Diagnosing Verb-second languages Anton Zim m erling (SMSUH Moscow/ Institute of linguistics, Moscow) 2 University of Tampere 04.12.2015 Guest Talk Venue University of Tam pere • Tim e: 4 Decem ber 20 15, 10 :0 0 -12:0 0 • Place: Pinni B50 0 5 3 University of Tampere 04.12.2015 Acknowledgement s • The paper is prepared with support from the Federal research project 2685 funded by the Russian m inistry of education and science. • I am grateful to the sem inar’s audience and personally, to Mikhail Mikhailov and Elena Kozerenko for their com m ents. The usual disclaim ers follow. • Anton Zim m erling’s research trip to Tam pere has been supported by the NCM-RU10 127 project. 4 University of Tampere 04.12.2015 2P: summary • I am going to render the notion and typological validity of the so called Second-Position Phenom ena, i.e. principles of linearization sensitive to the distance from the clausal (or phrasal) left edge rather than to the type of the preceding syntactic category. 5 University of Tampere 04.12.2015 2P clit ics • Wackernagel’s law (CL2) predicts clausal-second position of clustering clitics, i.e. weak stressed elem ents form ing clusters. • I propose a classification of word order system s with clustering clause-level clitics attested in genetically not related worlds language’s, including Slavic languages from the IndoEuropean fam ily, and discuss the triggers and param eters of clitic-external ordering and cliticinternal ordering. 6 University of Tampere 04.12.2015 V2 • Verb-Second constraint (V2) is in its canonic form attested in Germ anic languages (except for Gothic, English, Old English and Elder Futhark) and Kashm iri. • The so called sentence cartography is a universalist version of the Tem plate analysis proposed in 1930 -1950 s by the structuralists from the Prague and Copenhagen Linguistic Circles (Da. sætningsskem aet). 7 University of Tampere 04.12.2015 Subt ypes • I argue against broad definitions of V2. Strict V2 languages (cf. Danish, Germ an, Kashm iri) and V1-V2 languages (cf. Old Danish, Icelandic, Yiddish) exist. • ‘Partial V2 languages’, ‘residual V2 languages’ etc., where structures overtly resem bling V2 are attested only in one part of declarative clauses do not exist. 8 University of Tampere 04.12.2015 Synt act ic const raint s • Neither CL2 nor V2 are principles of Universal Gram m ar. They are ‘shallow’ constraints of narrow syntax, specific for classes of world’s languages. • I discuss syntactic, prosodic and inform ation structure m echanism s, which trigger secondposition phenom ena, in particular, the so called verb m ovem ent and constraints on the cooccurrence of syntactic categories (Com p vs V, Com p vs CL, V vs CL, Com p vs Wh-words etc). 9 University of Tampere 04.12.2015 1. Phonet ic and non-phonet ic 2P • Placem ent of SINGLE weak-stressed elem ents of a phrase – level can be non-syntactic (=phonetic). • Placem ent of CLUSTERING elem ents which are arranged in a rigid order (x, y), when they take contiguous positions, cannot be non-syntactic, since clusterization is bound to given syntactic position. • Placem ent of SINGLE elem ents representing given syntactic categories (verbs, com plem entizers, whwords etc) is always syntactic. 10 University of Tampere 04.12.2015 Clit ics, 2P and clust ering • 2P Finite verbs, 2P com plem entizers, 2P relative pronouns norm ally do not clusterize. • 2P wh-words occasionally clusterize (so called superiority effects). • Clitics and so called weak elem ents can both be clustering and non-clustering. 11 University of Tampere 04.12.2015 Clit ics • An extension of the enclitics vs proclitics classes (Zwicky). • Presum ably a universal class (though not necessarily present in all languages). • A com bination of prosodic deficiency and sy ntactic deficiency . • Special clitics vs sim ple clitics (Zwicky). 12 University of Tampere 04.12.2015 Mult if act or classif icat ions of clit ics • There are about 10 -20 (or m ore) diagnostic features of clitics. • It is plausible that phonetic clitics and syntactic clitics represent different sets of elem ents (cf. Sadock; Zalizniak, Zim m erling). 13 University of Tampere 04.12.2015 Clust ering vs non-clust ering clit ics • Clustering and non-clustering clitics usually represent different sets of elem ents in one and the sam e language. • NB! This fact can be used for autom atic parsing of clitic vs non-clitic sequences. • Adjacent clitics do not always belong to the sam e dom ain. Clitics in a cluster always represent the sam e dom ain. 14 University of Tampere 04.12.2015 Clit ics and weak element s • The so called weak form s (Cardinalletti & Starke) are a subclass of clitics which do not conform to strict m orphosyntactic criteria of presum ably canonic clitics. • Weak elem ents lack special m orphosyntactic double form s typical of clitic pronouns in Rom ance and Slavic languages. Cf. ORu. 1 Dat.Sg. Mně (non-clitic form ) vs 1 Dat.Sg. =m i (clitic). 15 University of Tampere 04.12.2015 Clust ering weak f orms: German • Modern Germ an (Haider 20 10 ) • NOM > ACC > DAT Daß=er=sie=ihr ja ausgesetzt hat ‚That he has exposed them to her‘ ?? daß=er=ihr=sie ja ausgesetzt hat. 16 University of Tampere 04.12.2015 Clust ering weak f orms: Middle Norwegian • Tem plate of clustering: • Subject pronoun > reflexive > object pronoun > deictic particles nu/ da > general negation / particle vel > postverbal particle • The tem plate is just as strict as clitic tem plates in Clitic-Second languages 17 University of Tampere 04.12.2015 Modern Norwegian • # Oc m eente | jeg m ig nu | hos hannem at vΦre fri for saadan sag (PCl 27) V1 lit."and considered | SUBJ REFL NOW| by him to be free from such things". • thi en part lade | sig det vel| befalde (PCl 96) V2 lit. "because a part of them let | REFL OBJ PART | please" Erling lod | sig det vel| befalde (PCl 21) V2 • Lit. "E. Let | REFL OBJ PART | please" 18 University of Tampere 04.12.2015 Case st udy 1: Slovenian • CL 2 (Com p - CL), no V2, SVO. Split NP/ DPs. • (Syntactic ) 2P clustering clitics (aux, pron, refl) which can be fronted in certain affirm ative contexts. • A. Videl=sem =ga. ‘I have seen him ’. • B. # Sem =ga=videl ‘I have indeed seen him ’. • Syntactic 2P non-clustering auxiliary clitic m ora which invariably takes 2P and cannot be fronted. 19 University of Tampere 04.12.2015 Case st udy 2: Osset ic • CL2, no V2, SOV, no split NP/ DPs, split verbal com plexes (endoclitics). • Preverbal com plem entizers, preverbal whwords, preverbal general negation/ Negation phrases, preverbal focus adverbials. • The verb m oves: SOV…, …V…, # V… , # ...V# • Preverbal Com ps are in com plem entary (!) distribution whith wh-words, not with verbs. 20 University of Tampere 04.12.2015 Case st udy 3: Kashmiri • SOV, consistent verb-final (left-branching), V2 in root clauses and som e subordinate clauses. • No clustering clitics (?). • Scram bling in the m iddle field. Neither subject nor object DPs have fixed slots if placed between the finite and non-finite verbs. • XP – Vfin {MIDDLE FIELD S + O + ADV} Vinf 21 University of Tampere 04.12.2015 Case st udy 4: Russian • SVO, no V2, no 2P clitics, • Split NP/ DPs, unbounded scram bling. • Verb m ovem ent. # VSO, # VOS, SVO, OVS, SOV, OSV. • [- EPP], generalized inversion (Kovtunova 1976, Bailyn 20 0 4) OVS ~ SVO, LocVS ~ SVLoc. • NB! Under a ‘broad’ definition of V2 Russian will be identified as ‘partial’ or ‘residual’ V2 language. 22 University of Tampere 04.12.2015 Case st udy 4: Old Icelandic • SVO, V1/ V2, unbounded scram bling, split NP/ DPs. • ‘Narrative inversion SV ~ VS. A wide com m unicative scale for V1 sentences. • Null subjects with role-and-reference properties AND pro (null them atic subject). • V2/ V3 clustering clitics (or clustering clauseinternal weak elem ents). 23 University of Tampere 04.12.2015 Case st udy 5: Jaure (Sout h Mande) • Finite operators (of pronom inal origen) in clausal 2P. • The initial position is invariably taken by gram m atical subject. • S AUX/ FIN O V • No verb m ovem ent. • No V2 effects, no XP-m ovem ent. 24 University of Tampere 04.12.2015 Syst ems wit h clust ering clit ics: Overview • Wackernagel: Syntactic distance counts from clausal left boarder: # XP-V; *XP-YP-V; # X/ XPCL, *X/ XP – Y -CL... • ‘Backernagel ~ Anti-Wackernagel’. Syntactic distance counts from clausal right border. • ...CL...# . Cf. Cysouw (20 0 5), Zim m erling (20 13). 25 University of Tampere 04.12.2015 W-Syst ems • W, W + , W*, V-systems with languages with clustering clitics (or clustering weak elem ents). • Standard W-system s with Wackernagel’s law: (W stays for Wackernagel’ or ‘Wor’d) • # X°/ XP – CL …..V. • Over 70 languages in different areas. 26 University of Tampere 04.12.2015 W+–syst ems: Cent ral Philippine Type • Clitic-verb adjacency superim posed on Wackernagel’s law: • Central-Philippine – Bulgarian type: the clitic has fixed position in a clause, the verb does not. # X°/ XP– [V – Cl] ~ # [V-Cl]. * X°/ XP – Y – CL • *# CLV; *# X°/ XP– CL…V; *# X°/ XP– V; # ....V….# 27 University of Tampere 04.12.2015 CL1/ CL2 languages • Macedonian is a partial parallel to Germ anic (and presum ably, to other) V1/ V2 languages: • # V-CL ~ CL-V • CL/ CL2 languages are rare, V2/ V2 languages are relatively com m on. 28 University of Tampere 04.12.2015 W+–syst ems: Old Germanic t ype • Old Germ anic (and likely, Middle Rom ance) type: the verb has fixed position(s) in a clause, the clitic does not – it takes adjacent positions. Mostly V1/ V2 languages • # XP – V – CL ~ # V- CL; * …V….CL 29 University of Tampere 04.12.2015 W*-syst ems. • Degraded Wackernagel system s, where fixed placem ent of clustering clitics is just one of the options: • Mudburru, Gurindi. • Polish, Old Church Slavonic. • Middle Greek. 30 University of Tampere 04.12.2015 V-syst ems • VP – internal (or vP- internal clitics). Verb m ovem ent. Neither the verb, nor the clitic have fixed position in a clause. • # ...[ VP CL + V]… • The clitic m oves together with the verb. • Most Modern Rom ance languages (with few exceptions – tentatively – Rom anian, RaetoRom ance, Ladino, European Portuguese). 31 University of Tampere 04.12.2015 C -syst ems • Scram bling languages without clustering clitics and fixed position of V. • Word order alternations (scram bling patterns) are predicted by inform ation structure (usually harm onized by m ovem ent constraints), cf. Kovtunova 1976. • Russian, Ossetic, Finnish, Estonian, Basque. 32 University of Tampere 04.12.2015 Barrier ef f ect s: general condit ions • Barrier effects: XP – CL….V >> { BARRIER XP} – V- CL; Cf. Zim m erling (20 0 2), Zim m erling (20 13), Zim m erling & Kosta (20 13). • In a num ber of languages with 2P clustering clitics initial topical constituents trigger late placem ent of clitics com bined with verb m ovem ent to 2P which gives rise to V2 & CL3 order in CL2 languages: the clitics/ clitic clusters adjoin not to the initial constituent, but to the m oved verb. 33 University of Tampere 04.12.2015 Barrier ef f ect s: Modern Slovak XP a. CL V [Vodic] =si zapálil cigaretu. driver REFLlit-PRF cigarette ‘The driver lit <him self> a cigarette’. XP- BARRIER V CL … zapálili=si b. {BARRIER[Vodic autobusu]} t i cigaretu. ‘The bus driver lit <him self> a cigarette’. 34 University of Tampere 04.12.2015 Barrier Ef f ect s: Old Novgorod Russian XP CL V a. životinoju = mi =s’a plati Cattle-INSTR CL.1Sg.DATCL.REFL.ACC pay-IMP ‘Pay m e with cattle!’ XP- BARRIER V CL platii = m i =s’a b. {BARRIER[v nedoborexъ]} životinoju t i crop setback-LOC pay CL.1Sg. DAT CL.REFL.ACC cattle-INSTR ‘In case of crop setback, pay m e with cattle!’ 35 University of Tampere 04.12.2015 Clit ic-verb adj acency or verb-clit ic adj acency? • In Svk and ONR, the verb does not have a fixed position in a clause, but it gets such position (i.e. V2) in derived structures with clitics and an initial topical Barrier: [BARRIER ] – V – CL. • Clause-level clitics in Bulgarian, Tagalog, Cebuano do not m ove out of 2P in these languages and attract the verbs to clitic-adjacent positions: XP- V-CL ~ VCL already in the basic word order # ...CL2, cf. Zim m erling & Kosta (20 13). When clitics are absent, the verb does not a fixed position. 36 University of Tampere 04.12.2015 A V2 perspect ive f rom t he clit ic t unnel • In languages, where the clitic linearization constraints are m ore prom inent (i.e. gram m aticalized) than verb linearization constraints (e.g. V2, V1/ V2, V1, V# etc), clitics attract verbs to clitic-adjacent positions, which explains the generation of V2 orders in standard W-system s with Wackernagel’s law (Slovak, Serbian/ Croatian/ Bosnian, Old Novgorod Russian) • and V1 ~V3 orders in W+-system s which have a com bination of 2P clitics and clitic-verb-adjacency. 37 University of Tampere 04.12.2015 Germanic V2 languages • It is plausible that Germ anic languages at som e m om ent of their history generalized V2 as an effect of the initial topical Barrier m echanism . • The Wackernagel’s parallels do not tell us, whether this m echanism of V2 gram m aticalization is the only one possible. 38 University of Tampere 04.12.2015 Phonet ic words and XP • Both 2P clitics and 2P verbs are functional words with reduced com m unicative value. They are placed to clausal 2P. • Attem pts of Anderson (1993) and Eyþórsson (20 15) to deduce 2P from local constraints one prosodic dom ains (XP + V allegedly had only one phrasal stress) are not verifiable. 39 University of Tampere 04.12.2015 V2 languages • Old em piric observations: a bottleneck condition = only one constituent before the finite verb. • Old generative interpretations: Head – m ovem ent + XP m ovem ent. • The first constituent m ust be m oved in order to fit the Agree m echanism (‘Spec – head agreem ent’). 40 University of Tampere 04.12.2015 Mot ivat ion f or verb movement in V2 languages • Holm berg (20 12, 20 15): • (i) A head within the C-dom ain bears a uPhifeature. • (ii) This head bears also an +EF • Sem i-typology: this approach does not specify, which C-head is the locus of these features – cross-linguistic variation. <if you rely on cartography and fine structure of C>. 41 University of Tampere Bot t leneck condit ion • Raeto-Rom ance (after Kaiser 20 15) 1a.[La m essa] ha it retg udiu The m ass has the king heard 1b. *La m essa il retg ha udiu 2a. [En Frontscha] ei igi im perator ius ‚The em peror has gone to France‘ 2b. *En Frontscha igi im perator ei ius. 04.12.2015 42 University of Tampere 04.12.2015 V2: generat ive & cart ographic account s • Germ anic: (after Roberts 20 12, 20 15). • Fine structure of the left-periphery (C-dom ain) Force …> Focus …> Fin • Fo rce has EF which triggers m ovem ent of exactly one category, which m ay in principle be m axim al or m axim al or m inim al, to its specifier. • Fo cu s m ay attract exactly one wh-phrase to its specifier. • Fin attracts V, and weak pronouns to its specifier (West Germ anic Only). Has p h -fe atu re s w h ich Agre e w ith th e ve rb. 43 University of Tampere 04.12.2015 V1/ V2 languages in cart ography • Old Icelandic, Old Danish, Yiddish, Faeroese, Icelandic, Old High Germ an etc. • If we believe in generative accounts of V2, we have to postulate silent operators for V1 orders in V1/ V2 languages. • V2 in subordinate clauses in V1/ V2 languages: Com ps are first-m erged in Fin and raise to Force (except in Yiddish and Icelandic). 44 University of Tampere 04.12.2015 Cart ographic f ailures • Focus as a site for the m oved wh-wordes is not exactly af ‘rhem a’ (only for declaratives). • Multiple topic positions intervening between Force, Focus and Fin are superfluous for V2 orders, but necessary for explaining the deviation from V2 in ‘partial V2 languages’. • Ad hoc explanation of V1 orders by postulating silent operators instead of recognizing overt verb m ovem ent. 45 University of Tampere 04.12.2015 Hypot het ical silent operat ors wit h V1 orders • Platzack (20 0 8) “…the edge feature of C is elim inated by an invisible operator like the Question Operator Q, Im perative Operator IMP… and elim ination of a <fram e topical pro>”. Alternative explanation: Verb m ovem ent, just as in non V2 languages. 46 University of Tampere 04.12.2015 Quest ion and ot her silent operat ors: Swedish • • • • a. # Har du köpt en ny cykel? <Q> b. # Kom m er han i m orgen? <Q> c. # Öppna fönstret! <Im p> d. # Fattar inte var jag ska få tag på det. <pro> • e. Var är tidningen? # Så jag nyss på bordet. <pro> • f. Nu är blom kålen m juk. # Tar m an upp den såhär. <fram e topic> 47 University of Tampere 04.12.2015 Embedded V2 and grammar t heory • V2 is presum ably at root phenom enon, hence the com plem entary distribution of V and Com p (which com pete for the T/ Fin slot). • But: if you see V2 orders in subordinate clauses, just claim that they have the ‘m ain clause order’. Explain the deviations from V2 in root clauses by the sam e way. • V2 in subordinate clauses in V1/ V2 languages: Com ps are first-m erged in Fin and raise to Force (except in Yiddish and Icelandic). 48 University of Tampere 04.12.2015 Comp vs V2 • Verb Movem ent + XP m ovem ent • Finite Operator takes a different position than Lexical (infinite) form of the verb • Finite Verbs and Com plem entizers com pete for one and the sam e position. • But see num erous stipulations for the deviating distribution. 49 University of Tampere 04.12.2015 Embedded V2: Germanic • Mainland Scandinavian: m ust be em bedded under a com plem entizer Sw. Han sa *(att) denna film har barnen sett. • Germ an: m ust be em bedded without a com plem entizer • Ger. Er sagt (*dass) diesen Film hat das Kind gesehen. • NB: The variation m ight be due not to different features ascribed to the com plem entizer, but to different param eter settings for the left periphery of subordinate clauses. 50 University of Tampere 04.12.2015 Embedded V2 (Kashmiri) • Root clauses (1a) vs em bedded that-clauses (1b) XP Vfin S O Vinf raath d yu t laRk-an tswaTh daar-yith. yesterday gave boy-Erg waste-Nom . throw ‘Yesterday the boy threw out waste’. XP Vfin Com p XP Vfin S O Vinf tem dop [ CP ki raath d u yt laRk-an tswaTh daar-yith]. ‘He said that yesterday the boy threw out waste’ 51 University of Tampere 04.12.2015 Similarit ies in t he mot ivat ion f or CL2 and V2 • Head-m ovem ent + XP-m ovem ent in {-EPP} languages, i.e. in languages where the preverbal position is NOT a subject position. • Roughly the sam e prescribed com plem entary distribution after Com p (but it is too often violated…) • Com p – Vfin ~ XP – Vfin • Com p – Cl ~ XP - Cl 52 University of Tampere 04.12.2015 Dif f erences in t he mot ivat ion f or CL2 and V2 • Clitic phrases (ClP) and clitic-internal ordering in 2P (Fin or Force). No finite operator clusterization. • CliticPs can include elem ents with interpretable features, finite verbs cannot (?). • Cf. illocutionary (Force and Focus) clustering clitics ZHE, LI, BO, TI, BY in Old Russian vs clustering ORu pronouns and auxiliaries. 53 University of Tampere 04.12.2015 Cl2 mot ivat ion in recent generat ive f rameworks: Force • Serbian (after Roberts 20 12, 20 15) • Fo rce has EF which triggers m ovem ent of exactly one category, which m ay in principle be m axim al or m axim al or m inim al, to its specifier. • In wh-interrogatives, it attracts exactly one whphrase to its specifier and attracts both Fin and Focus. • In yes-no interrogatives it host a null operator in its specifier and attracts T, if da is not m erged there. 54 University of Tampere 04.12.2015 Cl2 mot ivat ion in recent generat ive f rameworks: Foc and Fin • Serbian (after Roberts 20 12, 20 15). • Fo cu s hosts li and attracts all wh-phrases to its specifier. • Fin attracts clitics in virtue of its D- and phifeatures, except je, which is directly m erged there. The ordering of clitic m ovem ent reflects the cyclic derivation, so the operations are the reverse of the resulting surface order: nom (=aux) > dat> acc > gen > je. 55 University of Tampere 04.12.2015 Broad def init ion of V2: ‘ part ial V2 languages’ • There is a trend to broaden a definition of V2 by claim ing that the underlying m echanism s for verb m ovem ent are universal or at least widespread. • It is true that verb m ovem ent underlies V2, but only a vast m inority of SVO AND SOV languages has V2. 56 University of Tampere 04.12.2015 V2 crit eria and movement • • • • • The finite verbs take m ore than one position within the sam e type of clauses. Declarative clauses. OR Root declarative clauses. OR (…) Bottleneck condition (XP-m ovem ent). Gram m aticalizion and gram m ar-internal m otivation for verb m ovem ent (not just prosody or inform ation structure). 57 University of Tampere 04.12.2015 Part ial and residual V2 languages • Faking V2 criteria: PARTIAL V2 orders (= orders resem bling V2 in canonic V2 languages) attested only in one part of diagnostic clauses. RESIDUAL V2 orders: deviations / violations of V2 are ascribed to the fact that the language once had canonic V2, but lost it. • Dubious idea rather widely applied to English, Old English, Modern and Old Rom ance languages, etc. 58 University of Tampere 04.12.2015 Negat ive and int errogat ive clauses: AUX-insert ion • English is not a V2 language (and probably never has been), but is shows AUX2 (BE, DOinsertion) in negative clauses. • This is a relatively recent developm ent from the Middle English period, not just a survival of on older (and never attested V2 system ) 59 University of Tampere OVS inversion in Romance • Spanish • 1a.*La carta escribió J uan • 1b. La carta i, la=escribió i J uan Int.’ the letter wrote J ohn’. 04.12.2015 60 University of Tampere 04.12.2015 VS and AdvVS inversion in Romance Spanish: SV inversion is not m andatory, nor is it contingent on the presence of a fronted constituent. • 1a. Te m p ran o salió J uan de casa ‘J uan left the house e arly’. • 1b. Tem prano, salió J uan de casa. • 2. Escribió J uan la carta (V1). 61 University of Tampere 04.12.2015 Embedded V2 in Romance languages • Spanish: V2 is optional in independent clauses but m andatory in som e subordinate clauses. • 1a. No sé [qué com pró María] ‘I don’t know what Mary bought’ • 1b. *No sé [qué com pró María]. 62 University of Tampere 04.12.2015 Verb inversion in wh-quest ions: t est dat a f rom non-V2 languages • Not the sam e m echanism as V2, though in som e V2 languages this is m asked. • Standard and Central Basque (SOV, not V2). 1a.Nor ikusi du J onek? ‘Who seen AUX J ohn’ b. *Nor J onek ikusi du? • Northern Basque (SOV, not V2, a different AUX placem ent) 2a.Nor du J onek ikusi? b.*Nor J onek ikusi du? (after Kaiser 20 15). 63 University of Tampere 04.12.2015 Romance languages • Modern Rom ance languages generally ban OVS orders without clitic doubling, while Old Rom ance languages perm itted it. But this fact does not prove that f Old Rom ance language had V2. • Old. Sard. [Custu totu[ deti pro=ssa anim a sua a sancta Maria de Bonorcantu • ‘He gave [all this] for the good of his soul to St. Mary of Bonorcado’. 64 University of Tampere 04.12.2015 XP-V-S orders • A fixed position of postverbal subjects in partial V2 language is not a valid criterion, since V2 ~ V1/ V2 does not always bring about contraints for subject placem ent/ adjacent subject-verb orders. • Danish, Swedish (fixed slots for S) vs Germ an, Kashm iri, Old Danish, Old Swedish (scram bling in the m iddle field). 65 University of Tampere 04.12.2015 Old French: 1170 Bible • • • • • SXV orders É [Saül] [set jours] l’=atendi (Bible 1170 ) ‘And Saul seven days him awaited’ SXYZV (apparently verb-final) [Goliath] [par quarante jurs] [le m atin é le vesper] [a l’ost de Israel] vint (Bible 1170 ) • ‘For forty days, Goliath cam e to Israel’s arm y in the m orning and in the evening’. 66 University of Tampere 04.12.2015 Old French: f ict ion t ext s • no obvious exam ples of SXV (p. 239) • Cil dui, # sanz faille# , passerent de beauté toz les autres (Tristan 244) • ‘These two were no doubt m ore beautiful than all the others’ (presum ably a parenthesis) • XSV: [Et quant il parole], il dit And when he speaks, he says • [Mes Diue m erci et le sense Yselt ] il est tornez a garrison (Tristan 318) • ‘But because of God’s m ercy and Isolde’s knowledge he started to heal’ 67 University of Tampere 04.12.2015 Old English: a V2 language??? • • • • • • Bech & Salvesen (20 14) OSVAdvV Þæt Om erus se scop sweotelicost sægde (Or. 31) ‘Hom er the poet spoke that m ost clearly’. SO1O2VO2 7 [hy] [him ] [] forguldon [þone wigcræft þe hy æt him gæfter þæm grim m e leornodon] (Or. 22) • ‘And after that, they bitterly repaid him for the art of war that they learned from him ’. 68 University of Tampere 04.12.2015 An ult ra-cart ographic approach t o V2 variat ion • An attem pt to predict the param eters licensing V1 and V3 in presum ably residual V2 languages – Wolfe (20 15), exploiting the ideas of Roberts (20 12). • C-Fin V2 languages vs C-Force V2 languages. • Interesting, but the em piric base and the criteria are not quite valid. 69 University of Tampere 04.12.2015 C-Fin V2 languages according t o Wolf e 2015 • C-Fin V2 languages: Old Occitan, Old Sicilian. • V1 which is not lim ited to non-assertions. Topiccontinuity and Rhem atic V1. Sic. Tornau al m onisteriu. V1 ‘(they)returned to the m onastery’. Sic. Passau la sicundu e lu terzu iornu THETIC ‘the second and the third day passed by’. V3 is not just lim ited to Fram e-Setting Elem ents. V4 clauses result from the cooccurrence of a Fram eSetter, Topic and Focus. 70 C-Force-V2 languages according t o Wolf e 2015 University of Tampere 04.12.2015 • Old French, Old Spanish and Old Venetian. • V3 is a productive clause type. Triggered by elem ents first-m erged in the upper portion of the CP (the fram e topics). V4 not found, since orders whereby Topic and Focus co-occur bad. • V1 is either unattested in the later OFrench or found only in cases where the clause is nonassertive. (verba dicendi, after negation. Two interpretations. A null discourse operator or no null elem ents, underspecified for assertive force. 71 University of Tampere 04.12.2015 Conclusions, part 1 1.V2 and CL 2 are shallow constraints of narrow syntax, specific of their language classes, and not principle of UG. 2. Cartographic accounts do not save the idea of ‘partial’ or ‘residual’ V2 languages. 3. No operational diagnostics for partial V2 languages contra SVO languages / languages with scram bling & verb m ovem ent. 72 University of Tampere 04.12.2015 Conclusions, part II • 4. Verb-subject adjacency and verb-subject inversion are not a such diagnostic for V2 languages. • 5. Syntactic system s with a bottleneck conditions (CL2 and V2) are sim ilar, but clause-level clitics clusterize, while verbs and com plem entizers usually do not. 73 University of Tampere 04.12.2015 Conclusions, part III 6. V2 languages are a restricted class (m ostly Germ anic languages), CL2 languages and other system s with clustering clitics are m ore com m on. 7. The transition from CL2 system s to V2 system s is not explained yet, but Delbruck’s and Anderson’s m odel’s based on the gram m ar analogy (from clitic verb form s to all finite verbs are wrong). 74 University of Tampere 04.12.2015 Conclusion, part IV • 8. Inform ation structure and prosody are not sufficient triggers for CL2 and V2 orders, but they can influence som e derived C2 and (probably) V2 orders. • 9. Cartographic theories pretend for the role of UG principles, but rather look as extrapolations basing on ca. 10 -30 European languages, m ost of which have V2 and/ or clitics plus verb m ovem ent and wh-m ovem ent. 75 University of Tampere 04.12.2015 Ref erences (1) • Beninca, Paola & Poletto, Cecilia. Topic, Focus, and V2. In: L.Rizzi (ed.).The Structure of CP and IP: Oxford Studies in Com parative Sy ntax. Oxford. 20 0 4. P. 52-75. • Bech, Kristin & Salvesen, Christine Meklenborg. Preverbal word order in Old English and Old French. In: Inform ation Structure and Sy ntactic Change in Germ anic and Rom ance Languages. 20 14. Pp. 233 – 269. 76 University of Tampere 04.12.2015 Ref erences (2) • Bhatt, Rakesh M. Verb m ovem ent and the sy ntax of Kashm iri. Dordrecht. 1999. • Cardinaletti, Anna & Michal Starke. The Typology of Structural Deficiency: A Case Study of the Three Classes of Pronouns. In: H.van Riem dijk (ed.) Clitics in the Languages of Europe. Berlin, 1999. Pp. 145233. • Cysouw, Michael. 20 0 5. Morphology in the wrong place. A survey of preposed enclitics. In Wolfgang U. Dressler (ed.) Morphology and its Dem arcations, 17-37. Am sterdam : Benjam ins. 77 University of Tampere 04.12.2015 Ref erences (3) • Delbrűck, Berthold. Germ anische Sy ntax, Bd.II. Zur Stellung des Verbum s. Leipzig, 1911. 76 S. • Diderichsen, Poul. Sætningsbygningen i Skaanske Lov. Frem stillet som grundlag for en rationel dansk Syntaks / / Acta Philologica Scandinavica 15, Kρbenhavn, 1941. • Diderichsen, Poul. Elem entæ r Dansk Gram m atik. Kρbenhavn, 3 1976 ( 11946). • Faarlund J .T., Vannebo K.I., Lie S. N orsk referansegram m atikk. Universitetsforlaget, Oslo, 1997. 1223 p. 78 University of Tampere 04.12.2015 Ref erences (4) • Golston Chris & Brian Agbayani. Second position is first-position: Wackernagel’s Law and the role of clausal conjunction. Indogerm anische Forschungen 115 (20 10 ). 1-21. • Holm berg, Anders & Platzack, Christer. The Role of Inflection in Scandinavian Sy ntax. Oxford. 1995. • Legendre, Geraldine. Masked second-position effects and the linearization of functional features. In: Optim ality -theoretic sy ntax. Cam bridge. 20 0 1. Pp. 241277. • Holm berg, Anders. Verb Second. In: Sy ntax – an International Handbook of Contem porary Sy ntactic Research, T.Kiss & A.Alexiadou (eds.). Berlin: De Gruyter, 20 12. 79 University of Tampere 04.12.2015 Ref erences (5) • Lam brecht, Knud. Inform ation Structure and Sentence Form . Cam bridge. 1994. • Lyutikova, Ekaterina & Sergej Tatevosov. The clause internal left edge: Exploring the preverbal position in Ossetian / / ICIL3, Septem ber 11-13, Paris, 20 13. • Pollock J . Y. Verb Movem ent, Universal Gram m ar, and the structure of NP / / Linguistic inquiry 20 (1989), 365-424. 80 University of Tampere 04.12.2015 Ref erences (6) • Roberts, Ian. Phases, Head-m ovem ent and SecondPosition Effects. In: Phases developing the Fram ew ork. Angel E. Gallego (ed.). Berlin: De Gruyter, 20 12, 385-440 . • Rizzi, Luigi. The Fine Structure of the Left Periphery. In: Elem ents of Gram m ar: Handbook of Generative Sy ntax. Dordrecht. 1997. Pp. 281-337. • Vikner, Sten. Verb Movem ent and Expletive Subjects in the Germ anic Languages / / Oxford Studies in Com parative Sy ntax. New York, Oxford. Oxford University Press. 1995. 81 University of Tampere 04.12.2015 Ref erences (7) • Wackernagel, J akob. Über ein Gesetz der indogerm anischer Wortstellung / / Wackernagel J . Kleine Schriften. Bd. I, Basel, 1953, 1-10 3. • Wanner, Dieter. Second Position Clitics in Medieval Rom ance. In: Approaching Second: Second Position Clitics and Related Phenom ena. A.L.Halpern & A.M.Zwicky (eds.). Stanford. 1996. Pp. 537-578. • Zim m erling, Anton. Ty pological Scandinavian sy ntax. (In Russian). Moscow. 20 0 2. 82 University of Tampere 04.12.2015 Ref erences (8) • Zim m erling, Anton. Slavic w ord order sy stem s from the view point of form al ty pology . (In Russian) Moscow. 20 13. • Zim m erling, Anton, Kosta, Peter. Slavic clitics: a typology. STUF – Language Ty pology and Universals. Vol. 66. No 2. 20 13. Pp. 178-214. • Zim m erling, Anton. 1P orders in 2P languages. / / E.Lyutikova, A.Zim m erling, M.Konoshenko (eds.). Typology of m orphosyntactic param eters 20 15, vol. 2. Moscow: MSPU, Pp. 459-483.