrast, P1 is virtually absent from all of Eurasia and Australia, and the linguistically very diverse regions of West Africa and New Guinea. P1 can be traced back to the earliest reconstructible stages for many P1 languages. Adding just the P1 languages from the six language families Austronesian, Otomanguean, Afroasiatic, Mayan, Salish, and Nilotic, for which the reconstruction of P1 is generally accepted, accounts for 80% of P1 languages. The Austronesian family alone accounts for almost half (346, or 43%) of the world’s P1 languages. Independent emergence of P1 is theoretically possible, but practically highly unlikely. Perhaps the only case where the historical development of P1 can be reconstructed in detail—thanks to a long literary tradition—are the Celtic languages and Modern Greek. Put simply, the reason for the rare independent emergence of P1 is that the common syntactic processes that alter constituent order lead away from P1: fronting operations (topicalization, focalization) typically target participants rather than actions, because the former are more salient and continuous in narratives. Hence, there are only few syntactic pathways for P1 to develop language-internally across time. P1 is not a diachronic target but rather tends toward a subject-predicate-object order over time. As a result, P1 almost exclusively spreads through diachronic diversification of languages and language contact. This is the reason why P1 forms such distinct areal clusters, with relatively few typological outliers, compared to predicate-initial or predicate-final constituent orders.