Scrambling Verb-Final Languages and the Underlying Order of Objects in Ditransitive Constructions

University of Pennsylvania Working Papers in Linguistics Volume 17 Issue 1 Proceedings of the 34th Annual Penn Linguistics Colloquium 1-1-2011 Scrambling Verb-Final Languages and the Underlying Order of Objects in Ditransitive Constructions Effi Georgala Cornell University, This paper is posted at ScholarlyCommons. http://repository.upenn.edu/pwpl/vol17/iss1/13 For more information, please contact . Article 13 Scrambling Verb-Final Languages and the Underlying Order of Objects in Ditransitive Constructions Abstract In this paper I provide evidence from depictive stranding to show that German and Turkish, scrambling verbfinal languages which have been assumed to be an exception to the crosslinguistic generalization of IO>DO base order in double object constructions, in fact support the generalization. Following Georgala's et al. (2008) analysis of applicative constructions, which predicts that indirect objects (IOs) originate higher than direct objects (DOs), I argue that German and Turkish have two types of applied arguments (thematic and raising) with different underlying but the same surface position, namely [Spec, ApplP]. By showing that IO>DO is the base order of Turkish double object constructions, I also contribute to the discussion of the nature of scrambling in Turkish. In particular I corroborate Öztürk's (2005) view that scrambling in Turkish can be treated as either A-bar or A-movement. This working paper is available in University of Pennsylvania Working Papers in Linguistics: http://repository.upenn.edu/pwpl/ vol17/iss1/13 Scrambling Verb-Final Languages and the Underlying Order of Objects in Ditransitive Constructions Effi Georgala* 1 Introduction Among scrambling verb-final languages there have been a number of different proposals about the syntactic structure of double object constructions (DOCs) and the underlying order of objects. In this paper, I focus on German and Turkish. German and Turkish have a variety of DOCs with dative indirect objects (IOs) and accusative direct objects (DOs), as in examples (1–2). (1) German a. Ich habe dem Kind das Bombon I.NOM have the.DAT child.DAT the.ACC candy.ACC ‘I gave the child the candy.’ b. Ich habe das Kind einer Gefahr I.NOM have the.ACC child.ACC a.DAT danger.DAT ‘I exposed the child to a danger.’ (2) Turkish Ben çocuğ-a şeker-i verdim I child.DAT candy.ACC gave ‘I gave the child the candy.’ gegeben given ausgesetzt exposed In the recent literature on ditransitive constructions it has been claimed that German (den Dikken 1995, Müller 1995, McGinnis 1999, Tungseth 2008, among others) and Turkish (Issever 2003, Kornfilt 2003, Simpson et al. 2008) counterexemplify the generalization that IOs merge higher than IOs (Marantz 1993, Pesetsky 1995, Bowers 2010, among others). In this paper, I provide evidence from previously unnoticed data from stranded depictives in support of the view that IO>DO is the underlying order in German and Turkish, thus showing that the German and Turkish data in fact support <IO, DO> as the universal underlying order in DOCs. Following Georgala et al.’s (2008) account of applicative constructions which predicts IO>DO as the underlying order, I propose that German and Turkish have both low- (raising) and high-type (thematic) applicative constructions, but a single applicative head above the lexical VP. The depictive stranding facts strongly support the view that in the low-type applicative construction, ApplP has a strong EPP feature that attracts the recipient IO from its underlying position in [Spec, VP]. Section 2 gives a brief overview of Georgala et al.’s (2008) raising/thematic applicative hypothesis and shows how it applies to German and Turkish. In Section 3 I discuss the depictive stranding data, while in Section 4 I provide a new argument from DOCs in support of the view that Turkish has both A- and A-bar scrambling. In Section 5 I conclude. 2 Raising/Thematic Applicative Hypothesis and the Syntax of German and Turkish DOCs 2.1 Raising/Thematic Applicative Hypothesis Marantz (1993), based on evidence from Bantu languages with morphological applicatives, argues that in DOCs the IO is introduced by a (potentially silent) applicative head (3). * Many thanks to Molly Diesing, Jaklin Kornfilt, John Whitman, and the audience of PLC 34 for valuable feedback. I am grateful to Waltraud Paul and Michael Wagner for the German judgments, and Esra Kesici and Jaklin Kornfilt for the Turkish judgments. U. Penn Working Papers in Linguistics, Volume 17.1, 2011 106 EFFI GEORGALA (3) [VP IO [V’ APPL [VP DO V]]] (Marantz 1993) Pylkkänen (2002, 2008) extends Marantz’s approach by motivating two kinds of Appl heads: a “high” applicative which denotes a relation between an individual and the event denoted by VP (4a), and a “low” applicative which denotes a dynamic relation of transfer of possession between IO and DO (4b). (4) a. High applicative: [VoiceP DPAGENT [Voice’ Voice [ApplP DPBNF/LOC/INSTR… [Appl’ Appl [VP V DP]]]]] b. Low applicative: [VoiceP DPAGENT [Voice’ Voice [VP V [ApplP DPGOAL/SOURCE [Appl’ Appl DPTHEME]]]]] Georgala et al. (2008) reconcile the above two accounts by proposing that there is only one position for applicative heads above the lexical VP which come in two flavors: thematic and raising. Like Pylkkänen’s high applicatives (cf. (4a)), thematic applicatives introduce an extra argument above the lexical VP. The extra argument is base generated in [Spec, ApplP] and is assigned a theta-role (e.g., beneficiary, maleficiary, instrumental) by Appl. (5) [vP SUBJ [v’ v [ApplP IOBNF/LOC/INSTR… [Appl’ Appl [VP V DO]]]]] Unlike thematic applicatives, raising applicatives do not introduce an extra argument, but attract the recipient/possessor goal IO from its base position in [Spec, VP] to their specifier. (6) [vP SUBJ [v’ v [ApplP IOREC [Appl’ Appl [VP tIO [V’ V DO]]]]]] The raising/thematic applicative hypothesis, thus, preserves Marantz’s original structural insight, but at the same time it also accounts for Pylkkänen’s ample evidence for two distinct types of extra objects, one originating outside the lexical VP, another inside it. In the raising/thematic applicative analysis both types are licensed with a single position for the licensing head. 2.2 German, Turkish, and the Raising/Thematic Applicative Hypothesis In this section I present a syntactic analysis of dative DOCs in German and Turkish, focusing on showing how the raising/thematic applicative hypothesis applies to both languages. 2.2.1 Two Types of Dative DOCs in German Before I proceed, I should stress that there is fairly general consensus in the literature that German has two structurally distinct classes of dative DOCs (Wegener 1991, McFadden 2004, Cook 2006, Meinunger 2006, McIntyre 2006, among others).1,2 In this paper I show that German in fact has three structurally distinct classes of DOCs: (i) “low” dative DOCs,3 (ii) raising applicative constructions, and (iii) thematic applicative constructio (...truncated)