1 2 3 4 The Relative Order of Foci and Polarity Complementizers: 5 1 6 A Slavic Perspective 7 8 9 Abstract 10 11 According to Rizzi & Bocci’s (2017) suggested hierarchy of the left 12 periphery, fronted foci (FOC) may never precede polarity 13 14 complementizers (POL); yet languages like Bulgarian and Macedonian, 15 where POL may also follow FOC, seem to provide a counterargument to 16 17 such generalization. On the basis of a cross-linguistic comparison of ten 18 Slavic languages, I argue that in the Slavic subgroup the possibility of 19 20 having a focus precede POL is dependent on the morphological properties 21 of the complementizer itself: in languages where the order FOC < POL 22 23 is acceptable, POL is a complex morpheme derived through the 24 incorporation of a lower functional head with a higher one. The order 25 26 FOC < POL is then derived by giving overt spell-out to the intermediate 27 copy of the polarity complementizer rather than to the highest one. 28 29 30 Keywords : Left Periphery, Fronted Foci, Polarity Questions, 31 Complementizers, Bulgarian, Macedonian, Word Order. 32 33 34 I. Introduction 35 36 37 In this article, I will be concerned with accounting for cross-linguistic variation in the 38 relative distribution of two types of left-peripheral elements. The first are polarity 39 40 complementizers (POL), namely complementizers whose function is that of introducing 41 embedded polarity questions; the second are fronted types of constituents in narrow focus. 42 43 I provide an example of a configuration containing both elements in (1), from Italian; in 44 (1), the polarity complementizer “se” (=if) is marked in bold, whereas the fronted focus -in 45 this case, a PP- is in capitals. 46 47 48 1) Mi domando se A TROMSØ Espen sia nato (ITA) 49 50 Refl. I-wonder if IN TROMSØ Espen is(subv.) born 51 52 ‘I wonder if Espen was born IN TROMSØ’ 53 54 55 56 1 Thanks to Kjetil Rå Hauge, Yovka Tisheva and Victor Friedman for the helpful reviews and 57 comments. 58 59 60 61 62 63 64 65 1 2 3 Concerning the relative order of these two elements, we have two logical possibilities: either 4 5 the constituent in focus will follow the polarity complementizer, or it will precede it. In this 6 article, I will refer to the former configuration as alpha (2a), and to the latter as beta (2b). 7 Note that I use the symbol “<” to indicate linear precedence: 8 9 10 2) Order α: if < FOC (“se A TROMSØ”) 11 12 Order β: FOC < if (“A TROMSØ se ”) 13 14 According to Rizzi (2001) (see also Rizzi 2004, and Rizzi & Bocci 2017), foci and polarity 15 16 complementizers are associated with dedicated left-peripheral projections. In particular, the 17 18 complementizer “if” is said to be externally merged in the head of Int(errogative)P, a 19 projection sandwiched in between two other projections equally specialized for hosting 20 21 complementizers, namely Force and Fin . Unlike IntP, however, Force and Fin may 22 23 exclusively host declarative complementizers. Rizzi’s suggested derivation for fronted foci is 24 slightly more complex, in that these are not taken to be base-generated directly in the left 25 26 periphery. Rather, they are first externally merged in their argumental position, where they 27 28 can receive case and a theta role. Only at a second stage are they moved to the left 29 periphery, where they surface in the specifier of a dedicated Focus phrase. This dedicated 30 31 FocP is non-recursive, and is situated lower than IntP, as illustrated in (3). (3) represents 32 2 33 the most updated version of the cartographic hierarchy of the left periphery: 34 35 3) Force < TopP < IntP < TopP < FocP < TopP < ModP< TopP < QEmb < Fin < 36 37 IP 38 (Rizzi & Bocci 2017, ex. 29) 39 40 41 Evidence in favor of the idea that FocP must be lower than IntP comes from grammaticality 42 patterns such as the ones illustrated in (4) below. The pair in (4) shows that, in languages 43 44 like Italian –on which (3) is primarily based–, a constituent in narrow focus may only follow 45 46 the polarity complementizer “se”: 47 48 4) (ITA) 49 50 (a) Mi domando se QUESTO gli volessero dire 51 52 Refl. I-wonder if THIS to-him(cl) they-wanted to-say 53 ‘I wonder if they wanted to say THIS to him’ 54 55 56 57 2 At least at the time of writing (March 2019). 58 59 60 61 62 63 64 65 1 2 3 (b) *Mi domando QUESTO se gli volessero dire QUESTO 4 5 *Refl. I-wonder THIS if to-him(cl) they-wanted to-say THIS 6 (Rizzi 2001:289) 7 8 9 Based on the hierarchy in (3), we would then expect that the only way for a fronted focus 10 11 to precede a polarity complementizer would be for the focus to be moved to a left periphery 12 higher than the one in which POL is merged. In this case, the focus would linearly precede 13 14 the polarity complementizer simply by virtue of having being dislocated to a different –and 15 16 crucially higher– C-layer. This is illustrated in (5), where I use the notation “CP2” to 17 indicate a higher left periphery: 18 19 20 5) [CP2 … FOC … [CP1 … POL < FOC … [ TP … FOC …]]] 21 22 In this paper, I will be referring to configurations such as (1) and (4a) as relations of local 23 24 precedence, and to configurations such as (5) as non-local precedence. Based on the 25 26 hierarchy in (3), we would then expect that, in local configurations, only the α order should 27 be possible, whereas both α and β should be possible in non-local environments. 28 29 30 In this paper, we are interested in determining whether a restriction prohibiting foci from 31 32 locally preceding interrogative complementizers does indeed exist. To this end, we will be 33 investigating the relative distribution of foci and polarity complementizers in the Slavic 34 35 language group. We will see that while the beta configuration is indeed less frequent than 36 37 the alpha order, it is by all means attested. More careful analysis will however reveal that 38 the beta order is only possible in those Slavic languages (Bulgarian, Macedonian and 39 40 Bosnian-Serbo-Croatian) in which the polarity complementizer is a morphologically complex 41 42 word resulting from the fusion of two independent clausal markers. To account for the 43 specific subgroup in which the beta order is found, I will then argue that in these languages 44 45 the polarity complementizer is formed through the movement plus incorporation of a lower 46 47 functional head together with a higher focus marker. Configurations exhibiting the beta 48 order are then obtained by giving overt spell-out to the intermediate copy of the movement 49 50 chain, rather than to its head. In languages where only the alpha order is possible, on the 51 52 other hand, the polarity complementizer does not result from the incorporation of two 53 different clausal markers, but is rather generated directly in the position in which it 54 55 ultimately surfaces. 56 57 58 59 60 61 62 63 64 65 1 2 3 This article is structures as follows. In section II., I show that polarity questions are not 4 3 5 islands for focalization . This means that an intervention-effects analysis of the relative 6 order of left-peripheral constituents à la Abels (2012) is untenable, for focalization at least 7 8 (see author 2018). In section III., I investigate word order possibilities in ten different Slavic 9 10 languages, showing that only three languages in the group (Macedonian, Bulgarian and 11 partially Bosnian-Serbian-Croatian) have beta as a grammatical possibility. In section IV., 12 13 I establish a correlation between languages which display the beta order, and languages 14 15 which have the option of realizing the polarity complementizer not simply as a free particle, 16 but also an enclitic morpheme attaching to fronted XPs. Section V. illustrates the suggested 17 18 derivation of polarity questions featuring a fronted focus in Macedonian, Bulgarian and 19 20 Bosnian-Serbian-Croatian; in it, I argue in particular that the polarity complementizer in 21 these three languages result from the incorporation of the irrealis marker da , which is merged 22 23 low in the left periphery, together with the focus particle –li , which merged as the head of 24 25 the Focus projection. Section VI. is devoted to discussing the existence of apparently 26 problematic strings of clausal markers, which I use to further refine my analysis of the 27 28 irrealis particle da . Finally, in section VII., I briefly summarize the contents of the main 29 30 proposal. 31 32 33 34 35 II. Polarity Questions Are Not Islands For Focalization 36 37 38 In his (2012) paper, Abels provides an elegant and economical way of accounting for 39 precedence relations in the left periphery. Specifically, he argues that no constituent X will 40 41 ever precede a constituent Y if Y creates an island for X, as the movement of X across Y – 42 43 necessary to have X precede Y– would result in an intervention effect. We therefore predict 44 that the relative order of these two constituents will always be Y<X. To account for the 45 46 ungrammaticality of examples like (4b), Abels then argues that a focus may not precede 47 48 the polarity complementizer because focalization is blocked by weak islands, and polarity 49 questions, being questions, create weak islands.