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

Leftover agreement : spelling out Kartvelian number Bondarenko, Tanya; Stanislao, Zompì 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


73804-Bondarenko_T_et_al_Leftover_WCCFL38_2020_poster.pdf [ 130.83kB ]
JSON: 73804-1.0389839.json
JSON-LD: 73804-1.0389839-ld.json
RDF/XML (Pretty): 73804-1.0389839-rdf.xml
RDF/JSON: 73804-1.0389839-rdf.json
Turtle: 73804-1.0389839-turtle.txt
N-Triples: 73804-1.0389839-rdf-ntriples.txt
Original Record: 73804-1.0389839-source.json
Full Text

Full Text

Leftover Agreement: Spelling out Kartvelian numberTanya Bondarenko (tbond@mit.edu) & Stanislao Zompì (zompi@mit.edu) |Massachusetts Institute of TechnologyWCCFL 38, The University of British Columbia1 INTRODUCTIONStandard DM (Bobaljik 2000): VocabularyInsertion (VI) is fully replacive: all featuresof the head are replaced by the exponent.This poster: VI is only partially replacive: the expo-nent replaces only those features of the head that itsspecification matches.Evidence: unexponed features of v-probes get agreed with by higher #-probes in Kartvelian languages.4 ACCOUNTING FOR KARTVELIAN NUMBER AGREEMENT WITH LAv-agreement (following Béjar & Rezac 2009 — simplified): v agrees with participant NPs in φ-features(gv-/gw-/n-⇔ {1PL}, m-⇔ {1}, ž-/g-/r-⇔ {2}); it first searches in its complement, then in its specifier.T-agreement: T agrees with the highest non-oblique NP: the subject in (2)-(5).(8) a. Laz & Megrelian#PTPvPNP3SGv’VPNP1PLVvϕ: {1PL}⇔ m-Tϕ: {3SG}#PL ⇔ -a(n)b. Georgian & Svan#PTPvPNP3SGv’VPNP1PLVvϕ: {1PL}⇔ gv-/gw-/n-Tϕ: {3SG}#failed#-agreement: {3SG subject, 1PL object} (8)• Laz/Megrelian: the v-exponent m- only expones a subset of v’s bundle: only {1}. The unexponedPL is left over and remains visible to the #-probe. The #-probe copies PL and then expones it. ⇒ LA.• Georgian/Svan have v-exponents that are specified for the whole bundle {1st, PL}. Thus, afterSpell-Out, there are no leftover features on v that the # probe could agree with. ⇒ no LA.Why can’t the #-probe get PL from the object NP itself? We crucially assume Chomsky’s (2001) Weak PIC:the phase head # can access the lower phase head v and its specifier, but not anything in v’s complement.#-agreement: {3SG subject, 2PL object} — LA for all languages• None of the four languages have an exponent specified for the whole bundle {2st, PL}, so theunexponed PL feature copied from the 2PL object is left over and gets agreed with by the #-probe.2 THE KARTVELIAN DISCONTINUOUS-BLEEDING GENERALIZATION• Kartvelian verbs have three agreement slots; we assume the first slot (g- in (1)) corresponds to aprobe on v, the second one (-da) to a T-probe, and the third one (-t) to a higher, phasal #-probe.(1) (is)(3SG.NOM)(tkven)(2PL.ACC)gada-gv -c’er-daT-t#PVB-2-describe-COND.3SG-PLGeorgian‘(S)he would describe you (pl).’• Suffixal PL-agreement with an NP in Kartvelian shows up only if there is no v-prefix that canexpone PL-agreement with that NP — an instance of discontinuous bleeding (Noyer 1992).(2) Georgian (Aronson 1990: 172)gv-naxa /g-naxa-t1PL-see.AOR.3SG /2-see.AOR.3SG-PL‘(S)he saw us / you (PL).’(3) Svan (Testelets 1989: 9)gw/n-adgäri /ž-adgäri-x1PL.IN/EX-kill.PRS /2-kill.PRS-PL‘(S)he is killing us / you (PL).’(4) Laz (Lacroix 2009: 294)m-dziom-an /g-dziom-an1-see.PRS-PL /2-see.PRS-PL‘(S)he sees us / you (PL).’(5) Megrelian (Kipshidze 1914: 076)m-tS’ar@n-a(n) /r-tS’ar@n-a(n)1-write.PRS-PL /2-write.PRS-PL‘(S)he writes us / you (PL).’• In Georgian and Svan there is a 1PL prefix, and in those forms that have it, the #-suffix is bled by itspresence — unlike in Megrelian and Laz, which lack a 1PL prefix.• No language has a 2PL prefix, so all of them have suffixal PL-agreement with 2PL NPs.• The question: why does the suffix’s presence depend on the features exponed by the prefix?3 PROPOSAL: LEFTOVER AGREEMENT (LA)Leftover Agreement is agreement of a higher probe X with unexponed features on a lower probe Y.(6) a. XPYPYϕ: {F1:val1 , F2: val2 }⇔ /α/: {F1: val1}XF2:___b. XPYPYϕ: {F1:val1 , F2: val2 }⇔ /α/: {F1: val1}XF2: val2(7) Let Y be a probe that has copied a feature bundle ϕ through agreement with noun phrases in itsc-command domain, and let X be a higher probe, the head of the phase XP.a. X, being a phase head, triggers the Spell-Out of its complement, and the head Y thus getsmatched with the best exponent available, /α/. It turns out that this exponent is specified foronly a subset of the feature bundle ϕ. (E.g., in (6), /α/ is specified for F1, but not F2.)b. The features of Y that /α/ is not specified for — the leftovers, like F2 in (6) — remain accessiblefor further computation, and can get agreed with by X. This is Leftover Agreement (LA).5 KEY EVIDENCE• Previous accounts captured the pattern in (2)-(5) by morphological means (Halle & Marantz 1993;McGinnis 2008; Lomashvili & Harley 2011; Blix 2016; Thivierge 2019).Instead, we view the number suffix as a syntactically distinct probe (cf. Foley 2017), and thereforepredict LA to be subject to intervention effects and locality conditions. This is borne out.• Svan exhibits intervention effects: LA with object features on v is blocked by participant subjects.(9) ž-adgäri2-kill.PRS//ž-adgäri-x2-kill.PRS-PL(Testelets 1989: 9)‘I am killing you (PL) / (s)he is killing you (PL).’This is unexpected on morphological accounts, but follows if Svan’s highest probe is relativized toPL or PART, so that participant subjects — even if singular — can halt its search.• Evidence for sensitivity to locality comes from #-agreement with 3PL objects in Georgian.Such agreement is normally impossible: v does not agree with 3rd-person NPs, so no LA is possiblewith them, and 3PL objects that are inside vP are not directly accessible to # due to the Weak PIC.However, 3PL objects can be agreed with directly by # if moved out of vP:(10) [OBJ mesamethirdseri-isseries-GENnakt’v-eb-s]kform-PL-DAT[vP [SUBJ saertocommonpunkcia]function.NOMtk a-ertianeb]-tPFV-unite-PL‘A common function unites the forms of the 3rd series.’ (via Blix, forthcoming)


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items