On the functional vs. lexical nature of restructuring heads: evidence for a fine-grained classification of restructuring modals

ON THE FUNCTIONAL vs. LEXICAL NATURE OF RESTRUCTURING HEADS: EVIDENCE FOR A FINE-GRAINED CLASSIFICATION OF RESTRUCTURING MODALS Irene Balza* Abstract: This paper examines in detail the thematic and syntactic properties of a set of (non-)restructuring structures in Germanic, Romance and Basque. Based on the comparison of these structures, I propose a finegrained classification of (non-)restructuring constructions that include Functional Restructuring, SemiLexical Restructuring, Lexical Restructuring and Non-Restructuring constructions. The four types involve complements of different underlying structures (VPs, vPs or even as large as NegP/TPs). The results of the analysis argue against Cinque’s (2005) hypothesis that all restructuring verbs are exclusively functional, and contradicts Wurmbrand’s (1999) claim that, within the mixed class of restructuring verbs, modals must be raising verbs. Keywords: functional restructuring, (semi-)lexical restructuring, modals, Germanic, Romance, Basque 1. Introduction Recently advanced views on restructuring verbs fall into two main approaches: on the one hand, authors like Cinque (2004) propose that restructuring is universally restricted to functional heads; in contrast, Wurmbrand (2001, 2004) argues that, in languages like German, there also exists “lexical” restructuring (what she dubs LRI), where a lexical verb restructures with a small verbal complement. Within this debate, the analysis of Basque Modal Constructions (BMC) becomes particularly interesting. On the one hand, there is little research done concerning modal non-finite dependents in this language (see Etxepare and Uribe-Etxebarria 2009, 2010a and 2010b; henceforth E&UE); and, as we will see, the comparison between BMC and restructuring constructions in Germanic and Romance languages allows us to reach some firm conclusions regarding the underlying syntax of restructuring constructions in these languages. In addition, the analysis of BMC proves crucial in the debate concerning the functional vs. lexical nature of restructuring heads. In particular, I will show that the thematic properties of BMC contradict previous work where modal verbs are considered strictly functional (Cinque 2005) and raising heads (Wurmbrand 1999, 2004, Bobaljik and Wurmbrand 1999). In this paper1 , I adopt E&UE’s (2009, 2010a and 2003b) hypothesis that BMC involve different degrees of restructuring, Functional Restructuring (FRI), Lexical Restructuring (LRI) and Non-Restructuring (NRI), which differ with respect to the * UPV/EHU Bilboko Irasasleen Unibertsitate Eskola, Leioa (Bizkaia), . This research was funded by the Basque Government (Research Groups, GIC07/144-IT-210-07) the University of the Basque Country UPV/EHU (UFI11/14, HiTeDi/LingTeDi), and the Spanish Ministry of Science and Innovation, MICINN (FFI2011-29218). For feedback and discussion, I am grateful to Pablo Albizu, Maia Duguine, Ricardo Etxepare, Bryan Leferman, Celine Monoule, Beñat Oyharçabal and Myriam Uribe-Etxebarria. 1 72 Irene Balza underlying syntax of the complement selected by the modal (as argued by Wurmbrand 2001, 2004). However, I show that this classification needs to be expanded to include 4 different types of (non-)restructuring constructions: [VP …V0… ] F0 The modal is a functional head combining with a small (VP) size complement placed at its left side; Semi-Lexical Restructuring Type II constructions: [vP/VP …V0… ] V0 The modal is a semi-lexical head combining with a larger size complement (vP) placed at its left side; Lexical Restructuring Type III constructions: V0 [vP/VP …V0… ] The modal is a fully lexical head which imposes selectional restrictions to their subject and combines with a large size complement (at least a vP) located to its right; Non-Restructuring Type IV constructions: V0 [TP …V0… ] The modal is a lexical head which selects for a large size complement (at least a TP); these latter constructions present no restructuring effects. The paper is structured as follows: section 2 introduces the phenomenon of restructuring and summarizes the main approaches to restructuring proposed in the last years on the basis of Romance and Germanic languages. In section 3, I examine the (non)restructuring properties of BMC in detail. Following E&UE (2009, 2010), I first show that the transparency properties exhibited by these constructions vary depending on the word order the complement surfaces with respect to the modal. On the basis of this analysis, I present a classification of these constructions according to their level of (non)restructuring. In section 4, I introduce some of the typical tests used to determine the syntactic size of non-finite complements (presence/absence of temporal modification and the possibility of embedded negation within the non-finite complements), and I provide further support coming from multiple negation constructions. In section 5, I show that restructuring modals display different thematic properties and that we should distinguish between functional restructuring, semi-lexical restructuring, lexical restructuring and nonrestructuring. Finally, section 6 presents the main conclusions of the paper. Functional Restructuring Type I constructions: 2. The phenomenon of restructuring Some complex verb structures seem to behave like monoclausal structures in the sense that, although they involve more than one verb, they display clear transparency effects with respect to various syntactic operations. This phenomenon is commonly known as restructuring. In the next subsection I introduce some of the typical transparency effects exhibited by restructuring configurations in Romance languages and in German. On the functional vs. lexical nature of restructuring heads 73 2.1 Transparency effects associated with restructuring A typical transparency effect found in configurations involving restructuring is auxiliary switch, illustrated in (1). In (1a), the auxiliary that surfaces with the higher inflected verb is not determined by the modal verb but rather by the verb of the non-finite complement it combines with. Note that auxiliary switch is restricted to restructuring configurations and is not possible with verbs which select an inflected CP (1b)2. (1) a. Avrei / ?Sarei voluto andar -ci con Maria. AUXTR / AUX INTR1SG wanted goUNACC-CL with Maria b. ‘I would have wanted to go there with Maria.’ Avrei / *Sarei detestato andar -ci con Maria. AUXTR / AUX INTR1SG hated goUNACC-CL with Maria ‘I would have hated to go there with Maria.’ (Cardinaletti and Schlonsky 2004) Another typical clause-union effect, illustrated in (2) and (3), concerns clitic climbing. While in the Italian and Spanish examples (2a) and (3a) the clitics surface attached to the verb within the infinitival complement where they belong (andar-ci, decír-se-lo), in (2b) and (3b) they have moved out of the infinitival complements and appear immediately preceding the higher inflected verb. (2) a. b. Vorrei and (...truncated)