1 Relative clauses in Uralic Ksenia Shagal University of Helsinki 1. Introduction A relative construction is defined here as a construction consisting of a nominal ( head) and a subordinate clause interpreted as attributively modifying the nominal (relative clause) (cf. Lehmann 1986: 664). An important prerequisite for the construction is that the two clauses involved in it, the relative clause and the main clause, share an argument. This argument is referred to as the common argument (Dixon 2009: 314), and the syntactic position of this argument in the relative clause is the position relativized (Keenan and Comrie 1977: 63‒66). For example, sentence (1)a from Meadow Mari illustrates relativization on a locative participant in the inessive case, and in (1)b it is the object of postposition vokten that is relativized. Importantly, the common argument can be overtly expressed by a full noun phrase either in the main clause or in the relative clause, compare the use of the word pört ‘house’ in examples (1)a and (1)b respectively 1 : (1) Meadow Mari (Mari, Russia, Aralova and Brykina 2012: 522, 532) a. Pört jǝr, [ku-što tudo il-eš], šuko peledǝš ul-o. house near which-INE he live-PRS.3SG many flower be-PRS.3SG ‘Near the house in which he lives there is a lot of flowers.’ b. [Kudo pört vokte-n mašina šog-a] peš motor. which house near-INE car stand-PRS.3SG very beautiful ‘The house near which the car is standing is very beautiful.’ Relative clauses of the first type, e.g. (1)a, are, therefore, known as externally headed relative clauses, while (1)b is an example of an internally headed relative clause. Apart from being expressed by a noun phrase in one of the clauses, the common argument can have certain pronominal representation in the other one. For example, the head noun poškudǝlan ‘neighbor (DAT)’ in the internally headed relative clause in (2)a below is also referred to in the main clause by the anaphoric pronoun tudo ‘that’. This is a prototypical example of a so-called correlative relative clause, cf. Dryer (2013), which also differs from many other relative clause types in that it is clearly not embedded in the main clause, but is rather adjacent to it (Dixon 2009: 315). Within externally headed relative clauses, the head can be represented by an inflected relative pronoun, such as kušto ‘which (INE)’ in (1)a and (2)b. Structurally, these two examples differ in that (1)a features an embedded postnominal relative clause, which immediately follows the head noun, while in (2)b there is no embedding, and the relative clause is adjoined to the main clause. 1 The borders of dependent clauses are indicated by square brackets throughout the paper. When citing various printed sources, I generally keep the original glossing used by the authors (adapting it to follow the Leipzig Glossing Rules). When I refer to a participial marker in the text, in most cases I only use its most basic form, even if several morphophonological variants exist in the language. For example, I refer to Finnish agentive participles as -ma forms even though the marker changes to -mä after front vowels, e.g. syö-mä ‘eaten (by someone)’. The data used in this overview comes primarily from descriptive studies, but also from field work with language consultants (my own on Hill Mari and Daria Zhornik’s on Northern Mansi) and personal knowledge (Finnish). Importantly, some of the sources describe standard language varieties (Estonian, Finnish, Hungarian and North Saami), while for other languages the fullest available descriptions deal with colloquial data collected in the field (Pite Saami, Skolt Saami, and minority languages of Russia). Due to this discrepancy, in some aspects the data on different languages might not be directly comparable. However, a completely consistent study of Uralic relative clauses is hardly possible in the current state of the documentation and description of the family.