West Coast Conference on Formal Linguistics (WCCFL) (38th : 2020)

Long-distance scrambling in Balkar and the nature of edges Bondarenko, Tanya; Davis, Colin 2020

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
73804-Bondarenko_T_et_al_Long_distance_WCCFL38_2020.pdf [ 191.4kB ]
Metadata
JSON: 73804-1.0389838.json
JSON-LD: 73804-1.0389838-ld.json
RDF/XML (Pretty): 73804-1.0389838-rdf.xml
RDF/JSON: 73804-1.0389838-rdf.json
Turtle: 73804-1.0389838-turtle.txt
N-Triples: 73804-1.0389838-rdf-ntriples.txt
Original Record: 73804-1.0389838-source.json
Full Text
73804-1.0389838-fulltext.txt
Citation
73804-1.0389838.ris

Full Text

Long-distance scrambling in Balkarand the nature of edges*Tanya Bondarenko & Colin Davis / MIT / {tbond, colind}@mit.eduWCCFL 38 / University of British Columbia1 Introduction• In this project, we use fieldwork data about cross-clausal scrambling in Balkar (Turkic) toargue for several concepts about the constraints on movement.• A great deal of research has argued that certain constituents, in current terms phases (Chom-sky 2000, 2001, 2008, a.o.), are unique in only allowing constituents in their edge to beaccessible by later syntactic operations.• Generally, CP, vP, and sometimes DP are taken to be phases. If so, we expect that movementfrom these constituents will generally have to pass through their edge.⊲ The trees in (1) below illustrate this expectation for movement from CP, since movementfrom embedded clauses is our focus here:(1) General expectation: Must exit a phase via its edgea. Legal exit via edge...XXP1 ...... VPV CP[Phase]t1 C TPt1 ...b. Illegal exit from below edge...*XP1 ...... VPV CP[Phase]C TPt1 ...• We argue that the way in which long-distance scrambling interacts with the different embed-ded clause types in Balkar reveals insight into what happens at phase edges.*Authors listed alphabetically. We thank David Pesetsky, Norvin Richards and the members of the Lomonosov Moscow StateUniversity fieldtrip to Balkaria for their valuable feedback.1• Balkar has three types of embedded (nominalized) clause, which are differentiated based onthe case of their subject—nominative (nom), accusative (acc), or genitive (gen).(2) Three possibilities for embedded subjects: nom, gen or accUstazteacher.nom[Clause [fatima-n1Fatima-gensabij-i-∅child-3-nom/sabij-i-ni/child-3-gen/sabij-i-n]/child-3-accalma-n1apple-accaša-Kan-1-neat-nfut-3-acc] ešit-ti.hear-pst‘The teacher heard [that Fatima’s child ate her apple].’⊲ In (2) and throughout this presentation we used possessed subjects in embedded clauses,since in the absence of possessive marking accusative and genitive are syncretic.• Based on differences in how each variety of subject interacts with cross-clausal scrambling,we argue for the following general concepts about phases and their edges:⊲ #1: That CP is a phase which allows multiple specifiers provided that tucking-in applies(Richards 1997, 1999).⊲ #2: That the highest phrase in a multiple specifier structure is privileged, such that a higherspecifier must move before a lower one can be accessed (Bošković 2016, a.o.).⊲ #3: That DP is a phase which (at least in this context) does not allowA′-movement throughits edge (Bosque & Gallego 2014, Reeve 2018, van Urk 2019).• We will show that these general ideas make sense of a variety of facts about cross-clausalscrambling in Balkar, and facilitate an understanding of several related patterns.• First we consider scrambling of objects in detail, before turning to the properties of subjectmovement later on.1.1 Contents of this presentation• §2 - The main scrambling facts and the puzzle they present.• §3 - Background on Balkar embedded clauses.• §4 - Analysis of the constraints on object scrambling.• §5 - Locality and subject scrambling.• §6 - Conclusion, followed by the appendices.22 Subject case and constraints on scrambling• Each variety of embedded subject interacts with cross-clausal scrambling differently.• First, note that scrambling to position immediately preceding the embedded subject is pos-sible only if that subject is nominative:(3) No clause-internal scrambling over acc/gen subject, ok over nom subjectUstazteacher.nom[Clause [tauušnoiseet-dir-ip]make-caus-convalma-n1푘apple-acc[bala-s1child-3.nom/*bala-s1-n1/child-3-gen/*bala-s1-n]/child-3-acct푘 aša-Kan-1]-neat-nfut-3-accešit-tihear-pst‘The teacher heard that her child ate the apple loudly (lit. ‘while making noise’).’⊲ In (3), an adverb is used as a signpost for the embedded clause’s edge, clarifying that weare attempting clause-internal scrambling here.1• If scrambling to an edge position above only nominative subjects is possible, then given thehypothesis that embedded clauses are phases, it is unsurprising that only with a nominativesubject is scrambling into the matrix clause permitted:(4) No long-distance scrambling over acc/gen subject, ok over nom subjectAlma-n1푘apple-accustazteacher.nom[Clause t푘 [fatima-n1Fatima-gensabij-i-∅/*-ni/*-n]child-3-nom/*gen/*acct푘aša-Kan-1-neat-nfut-3-acc] ešit-ti.hear-pst‘The teacher heard [that Fatima’s child ate her apple].’⊲ Thus accusative/genitive subjects uniquely ‘plug’ the edge of the embedded clause, pre-venting scrambling from reaching the matrix clause by passing through that position.• Another trait of accusative and genitive subjects is that they themselves can scramble intothe matrix clause:(5) Long-distance scrambling of acc/gen subject[Fatima-n1Fatima-genbala-s1-n1/n]푘child-3-gen/accustazteacher.nom[Clause t푘 alma-n1apple-accaša-Kan-1-neat-nfut-3-acc]ešt-gen-di.hear-part-31See appendix A for evidence that this adverb can only occur within the embedded clause.3‘The teacher heard that Fatima’s child ate an apple.’! Importantly, when the accusative subject moves from the embedded clause, that clause’sobject can do so as well (6a). The final order S < O is necessary in this situation (6a vs. 6b):(6) Accusative subject scrambling feeds long object scramblinga. [Fatima-n1Fatima-gensabij-1-n]푘child-3-acctüneneyesterdayalma-n1 푗apple-accustazteacher.nom[Clause t푘 t 푗aša-Kan-1neat-nfut-3-acc] ešit-ti.hear-pst‘The teacher heard that Fatima’s child ate the apple yesterday.’b. *Alma-n1 푗apple-acctüneneyesterday[fatima-n1Fatima-gensabij-1-n]푘child-3-accustazteacher.nom[Clause t푘 t 푗aša-Kan-1neat-nfut-3-acc] ešit-ti.hear-pst‘The teacher heard that Fatima’s child ate the apple yesterday.’• But in contrast, genitive subject movement does not feed object scrambling of any form:(7) Genitive subject scrambling does not feed object scramblinga. *[Fatima-n1Fatima-gensabij-1-n1]푘child-3-gentüneneyesterdayalma-n1 푗apple-accustazteacher.nom[Clause t푘 t 푗aša-Kan-1neat-nfut-3-acc] ešit-ti.hear-pst‘The teacher heard that Fatima’s child ate the apple yesterday.’b. *Alma-n1 푗apple-acctüneneyesterday[fatima-n1Fatima-gensabij-1-n1]푘child-3-genustazteacher.nom[Clause t푘 t 푗aša-Kan-1neat-nfut-3-acc] ešit-tihear-pst‘The teacher heard that Fatima’s child ate the apple yesterday.’F These are the patterns we focus on deriving in this presentation. In summary:⊲ Scrambling the embedded clause’s object across a nominative embedded subject is licit.⊲ Accusative subjects block such object scrambling, unless they scramble as well.⊲ Genitive subject scrambling, though possible, never feeds object scrambling.43 Background: Characteristics of Balkar embedded clauses• In this section, we provide an understanding of the structure of each embedded clause type,which will serve as our foundation for an analysis of the patterns just described.3.1 Embedded clause structure• All the clauses we consider behave like nominals: they have agreement marking matchingthat seen in nominal phrases (specifically possessive constructions), appear in argumentpositions, and carry case morphology.• But all also have some verbal properties. We hypothesize the following structures:(8) Embedded clause contents (building from Bondarenko 2018)Case of the subject Nominal structure Verbal structurenom, acc NP CP-TP-AspP-vP-VPgen DP-NP AspP-vP-VP• All three clause types have at least enough verbal structure to host VP-level adverbs, as (9)below shows with a manner adverb:(9) VP-level adverb in all clause typesUstazteacher.nom[[bala-s1child-3.nom/bala-s1-n1/child-3-gen/bala-s1-n]/child-3-acctauušnoiseet-dir-ipmake-caus-convalma-n1apple-accaša-Kan-1-n]eat-nfut-3-accešit-ti.hear-pst‘The teacher heard that her child ate the apple loudly.’• All clause types also permit negation:(10) Negation in all clause typesUstazteacher.nom[[fatima-n1fatima-gensabij-ichild-3.nom/sabij-i-ni/child-3-gen/sabij-i-n]/child-3-accalmaappleaša-ma-Kan-1-n]eat-neg-nfut-3-acckör-dö.see-pst‘The teacher saw that Fatima’s child did not eat an apple.’• All three also plausibly contain some degree of functional material relating to tense/aspect,since all can be built from either a non-future participle (-Kan ‘nfut’) or a future-orientedone (-r1q ‘fut’).5• Most examples shown so far use -Kan. Example (11) below illustrates -r1q:(11) Future marking -r1q allowed in all clause typesOl(s)he.nom[bala-s1child-3.nom/bala-s1-n1/child-3-gen/bala-s1-n]/child-3-acc(tambla)(tomorrow)alma-s1-napple-3-accaša-riK-1-neat-fut-3-accajt-a-d1.say-ipfv-3sg‘(S)he is saying that (someone’s) child will be eating his/her apple (tomorrow).’⊲ Unlike nominalizations with acc and nom subjects, nominalizations with gen subjectscannot have temporal modification that is in contradiction with that of the matrix clause:(12) Tense of gen subject clause must match that of matrix clausea. KerimKerim.nomtüneneyesterday[[fatima-n1Fatima-genbala-s1child-3.nom/bala-s1-n]/child-3-acctamblatomorrowalim-niAlim-genkištig-i-ncat-3-accbaKar-l1q-1-n]feed-fut-3-accbil-di.know-pst‘Kerim found out yesterday that Fatima’s child will feed Alim’s cat tomorrow.’b. *KerimKerim.nomtüneneyesterday[[fatima-n1Fatima-genbala-s1-n1]child-3-gentamblatomorrowalim-niAlim-genkištig-i-ncat-3-accbaKar-l1q-1-n]feed-fut-3-accbil-di.know-pst‘Kerim found out yesterday that Fatima’s child will feed Alim’s cat tomorrow.’⊲ We therefore hypothesize that embedded clauses with genitive subjects uniquely lack T(or perhaps have one that is in some sense ‘defective’ and thus semantically deficient).• We also suggest that embedded clauses with genitive subjects may be in a sense ‘morenominal’ than the others, since they can more easily be used with elements like quantifiersand numerals (though this is not an absolute contrast):(13) Quantifiera. ?Tüneneyesterdayustazteacher[[fatima-n1Fatima-genbala-s1child-3.nom/bala-s1-n]/child-3-accalma-n1apple-accxareveryaša-Kan-1-n]eat-nfut-3-accešt-gen-dihear-nfut-3sg‘The teacher heard every eating of the apple by Fatima’s child yesterday.’b. Tüneneyesterdayustazteacher[[fatima-n1Fatima-genbala-s1-n1]child-3-genalma-n1apple-accxareveryaša-Kan-1-n]eat-nfut-3-accešt-gen-dihear-nfut-3sg‘The teacher heard every eating of the apple by Fatima’s child yesterday.’6(14) Numerala. ?Tüneneyesterdayustazteacher[[fatima-n1Fatima-genbala-s1child-3.nom/bala-s1-n]/child-3-accalma-n1apple-accekitwoaša-Kan-1-n]eat-nfut-3-accešt-gen-dihear-nfut-3sg‘The teacher heard two eatings of the apple by Fatima’s child yesterday.’b. Tüneneyesterdayustazteacher[[fatima-n1Fatima-genbala-s1-n1]child-3-genalma-n1apple-accekitwoaša-Kan-1-n]eat-nfut-3-accešt-gen-dihear-nfut-3sg‘The teacher heard two eatings of the apple by Fatima’s child yesterday.’• We hypothesize that genitive subject clauses are unique in containing the DP layer, whilethose with nominative and accusative subjects contain a more minimal amount of nominalstructure, perhaps just NP.• This thus leaves us with the following split between clause types, which will expand to athree-way distinction once we consider the properties of each subject type:(15) a. Clause with nom/acc subjectNPCP[Phase]TPAspPvP AspTCNb. Clause with gen subjectDP[Phase]NPAspPvP AspND3.2 Subject positions and case assignment• We posit a distinct position for each type of subject, based on which their interaction withscrambling, and additional related facts, will be derived.• First, we hypothesize that when the embedded clause’s subject is nominative, the subject isassigned case by and thus moves to the specifier of TP:7(16) Nominalized clause with nominative subject in spec-TPNPCP[Phase]TPSNOMAspPvPt푆 ...AspT푢푁푂푀CN• Second, we hypothesize that what distinguishes embedded clauses with nominative andaccusative subjects is that in the latter type, T lacks the ability to assign nominative case.• We thus assume that a subject gains accusative case marking by bypassing TP and landingin the edge of CP, where it is accessible for case assignment by the matrix V:(17) Nominalized clause with accusative subject in spec-CP...NPCP[Phase]SACCTPAspPvPt푆 ...AspTCNV푢퐴퐶퐶⊲ That accusative case on the subject is assigned by matrix V is supported by the fact thataccusative subjects are banned in clauses that are subjects (18), and the fact that when thematrix V cannot assign accusative case, an accusative subject is impossible (19):8(18) No acc subject within a clausal subject[[Fatima-n1Fatima-gensabij-ichild-3.nom/sabij-i-ni/child-3-gen/*sabij-i-n]/child-3-accalmaappleaša-Kan-1]eat-nfut-3.nomigi-di.good-3‘That Fatima’s child ate an apple is good.’(19) No acc subject if matrix V does not independently assign ACCAlimAlim[[fatima-n1Fatima-gensabij-ichild-3.nom/sabij-i-ni/child-3-gen/*sabij-i-n]/child-3-accmašinacarbuz-Kan-dan]break-nfut-ablqurqu-ruq-du.be.afraid-fut-3‘Alim will be afraid of Fatima’s child breaking a car.’• Finally, we assume that in embedded clauses with genitive subjects, an absence of T (orpresence of a highly defective one) is compensated for by merge of D, which assigns case toand triggers movement of the subject:(20) Nominalized clause with gen subject in spec-DPDP[Phase]SGENNPAspPvPt푆 ...AspND푢퐺퐸푁4 Analyzing the scrambling facts• Here we will argue that the facts about scrambling from embedded clauses are accuratelypredicted, given what we’ve proposed above about the properties of each clause type.⊲ Following the assumptions of much current work, we take CP and DP to be phases, butnot NP and TP, as marked in many of the diagrams shown above.2• We begin with nominative subjects, whose interaction with object scrambling is very simple.2This analysis has no bearing on the phasehood of vP.94.1 Object scrambling in clauses with nominative subjects• If nominative subjects sit in the specifier of a TP dominated by CP, such subjects are notexpected to have any interaction with scrambling of an object from the embedded clause.(21) Nominative subject in spec-TP[푁푃 [퐶푃 [푇푃 Snom [푣푃 t푆 O V-v ] T ] C ] N ] V ...• And indeed, scrambling across nominative subjects, presumably via spec-CP, is acceptable:(22) NOM subject does not interact with long-distance object scrambling[퐶푃 O ... [푉푃 [푁푃 [퐶푃 t푂OO [푇푃 Snom [푣푃 t푆 t푂OO V-v ] T ] C ] N ] V ] C ]4.2 Object scrambling in clauses with accusative subjects• In contrast, we have seen that object scrambling from a clause with an accusative subject isunacceptable under normal circumstances.• We have hypothesized that accusative subjects skip spec-TP and instead land in spec-CP,where they are assigned case by the matrix V.(23) Accusative subject in spec-CP[푁푃 [퐶푃 Sacc [푇푃 [푣푃 t푆 O V-v ] T ] C ] N ] V ...⊲ If CP is a phase, any object scrambling from a clause with an accusative subject will haveto pass through the CP edge, which in this context the subject also inhabits.⊲ Furthermore, following Richards (1997, 1999, a.o.), secondary specifiers formed bymove-ment to a given head should be required to tuck-in to a lower specifier of that head:(24) Predicted tucking-in below acc subject prior to further object scrambling... [푁푃 [퐶푃 Sacc O [푇푃 [푣푃 t푆 t푂OO ... ] T ] C ] N ] ]• If such a structure is in fact the input to attempted scrambling across an accusative subject,we correctly predict the unacceptability of such scrambling with one additional concept.• Specifically, if in a multiple specifier configuration the outer specifier must move before theinner one can be accessed, then we indeed expect scrambling of the object to fail here, sincethis would require illegally extracting the object from a lower specifier of CP:(25) No scrambling object from spec-CP below acc subject* [퐶푃 O ... [푉푃 [푁푃 [퐶푃 Sacc t푂OO [푇푃 [푣푃 t푆 t푂OO V-v ] T ] C ] N ] V ] C ]10F Importantly, this understanding accurately predicts that if the accusative subject is movedinto the matrix clause, then such movement of the object becomes possible as well:(26) Object scrambling fed by acc subject movementX [퐶푃 Sacc O .... [푉푃 [푁푃 [퐶푃 t푆OO t푂OO [푇푃 [푣푃 t푆 t푂OO V-v ] T ] C ] N ] V ] C ]• Thatmovement of an inner specifier requiresmovement of the one above it as well is predictedby at least two theories:⊲ Bošković (2016) argues that this is simply how phase impenetrability is calculated: if thereare two edge constituents, only the highest (modulo traces) is visible.⊲ The same prediction is made by the Cyclic Linearization theory (Fox & Pesetsky 2005,a.o.), for which movement of a lower specifier of a phase will also require movement ofany higher ones, in order to ensure that their linearization is kept consistent.3• The Cyclic Linearization account accurately predicts that only the final order S < O ispermitted here, as shown once more below:(27) Accusative subject scrambling feeds long object scrambling with S < O ordera. [Fatima-n1Fatima-gensabij-1-n]푘child-3-acctüneneyesterdayalma-n1 푗apple-accustazteacher.nom[Clause t푘 t 푗aša-Kan-1neat-nfut-3-acc] ešit-ti.hear-pst‘The teacher heard that Fatima’s child ate the apple yesterday.’b. *Alma-n1 푗apple-acctüneneyesterday[fatima-n1Fatima-gensabij-1-n]푘child-3-accustazteacher.nom[Clause t푘 t 푗aša-Kan-1neat-nfut-3-acc] ešit-ti.hear-pst‘The teacher heard that Fatima’s child ate the apple yesterday.’• Bošković (2016) will also predict this ordering fact if the two moving phrases necessarilytuck-in through all subsequent phases they pass through.⊲ We will not adjudicate between these two possibilities here, since both appear plausible.3See Davis (2020) for detail on why the Cyclic Linearization theory predicts that the order derived in a phase edge position will bepreserved by subsequent movement.114.3 Object scrambling in clauses with genitive subjects• Above we hypothesized that embedded clauses with genitive subjects include DP, to whosespecifier the subject moves for case-related reasons, as shown once more below:(28) Movement of gen subject to spec-DP[퐷푃 Sgen [푁푃 [푣푃 t푆 O V-v ] N ] D ] V ...• We have seen that the genitive subject can scramble into the matrix clause, which is unsur-prising if it occupies the edge of the DP phase prior to the application of any A′-movement.• But more surprising is the fact that, as section 2 showed, object scrambling from such anembedded clause is impossible whether the genitive subject scrambles out or not:(29) GEN subject movement never feeds long-distance object scramblinga. * [퐶푃 O .... [푉푃 [퐷푃 Sgen t푂OO [푁푃 [푣푃 t푆 t푂OO V-v ] N ] D ] V ] C ]b. * [퐶푃 Sgen O .... [푉푃 [퐷푃 t푆OO t푂OO [푁푃 [푣푃 t푆 t푂OO V-v ] N ] D ] V ] C ]c. * [퐶푃 O Sgen .... [푉푃 [퐷푃 t푆OO t푂OO [푁푃 [푣푃 t푆 t푂OO V-v ] N ] D ] V ] C ]F This fact will be accurately predicted if A′-extraction cannot pass through spec-DP. In thissituation, the behavior of the genitive subject is simply irrelevant.• While it is not abundantly clear why this should be so, a fewworks havemade this suggestion.⊲ Bosque & Gallego (2014) argue that extraction from Spanish DPs cannot occur, and thatwhen it appears to have, reanalysis is involved.⊲ Reeve (2018) argues that nominal phrases are phases that uniquely lack edges, and proposesthat apparent extraction from them involves base generation in a higher position.⊲ van Urk (2019) recently points out that while nominal phrases have many of the hallmarksof phase-hood, it remains unclear if there is solid evidence for successive-cyclic movementthrough them.4• For the purposes of this presentation, we will leave a solution for this unique property of theDP to future work.4Rackowski & Richards (2005) and several works following argue that extraction from a phase can bypass its edge if and only if thatphase is first agreed-with. If DP uniquely lacks an edge position for A′-extraction, then it could be the case that all extraction from DPrequires agreement in this way. If so, this would entail that part of why extraction from DP seems relatively constrained is because, unlikeextraction from CP or vP, it is contingent on the availability of an independent agreement process. Such agreement would apparently benull in many cases, for instance, in English sentences like Who did you see a picture of?.125 Extension: Locality and subject scrambling• Here we extend the above concepts to account for some additional properties of embeddedsubjects.• First, on binding: If the principles of binding theory are evaluated at the phase level (Char-navel & Sportiche 2016, Bošković 2016, a.o.), then an anaphoric subject should have toinhabit the edge of its local clause if it is to be bound by an antecedent in the matrix clause.⊲ Recall that our analysis in the previous section used the idea that accusative and genitivesubjects inhabit the edge of their embedded clause (respectively in CP and DP), whilenominative subjects remain in TP, below CP.• These concepts together accurately predict the fact that only accusative and genitive subjectscan be anaphors bound by an antecedent in the matrix clause:(30) Matrix subject can bind only acc/gen subject anaphor5Ustaz푘teacher.nom[Clause [kesi-kes-i-ni/n/*∅]푘self-self-3-gen/acc/*nomalmaappleaša-Kan-1-neat-nfut-3-acc] ešit-tihear-pst‘The teacher heard herself eating an apple.’⊲ This connection between binding and phase edges is also evident in English, in which ananaphor originating in an embedded clause must move to the clause edge to be bound bya phrase in the matrix clause (Nissenbaum 2000):(31) Binding into embedded clause must take advantage of clause edgea. *Mary1 said [퐶푃 that we should keep [this picture of herself1]].b. Mary1 said [퐶푃 [which picture of herself1]2 we should keep t2].c. *He2 knows [퐶푃 (that) [this picture of himself2] is probably the best].d. He3 knows [퐶푃 [which picture of himself3]4 t4 is probably the best].• Related fact: Nominative subjects in Balkar are frozen in place, and thus unlike accusativeand genitive ones, cannot move from the embedded clause:65Note that the unacceptability of the nominative anaphor here cannot be attributed to an anaphor agreement effect, since these threesubject types are all targeted for agreement by the embedded clause.6For some speakers, while scrambling of the nominative subject is usually illegal, it becomes licit if and only if the object alsoscrambles, provided that O < S word order holds. At the moment, we have only speculations about this interesting pattern.i a. *[Fatima-n1Fatima-gensabij-1]푘child-3.nomalma-n1 푗apple-acctüneneyesterdayustazteacher.nomt푘 t 푗 aša-Kan-1neat-nfut-3-accešit-ti.hear-pst‘The teacher heard that Fatima’s child ate the apple yesterday.’b. Alma-n1 푗apple-acctüneneyesterday[fatima-n1Fatima-gensabij-1]푘child-3.nomustazteacher.nomt푘 t 푗 aša-Kan-1neat-nfut-3-accešit-ti.hear-pst13(32) No scrambling of nom subject*[Fatima-n1Fatima-genbala-s1]푘child-3.nom(tünene)(yesterday)ustazteacher.nom[t푘 alma-n1apple-accaša-Kan-1-n]eat-nfut-3-accešt-gen-di.hear-nfut-3‘The teacher heard that Fatima’s child ate an apple (yesterday).7• The impossibility of both the binding of the nominative subject in (30), and its scramblingfrom the embedded clause in (32), would be predicted if there were an independent reasonto expect the unavailability of movement from spec-TP to spec-CP.⊲ If this movement is banned, the nominative subject cannot reach the edge of its local phaseand thus would not be accessible for dependencies relating to the higher clause.(33) To be ruled out: Movement from spec-TP to spec-CP*CPDP푁푂푀TPt푆. . .TC⊲ Movement of precisely this sort is ruled out by the formulation of anti-locality in Brillman& Hirsch (2016) and Erlewine (2016), who argue for a ban on movement from a specifierof a given phrase to one of the next highest phrase:8,9‘The teacher heard that Fatima’s child ate the apple yesterday.’(Also OK: Fatima’s child heard that the teacher ate the apple yesterday.)7This sentence is possible under an interpretation that does not involve scrambling: “Fatima’s child heard that the teacher ate anapple (yesterday).”8If such anti-locality is a real constraint, we might ask why many languages do allow cross-clausal scrambling of nominativesubjects. English is, of course, such a language. Brillman & Hirsch (2016) suggest following Doherty (1997) that embedded clauses inEnglish may lack a CP layer and thus involve extraction of subjects directly from spec-TP (yielding the that-trace effect):(i) Who1 does Bill think [푇 푃 t1 saw John ]?Another account consistent with the version of anti-locality used here comes from McCloskey (2000), who suggests based on facts aboutstranding in West Ulster English that subject extraction may proceed directly from spec-vP to spec-CP. We thus have two potential waysof understanding why languages like English permit the movement in (i). Correspondingly, we expect such derivations to be unavailablein Balkar (though see the footnote in 6 above for a puzzle which might be relevant on this note).9An alternative account of the ban on nominative subject scrambling might come from processing: perhaps pressure to parse thescrambled nominative subject as the subject of the matrix clause causes a garden path effect. Under such an analysis, it is not obviouswhy English speakers would not have comparable trouble with a sentence like the following:(i) John1, Mary thinks [ t1 likes this kind of food].Yet another alternative account would be that nominative subjects are not frozen, but rather, simply gain accusative case marking if they14(34) Spec-to-spec anti-locality*XPZP1 X YPt1 Y ...• In contrast, the fact that accusative and genitive subjects can both be bound by the matrixsubject, and can both scramble out of their local clause, are predicted if these subjects bypassspec-TP and instead A-move to the edge of their local phase, as argued above.5.1 Supporting evidence that nominative subjects are frozen5.1.1 QR• Baseline: in a simple clause, both direct and inverse scopes are available:(35) Ekitwoq1zgirlxareveryžaš-n1boy-acckör-gen-di-le.see-nfut-3-pl1. Two >every: ‘There were two girls such that they saw every boy.’2. Every >two: ‘For every boy, two (potentially different) girls saw him.’• When an embedded subject is a quantifier phrase, its case matters for possible scopeswith respect to the matrix subject.• When the embedded subject is genitive, both scopes are available; and that is independentof whether the genitive subject undergoes scrambling.(36) Ekitwoq1zgirlfatima-n1Fatima-genxareveryžaš-1-n1boy-3-genšaxar-Kacity-datbar-Kan-1-ngo-nfut-3-accešit-ti-le.hear-pst-pl1. Two >every: ‘There were two girls such that they heard that Fatima’s every boy wentto the city.’2. Every >two: ‘For Fatima’s every boy, there were two (potentially different) girls thatheard that he went to the city.’(37) Fatima-n1Fatima-genxareveryžaš-1-n1푘boy-3-genekitwoq1zgirlt푘 šaxar-Kacity-datbar-Kan-1-ngo-nfut-3-accešit-ti-le.hear-pst-pl1. Two >every: ‘There were two girls such that they heard that Fatima’s every boy wentto the city.’2. Every >two: ‘For Fatima’s every boy, there were two (potentially different) girls thatheard that he went to the city.’move through spec-CP in order to exit the clause. If nominative subjects are thus perfectly mobile modulo a morphological confound,then the facts in section 5.1, which shows that nominative subjects cannot undergo covert movement either, would be mysterious.15• The same pattern is observed for when the embedded subject is accusative: both scopesare available independent of scrambling.(38) Ekitwoq1zgirlfatima-n1Fatima-genxareveryžaš-1-nboy-3-accšaxar-Kacity-datbar-Kan-1-ngo-nfut-3-accešit-ti-le.hear-pst-pl1. Two >every: ‘There were two girls such that they heard that Fatima’s every boy wentto the city.’2. Every >two: ‘For Fatima’s every boy, there were two (potentially different) girls thatheard that he went to the city.’(39) Fatima-n1Fatima-genxareveryžaš-1-n푘boy-3-accekitwoq1zgirlt푘 šaxar-Kacity-datbar-Kan-1-ngo-nfut-3-accešit-ti-le.hear-pst-pl1. Two >every: ‘There were two girls such that they heard that Fatima’s every boy wentto the city.’2. Every >two: ‘For Fatima’s every boy, there were two (potentially different) girls thatheard that he went to the city.’• When the embedded subject is nominative, only the direct scope is available. Thecorrelation with overt scrambling holds: nominative subjects cannot be scrambled.(40) Ekitwoq1zgirlxareveryžašboy-3.nomšaxar-Kacity-datbar-Kan-1-ngo-nfut-3-accešit-ti-le.hear-pst-pl1. Two >every: ‘There were two girls such that they heard that every boy went to thecity.’2. *Every >two: *‘For every boy, there were two (potentially different) girls that heardthat he went to the city.’(41) *Xareveryžaš푘boy-3.nom(tünene)(yesterday)ekitwoq1zgirlt푘 šaxar-Kacity-datbar-Kan-1-ngo-nfut-3-accešit-ti-le.hear-pst-pl‘Two girls heard that every boy went to the city.’ (ungrammatical under both scopes)5.1.2 NPI licensing• Due to syncretism of genitive and accusative pronouns, here and in the next section wewill not be able to show genitive and accusative subjects separately. But will show that aform that is genitive/accusative has different properties compared to the nominative one.• Basic facts about NPIs:◦ Kiši-da (man-PTCL) is an NPI pronoun:◦ It cannot be used in upward entailing contexts;(42) a. *Kiši-daman-ptclalmaappleaša-Kan-d1.eat-nfut-3Exp.: ‘Someone ate an apple.’16b. *AlimAlimkiši-ni-daman-acc-ptclkör-gen-di.see-nfut-3Exp.: ‘Alim saw someone.’10◦ but it can be used when, e.g., negation is present.(43) a. Kiši-daman-ptclalmaappleaša-ma-Kan-d1.eat-neg-nfut-3‘Nobody ate an apple.’b. AlimAlimkiši-ni-daman-acc-ptclkör-me-gen-di.see-neg-nfut-3‘Alim didn’t see anyone.’• When the embedded subject is an NPI of a GEN/ACC form, it can get licensed either myembedded or by matrix negation:(44) a. Ustazteacherkiši-ni-daman-gen/acc-ptclalmaappleaša-ma-Kan-1-neat-neg-nfut-3-acckör-gen-di.see-nfut-3‘The teacher saw that no one ate an apple.’b. Ustazteacherkiši-ni-daman-gen/acc-ptclalmaappleaša-Kan-1-neat-nfut-3-acckör-me-gen-di.see-neg-nfut-3‘The teacher didn’t see of any x that x ate an apple.’• Howeverwhen an embedded subjectNPI isNOM, it can get licensed only by the embeddednegation:(45) a. Ustazteacherkiši-daman.nom-ptclalmaappleaša-ma-Kan-1-neat-neg-nfut-3-acckör-gen-di.see-nfut-3‘The teacher saw that no one ate an apple.’b. *Ustazteacherkiši-daman.nom-ptclalmaappleaša-Kan-1-neat-nfut-3-acckör-me-gen-di.see-neg-nfut-3Expected: ‘The teacher didn’t see of any x that x ate an apple.’5.1.3 Wide-scope indefinites• Kim ese da is a wide-scope indefinite: e.g., it normally takes high scope negation w.r.t.negation in a simple clause.• When this indefinite is an embedded subject of the nominalization, its scope with respectto matrix negation is determined by its case.10This sentence however is possible under a reading where kiši is interpreted literally, as meaning ‘man’: ‘Alim saw a man too.’17◦ When kim ese da is in the genitive/accusative form, it obligatorily takes wide scopewith respect to matrix negation.(46) Ustazteacherkim-ni-ese-dawho-gen/acc-ptcl-ptclalmaappleaša-Kan-1-neat-nfut-3-accešit-me-gen-di.hear-neg-nfut-3‘The teacher didn’t hear that someone ate an apple.’∃ >¬: ‘There exists someone about whom the teacher didn’t hear that theyate an apple.’¬ >∃: *‘The teacher didn’t hear that anyone ate an apple.’◦ When kim ese da is in the nominative form, it obligatorily takes narrow scope withrespect to matrix negation.(47) Ustazteacherkim-ese-dawho.nom-ptcl-ptclalmaappleaša-Kan-1-neat-nfut-3-accešit-me-gen-di.hear-neg-nfut-3‘The teacher didn’t hear that someone ate an apple.’∃ >¬: *‘There exists someone about whom the teacher didn’t hear that theyate an apple.’¬ >∃: ‘The teacher didn’t hear that anyone ate an apple.’6 Conclusion• To recap, in this presentation we have argued that Balkar reveals the following principlesabout movement:⊲ #1: That CP is a phase which allows multiple specifiers provided that tucking-in applies(Richards 1997, 1999).⊲ #2: That the highest phrase in a multiple specifier structure is privileged, such that a higherspecifier must move before a lower one can be accessed. (Bošković 2016, a.o.).⊲ #3: That DP is a phase which (at least in this context) does not allowA′-movement throughits edge (Bosque & Gallego 2014, Reeve 2018, van Urk 2019).• These proposals stem from an understanding whereby accusative and genitive subjects, butnot nominative ones, inhabit the edge of their local nominalized clause.• We also related this proposal to asymmetries in the bind-ability and mobility of subjects,which we argued additionally supplied evidence for the influence of anti-locality.• A puzzle: We predict that any variety of non-subject scrambling should in principle behaveexactly the same as object scrambling as described here. Our data on this is incomplete, butsome speakers indeed fit this prediction, while others showed more variability.187 Appendix A: Adverbs as a diagnostic for clause-internalscrambling• Temporal adverbs cannot be used as a diagnostic because of nominalizations with genitivesubjects: they cannot have adverbial modification that contradicts adverbial modificationof the matrix clause.• Lower adverbs can be used, because they are present in all the three nominalizations.• (48) shows that adverbs like ‘loudly’ (lit. ‘while causing the making of noise’) can occurat the edge of the embedded clause, but cannot scramble outside the embedded clause.(48) Long-distant adjunct scramblinga. ustazteacher.nombala-s1child-3.nomtauušnoiseet-dir-ipmake-caus-convalma-n1apple-accaša-Kan-1-neat-nfut-3-accešit-tihear-pst‘The teacher heard that her child ate the apple loudly.’b. ustazteacher.nomtauušnoiseet-dir-ipmake-caus-convbala-s1child-3.nomalma-n1apple-accaša-Kan-1-neat-nfut-3-accešit-tihear-pst‘The teacher heard that her child ate the apple loudly.’c. *tauušnoiseet-dir-ipmake-caus-convustazteacher.nombala-s1child-3.nomalma-n1apple-accaša-Kan-1-neat-nfut-3-accešit-tihear-pstExpected: ‘The teacher heard that her child ate the apple loudly.’(But possible under the interpretation where the teacher is the one making noise.)⇒ These indicates that these adverbs can be used as a marker of the embedded clauseboundary.8 Appendix B: Possessors of embedded subjects can move• Scrambling of the possessor of the NMN subject is possible irrespective of the case ofthe NMN’s subject.(49) Scrambling of the possessor of the nmn subject19fatima-n1푘Fatima-gentüneneyesterdayustazteacher[t푘 sabij-ichild-3.nom/sabij-i-ni/child-3-gen/sabij-i-n]/child-3-accalma-s1-napple-3-accaša-Kan-1neat-nfut-3-accešit-tihear-pst‘The teacher heard that Fatima’s child ate his apple yesterday.’• The same movement is possible from regular NPs as well: possessors of GEN possessorsof NPs can scramble out, (50).(50) Scrambling of a possessor of an NP’s possessorfatima-n1푘Fatima-gentüneneyesterdayasiatAsiat.nomt푘 sabij-i-nichild-3-gentat1ulutastyalma-s1-napple-3-accaša-d1eat-pst‘Asiat ate Fatima’s child’s tasty apple yesterday.’• Ability of possessors of the nmn’s subject to scramble correlates with them being able toQR and take wide scope with respect to the matrix subject, (51)-(52).(51) Baselineekitwoq1zgirl.nomxareveryoquc-u-nustudent-3-gennöger-ler-i-nfriend-pl-3-acckör-gen-di-lesee-nfut-3-pl‘Two girls saw a friend of every student.’1. 2 >∀: There were two girls such that they saw a friend of every student.2. ∀ > 2: For every student, there were two girls who saw a friend of theirs.(52) Targetekitwoq1zgirl.nomxareveryoquc-u-nustudent-3-gennöger-ifriend-3.nom/nöger-i-ni/friend-3-gen/nöger-i-n/friend-3-accalmaappleaša-Kan-1-neat-nfut-3-accešit-tihear-pst‘Two girls heard that a friend of every student ate an apple.’1. 2 >∀: There were two girls such that they heard that a friend of every studentate an apple.2. ∀ > 2: For every student푘 , there were two girls 푗 such that they 푗 heard thattheir푘 friend ate an apple.• Moreover, possessors of all three nominalizations can be anaphors bound by the matrixsubject, (53).1111Unfortunately, NPI and wide-scope indefinites could not be tested: making possessors out of those pronominal items has failed.20(53) BindingkerimKerim.nomkesiselfkes-i-niself-3-genbala-s1child-3.nom/bala-s1-n1/child-3-gen/bala-s1-n/child-3-acckitabbookoqu-Kan-1-nread-nfut-3-accešit-gen-dihear-nfut-3‘Kerim푘 heard that his푘 child read a book.’• All these properties indicate that possessors of embedded subjects can occupy the edgeof the nominalization and get out of it.⇒ The fact that possessors of nom subjects can get to the edge suggests that they, unlike thenominative subjects themselves, are not subject to the antilocality restrictions.⇒ Perhaps this is so because movement from within the nominative subject, given that ittakes off from a more deeply embedded position, is effectively ‘longer’ than movementof the subject itself.9 Appendix C: Similiar patterns in Turkish & Buryat• Patterns similiar to Balkar are also observed in other languages.• In Turkish scrambling of the object is possible from CPs with nominative subjects, butnot from the ones with accusative subjects.(54) Nominative Subject (CP)a. AhmetAhmet.nom[AliAli.nomkitab-ıbook-accoku-du]read-pstbiliyorknows‘Ahmet believes that Ali read the book.’b. AhmetAhmet.nom[kitab-ı푘book-accAliAli.nomt푘 oku-du]read-pstbiliyorknows‘Ahmet believes that Ali read the book.’c. kitab-ı푘book-accAhmetAhmet.nom[AliAli.nomt푘 oku-du]read-pstbiliyorknows‘Ahmet believes that Ali read the book.’(55) Accusative Subject (CP)a. AhmetAhmet.nom[Ali-yiAli-acckitab-ıbook-accoku-du]read-pstbiliyorknows‘Ahmet believes that Ali read the book.’b. *AhmetAhmet.nom[kitab-ı푘book-accAli-yiAli-acct푘 oku-du]read-pstbiliyorknows‘Ahmet believes that Ali read the book.’21c. *kitab-ı푘book-accAhmetAhmet.nom[Ali-yiAli-acct푘 oku-du]read-pstbiliyorknows‘Ahmet believes that Ali read the book.’• In Buryat (Mongolic), object scrambling out of finite CPs is possible only if the embeddedsubject is nominative (but not accusative).(56) Nominative Subject (CP)a. badmaBadma.nom[sajanaSajana.nomtum9n-ij9Tumen-accxara-xasee-futg9ž9]comphan-a:think-pst‘Badma thought that Sajana will see Tumen.’b. badmaBadma.nom[tum9n-ij9푘Tumen-accsajanaSajana.nomt푘 xara-xasee-futg9ž9]comphan-a:think-pst‘Badma thought that Sajana will see Tumen.’c. tum9n-ij9푘Tumen-accbadmaBadma.nom[sajanaSajana.nomt푘 xara-xasee-futg9ž9]comphan-a:think-pst‘Badma thought that Sajana will see Tumen.’(57) Accusative Subject (CP)a. badmaBadma.nom[sajan-i:j9Sajana-acctum9n-ij9Tumen-accxara-xasee-futg9ž9]comphan-a:think-pst‘Badma thought that Sajana will see Tumen.’b. *badmaBadma.nom[tum9n-ij9푘Tumen-accsajan-i:j9Sajana-acct푘 xara-xasee-futg9ž9]comphan-a:think-pst‘Badma thought that Sajana will see Tumen.’12c. *tum9n-ij9푘Tumen-accbadmaBadma.nom[sajan-i:j9Sajana-acct푘 xara-xasee-futg9ž9]comphan-a:think-pst‘Badma thought that Sajana will see Tumen.’13• Similar restrictions hold for nominalized clauses: objects cannot scramble over accusativesubjects, which are on the edge of the embedded clause, but can scramble over genitivesubjects, which occupy position lower than the edge (Bondarenko 2017).(58) Genitive Subject (nmn)a. badmaBadma.nom[sajan-i:nSajana-gentum9n-ij9Tumen-accxar-a:S-i:j9]see-part-acchan-a:think-pst‘Badma remembered that Sajana saw Tumen.’b. badmaBadma.nom[tum9n-ij9푘Tumen-accsajan-i:nSajana-gent푘 xar-a:S-i:j9]see-part-acchan-a:think-pst‘Badma remembered that Sajana saw Tumen.’12This sentence is grammatical under a different reading: Badma thought that Tumen will see Sajana.13This sentence is grammatical under a different reading: Badma thought that Tumen will see Sajana.22c. tum9n-ij9푘Tumen-accbadmaBadma.nom[sajan-i:nSajana-gent푘 xar-a:S-i:j9]see-part-acchan-a:think-pst‘Badma remembered that Sajana saw Tumen.’(59) Accusative Subject (nmn)a. badmaBadma.nom[sajan-i:j9Sajana-acctum9n-ij9Tumen-accxar-a:S-i:j9]see-part-acchan-a:think-pst‘Badma remembered that Sajana saw Tumen.’b. *badmaBadma.nom[tum9n-ij9푘Tumen-accsajan-i:j9Sajana-acct푘 xar-a:S-i:j9]see-part-acchan-a:think-pst‘Badma remembered that Sajana saw Tumen.’c. *tum9n-ij9푘Tumen-accbadmaBadma.nom[sajan-i:j9Sajana-acct푘 xar-a:S-i:j9]see-part-acchan-a:think-pst‘Badma remembered that Sajana saw Tumen.’10 ReferencesBondarenko, Tatiana. 2017. ECM in Buryat and the optionality of movement. In Proceedingsof WAFL 12, MIT Working Papers in Linguistics, #83, P. 31-42.Bondarenko, Tatiana. 2018. Subject marking and scrambling effects in Balkar nominalizations.In Faruk Akkuş, İsa Kerem Bayırlı, Deniz Özyıldız (eds.) Proceedings of the first workshopon Turkish, Turkic, and the languages of Turkey (Tu+1), Graduate Linguistics StudentAssociation, University of Massachusetts. P. 27-42.Bosque, Ignacio, and Ángel J. Gallego. 2014. Reconsidering subextraction: Evidence fromSpanish. Borealis 3(2): 223–258.Brillman, Ruth & Aron Hirsch. 2016. An anti-locality account of English subject/non-subjectasymmetries. Proceedings of CLS 50.Boškovic`, Željko. 2016. Getting really edgy: On the edge of the edge. Linguistic Inquiry 45.Charnavel, Isabelle, and Dominique Sportiche. 2016. Anaphor Binding: What FrenchInanimate Anaphors Show. Linguistic Inquiry 47, 35–87. doi:10.1162/ling푎00204.Chomsky, Noam. 2000. Minimalist Inquiries. In Roger Martin, David Michales, JuanUrigareka & Samuel Jay Keyser (eds.), Step by Step: Essays on Minimalist Syntax in Honorof Howard Lasnik, 89–156. Cambridge, MA: MIT Press.Chomsky, Noam. 2001. Derivation by Phase. In Michael Kenstowicz (ed.), Ken Hale: A lifein language, Cambridge, MA: MIT Press.Chomsky, Noam. 2008. On Phases. In Robert Freidin, Carlos P Otero & Maria LuisaZubizarreta (eds.), Foundational issues in linguistic theory: essays in honor of Jean-RogerVergnaud, Cambridge, MA: MIT Press.23Davis, Colin. 2020. Crossing and stranding at edges: On intermediate stranding and phasetheory. Glossa, 5(1), 17. DOI: http://doi.org/10.5334/gjgl.854Doherty, Cathal. 1997. Clauses without complementizers: Finite IP complementation inEnglish. The Linguistic Review 14, 197–220.Erlewine, Michael Yoshitaka, 2016. Anti-locality and optimality in Kaqchikel Agent Focus.Natural Language & Linguistic Theory 34, 429–479.Fox, Danny and David Pesetsky. 2005. Cyclic Linearization of Syntactic Structure. TheoreticalLinguistics 31, 1-45.McCloskey, James. 2000. Quantifier Float and Wh-movement in an Irish English. LinguisticInquiry 51.Nissenbaum, Jonathan. 2000. Investigations of covert phrase movement. PhD Dissertation,MIT.Rackowski, Andrea and Norvin Richards. 2005. Phase Edge and Extraction: A Tagalog CaseStudy. Linguistic Inquiry 36, 565-599.Reeve, Matthew. 2018. An agreement-based analysis of extraction from nominals. NaturalLanguage and Linguistic Theory.Richards, Norvin. 1997. What moves where in which language? PhD Dissertation, MIT.Richards, Norvin. 1999. Featural cyclicity and ordering of multiple specifiers. In Samuel D.Epstein & Norbert Hornstein (eds.), Working Minimalism, Cambridge, MA.van Urk, Coppe. 2019. A taxonomy of successive cyclicity effects. Unpublished manuscript.<http://cvanurk.sllf.qmul.ac.uk/successivecyclicityacrossdomains.pdf>24

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.73804.1-0389838/manifest

Comment

Related Items