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

On the seemingly indefinite readings of Kannada bare singulars Srinivas, Sadhwi; Rawlins, Kyle 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-Srinivas_S_et_al_Kannada_singulars_WCCFL38_2020_handout.pdf [ 207kB ]
JSON: 73804-1.0389899.json
JSON-LD: 73804-1.0389899-ld.json
RDF/XML (Pretty): 73804-1.0389899-rdf.xml
RDF/JSON: 73804-1.0389899-rdf.json
Turtle: 73804-1.0389899-turtle.txt
N-Triples: 73804-1.0389899-rdf-ntriples.txt
Original Record: 73804-1.0389899-source.json
Full Text

Full Text

March 8, 2020On the seemingly indefinite readings of Kannada baresingularsSadhwi Srinivas and Kyle Rawlins{sadhwi@jhu.edu, kgr@jhu.edu}Johns Hopkins University, Cognitive Science DepartmentWCCFL 2020, UBC1 Overview• Bare singulars (more generally, bare nominals) in ‘determinerless’ languagescommonly occur in definite, indefinite and kind-denoting contexts (Chier-chia, 1998; Dayal, 2004; Jenks, 2018, 2015; Ahn, 2019; Deal and Nee, 2018).• Two types of prominent analyses in the literature:– One views them as ambiguous between these readings (Gerstner andKrifka, 1987; Wilkinson, 1991; Diesing, 1992).– Another views them as ambiguous only between the definite and kindreadings, with indefinite readings explained as arising via the kind read-ing (Carlson, 1977; Chierchia, 1998; Dayal, 2004).• Kannada (Dravidian; mainly spoken in India with about 44 million speakers):all 3 readings are descriptively compatible with the bare singular.• A partial analysis by Lidz (2006) suggests that the Kannada bare singulardirect objects are ambiguous between definite and indefinite readings.• However: we claim, following Dayal’s (1992; 1999; 2004; 2011) account ofHindi, that Kannada bare singulars only have definite and kind readings butnot true indefinite readings.– They do not allow (non-trivial) wide- or intermediate-scoped indefinitereadings.– Examples that Lidz (2006) analyzed as wide-scoped/specific indefinitereadings in fact involve hearer-old definite readings.• To explain the apparent non-specific readings of the bare singulars:– We first consider three existing proposals in the literature—two by Dayal(2004) for Hindi data, one by Aguilar Guevara (2014) for related defi-nite data in English.– None of these are entirely satisfactory, so we then propose an alternativesolution that uses Predicate Restriction (Chung and Ladusaw, 2003).2 Core data: Kannada bare singular2.1 Definite usesExamples (1)-(2) show that bare singulars in Kannada have descriptively definitereadings in (larger or immediate situation) uniqueness contexts (Russell, 2005;Roberts, 2003; Evans, 1977; Barker, 2004; Lo¨bner, 1985; Kadmon, 1990), as wellas in anaphoric contexts (Christophersen, 1939; Heim, 1982; Kamp, 1981).In both cases, the bare noun picks out a uniquely identifiable referent:(1) SuryaSuniDiialldinadayhorageeoutside.EMPHbandillacome-PST-NEG“The sun did not come out all day.”(2) NenneYesterdaynaanuI.NOMonduonecchandadaprettyungurajringkonDukonDe.boughtIvattutodayungurajringelloosomewherekaLedulosthoytu.went“Yesterday I bought a pretty ring. Today, the ring has gotten lost somewhere.”They can also felicitously occur in “donkey sentences” like in (3).(3) Katte(y)uLLuvadonkey.havingpratiyobbaeveryraitafarmerkatte-gedonkey-DATooTafoodhaakuttaanegives“Every farmer who has a donkey feeds the donkey.”(4) further establishes their status as definites, distinguishing the bare singulars fromdemonstratives & indefinites (Dayal 2004).(4) #(Aa/Ondu)(That/One)naayidogmalkonDidesleeping.COPmatteand#(aa/ondu)(that/one)naayidogbogaLtabarkingide.COP“That/One/#The dog is sleeping and that/one/#the dog is barking.”2.2 Indefinite(-like) usesIn (5)-(6), the available reading of the bare singular descriptively looks like a non-specific indefinite.(5) Room-alliRoom-LOCilimouseideCOP-PRES.“There is a mouse in the room”(6) a. (Discourse initial) There is something wrong with today. The sun isn’tshining, the flowers aren’t blooming, and...b. HoragaDe,OutsidehakkibirdjooraagiloudlyhaaDtasingingilla.not“It is not the case that there is a bird singing loudly outside.”Page 1Sadhwi Srinivas and Kyle Rawlins (Johns Hopkins University)On the seemingly indefinite readings of Kannada bare singulars March 8, 2020In examples like (7), a definite reading fails to arise despite the presence of a poten-tial antecedent. The bare nominal nonetheless gets what appears to be a non-specificindefinite reading.(7) RoomgaLalliOf.the.roomsonduonemaatraonlytumbaverychennagittu.good.was1920-alli1920-LOConduoneraatrinightNehruNehruroom-alliroom-LOCmalagiddasleptanteit.seems“One of the rooms was especially nice. In 1920, Nehru supposedly slept anight in #the room/ some room.”Two things to note:• Examples like (7) are not descriptively incompatible with an alternative defi-nite analysis – as long as there are appropriate constraints on the bare singu-lar’s definite interpretation.– We believe that whether or not the definite reading can arise dependson uniqueness considerations within suitable referential domains. Aprecise characterization of which domains count as suitable is beyondthe scope of the current talk.• Second, notice that we haven’t shown any (non-trivial) wide-scoped indefi-nite examples above. This is because (we claim) no such readings exist—apoint taken up in more detail in §4.2.3 Generic usesThe bare singular can appear with kind-level predicates like common:(8) NaayiDogonduonesaamanya-(v)aadacommon-COPpraaNianimal“The dog is a common animal.”With object-level predicates like (9)-(10), imperfective aspect or individual-levelpredicates can lead to generic reading.(9) NaayiDogbogaLatte.barks“Dogs bark.”(10) Naayi-geDog-DATbekk-inacat-GENjotewithjagaLafightaaDuvuduplayingishTa.like“Dogs like to fight with cats.”2.4 Predicative usesThe bare singular in Kannada may also occur in predicative sentences like (11).(11) MotiMotinaayidogaadroobeing.stillbekkinacat-GENhaagelikeaaDatte.behaves(Tanna.paaDi-ge(its.own.self-DATtaanuitkootiratte.)sits)“In spite of being a dog, Moti acts like a cat. (It keeps to itself.)”3 Related previous discussion: Lidz 2006Lidz (2006) includes one of the few existing theoretical discussions of bare singulardirect objects in Kannada.• Key starting assumption: bare singulars have indefinite readings. Builds uponthis assumption to argue that the morphology associated with these itemsand their syntactic positions separately determine if they are specific or non-specific (indefinites).• When (inanimate)1 bare singulars appear as direct objects without case-markingas in (12), Lidz argues that they can be specific or non-specific indefinites.(12) RashmiRashmikathepustakastory.booktarlilladid.not.bring“Rashmi did not bring a story book.”• But when the bare singular is case-marked as in (13), only specific readingsare said to be allowed.(13) RashmiRashmikathepustaka-(v)annastory.book-ACCtarlilladid.not.bring“There was a storybook that Rashmi did not bring.”• Similarly, higher syntactic position of the bare singular also forces specificindefinite readings.(14) NaanuIchennaagiwellpustakabookoodideread“I read a book well.” (non-specific book)(15) NaanuIpustakabookchennaagiwelloodideread“I read a book well.” (specific book)1ACC-marking is said to be obligatory with animate bare singulars, and does not contribute todisambiguating specific from non-specific readings in such nouns. But see footnote 2 for clarification.Page 2Sadhwi Srinivas and Kyle Rawlins (Johns Hopkins University)On the seemingly indefinite readings of Kannada bare singulars March 8, 2020Our main point of conflict with Lidz (2006), which we develop further below, isdue to our claim that the so-called specific indefinite readings of the bare singularare in fact hearer-old, scopally-inert definite readings.4 Proposed analysis for the bare singular4.1 Gist• The Kannada bare singular is ambiguous between definite, kind and non-specific (narrow-scoped) ‘indefinite readings’, but does not have specific (wide-scoped) indefinite readings.• The bare singular has a basic predicative type 〈e,t〉, and definite/kind readingsarise due to Partee-style type-shifts (Partee 1986).• Apparent ‘non-specific’ readings in non ACC-marked bare singular objects inatelic sentences arise not through indefiniteness, but through pseudo-incorporation(Dayal, 2011).• The apparent ‘non-specific’ readings in other positions are explained notby indefiniteness, but by appealing to composition via Predicate Restriction(Chung and Ladusaw, 2003).4.2 New data: restrictions on ‘specific’ readingsWe claim that the so-called specific readings in (12)-(13) and (15) are not in fact dueto a wide-scoped existential quantifier, but instead instances of definiteness wherethe referent is unique & hearer-old.• Severe restrictions on scope that don’t match specific indefinite behavior.• What ‘specific’ readings there are better match behavior of definites.Case 1: intermediate scope. What Lidz (p. 13) described as specific indefinitereadings (with intermediate scope), as in (16), are better accounted for by bridgingdefinite descriptions.(16) PratiyobbaEveryvidyarthistudentpustakabookood-al-illaread-INF-NEGNOT: ‘For every student, there was a book the student did not read’BUT: ‘Every student did not read the book (that they had been assigned)’• In the absence of prior knowledge that every student had been assigned aspecific book, (16) only has a reading where no student read any book or onewhere no student read a specific, hearer-old book.Case 2: scope with respect to negation. (17) illustrates that a specific indefinitereading under negation cannot be obtained without the nominal modifier ondu.(17) RashmiRashmiheccukammimore.or.lessellaallpustakabookskonDkonDlu,boughtaadrebut#(ondu)(one)pustakabookkonDkoLLalilla.did.not.buyWithout ondu: #‘Rashmi bought almost all the books, but she didn’t buya(ny) book/the book’With ondu: ‘Rashmi bought almost all the books, but there was a book shedid not buy’Case 3: referential properties in discourse. Bare singulars also do not showproperties generally associated with ‘true’ indefinites: namely, they are infelicitouswithout explicitly or implicitly established discourse referents in (18-b), and do notpattern like indefinites in (19). (Data after Dayal on Hindi)(18) a. (Discourse-initial) How was the party?b. Party-alliParty-LOCtumbamanyjanar-iddaru.people-COP#(obba)(one)huDugaboy#(ondu)(one)huDugigirljotewithdancedancemaaDtidda.doing.COP‘There were many people at the party. #The boy was dancing with thegirl.’(19) *(Aa/Ondu)(That/One)naayidogmalkonDidesleeping.COPmatteand*(aa/ondu)(that/one)naayidogbogaLtide.barking.COP‘That/One/*The dog is sleeping and that/one/*the dog is barking.’4.3 Explaining the apparent non-specific indefinite readings4.3.1 Pseudo-incorporation in non-case marked direct objects• When the bare singular is a non-ACC marked object, the non-specific read-ing is (often) due to pseudo-incorporation of the kind reading with the verb(‘pseudo’ because the N and V do not form a syntactic unit).• Strong evidence for such incorporation comes from the number neutrality ofthe bare singular in atelic contexts (Dayal, 2011, 2015), as seen in (20).(20) NaanuI.NOMbaTTeclothesogiitidiiniwashing‘I am washing clothes’ (many different clothes)• In spite of the singular (null) morphology on “book” in (21), it can occur withplural predicates like ‘gather’/‘arrange’.(21) AmmaMotherAnu-geAnu-DATsariyaagiproperlypustakabookjoDsakkearrangeheLidaLehas.asked“Mom has asked Anu to arrange books properly.”Page 3Sadhwi Srinivas and Kyle Rawlins (Johns Hopkins University)On the seemingly indefinite readings of Kannada bare singulars March 8, 2020• Upon combining with atelic predicates, it can denote a plurality of events.(22) is understood as everyone going to do some shoe-buying, so each personbuys shoes for themselves.(22) NaavellaruAll.of.uschaplislipperskonDkoLakketo.buyhogtagoingidiiviare“We are all going to buy slippers.”• It is also awkward to try to refer back to the bare singular with a pronounwhen it is not case-marked, as is expected for incorporated nouns.(23) AnuAnupustakabookoodtidaLe.is.reading??AduItavaL-igeshe-DATtumbaveryishTa.like“Anu is reading a book. She likes it a lot.”• Such incorporation is more productive with inanimate nominals, but also oc-curs in some animates (e.g., magu noDikoLLu; “watch a child”) Elsewhere• In the case-marked nominal in (24) that lacks a number-neutral reading, non-specificity cannot be due to pseudo-incorporation.(24) NaanuI.NOMbaTTe-annacloth-ACCogiitidiiniwashing‘I am washing a (single) cloth.’• Non-specific readings can also arise in bare singular subjects, e.g. (25).(25) HoragaDe,OutsidenaayidogbogaLtittu.was.barking‘Outside, a dog was barking.’• An adequate analysis for Kannada bare nominals must be able to account forsuch non-incorporated yet non-specific readings.• Below, we discuss three potential accounts for explaining these readings,each with its own set of significant problems.• We then present an alternative proposal that uses Predicate Restriction.(I) Dayal (2004) #1 (Dayal 1992, 1999)Basic idea: the non-specific existential reading is derived from the kind readingusing Derived Kind Predication (DKP):2This shows, contra Lidz 2006, that some animate bare singular objects can appear without ACC-marking. However, these are not very productive. “Name-worthiness” (Dayal, 2015) seems to be animportant requirement for this to be allowed—i.e., the bare noun + predicate combination must denotea “well-established activity”.(26) DKP: If P applies to objects and k denotes a kind,then P(k) = ∃x[∪k(x) ∧ P(x)] (Dayal 2004 10a)• The ∪ operation takes kinds and returns their instantiation sets in a givensituation, and the existential in DKP necessarily takes narrow scope.• The assumption here is that kind terms have access to their instantiation sets.• While there are no size restrictions on the instantiation set associated withbare plurals, the instantiation set associated with a bare singular must be re-stricted to a singleton set.• However, Dayal (2004) ultimately does not adopt this analysis, because ofentity-level apparently existential readings (see below).• Assumption that the bare singular kind term in languages like Hindi must beanalyzed similarly to singular definite generics in English. However, Englishsingular definite generics do not seem to have access to their instantiations,from reasoning as below:– Singular kinds are more “group-like” than “sum-like” (the bare pluralis “sum like”).– The relation between sums and their atoms is semantically transparentto predication, but the relation b/w groups and their members is opaque.• This is why the singular definite generic in English is unacceptable in episodiccontexts like in (27), but is possible with bare plurals as in (28).(27) The dog is barking.(Intended: A dog is barking).(28) Dogs are barking.Takeaway: some, but not all, ‘indefinite’ bare nominal readings may be explainableunder a kind-based analysis.(II) Dayal (2004) #2Basic idea: indefinite readings of bare singular subjects in Hindi, e.g. in (29) whichconveys presence of a hearer-new mouse, arise by an IOTA-like type-shift, wherethe type-shifter presupposes uniqueness but not necessarily familiarity.(29) kamreroommeinincuuhaamousehaiis‘There is a mouse in the room.’ (Hindi; Dayal 2004 ex. 19a)• In this, Hindi definites differ from those in English containing determinerthe—which presupposes uniqueness and familiarity.Page 4Sadhwi Srinivas and Kyle Rawlins (Johns Hopkins University)On the seemingly indefinite readings of Kannada bare singulars March 8, 2020• Under this account, non-specific interpretations are simply a consequence ofthe lack of familiarity of the referent.• Challenges of the account:– Indefinite readings of bare subjects can be obtained not just in caseslike (29) where there is no potential antecedent, but also in cases wherethere is one: e.g., (30) in Hindi.(30) KalyesterdayAbhinavAbhinavekonecuuhe-komouse-ACCgharhomelebroughtaaya.cameKamreRoommeinincuuhaamousehai.is‘Yesterday, Abhinav brought home a mouse. A mouse/#the mouseis in the room’– The non-specificity (speaker-newness) of the indefinite readings doesnot follow from absence of familiarity: IOTA is scopally inert; amountsto wide scope interpretation. E.g. interaction with negation does notmatch Kannada data.– We have also seen Kannada examples where familiarity/discourse-oldnessappears to be required (≈ the indefinite interpretation predicted by thisversion of IOTA is not available.)– Third, this analysis views definiteness in English as being fundamen-tally different from definiteness in ‘determinerless’ languages. How-ever, we prefer a (more) uniform analysis of definiteness.Takeaway: ‘indefinite’ bare nominal readings can be reduced to definite readings,at the cost of substantially weakening the notion of definiteness used (non-familiaruses possible). However, this weakening seems too strong, and still misses somedata.(III) after Aguilar-Guevara (2014)Basic idea: some indefinite-like definites (weak definites) are kind denoting, simi-lar to singular definite generics in e.g. English (the common noun denotes an entityat the taxonomic level and the determiner denotes ι). Not about bare singulars perse, but about similar phenomenon in overtly definite singulars in languages likeEnglish:(31) Take the elevator to the third floor.(uttered in a context where there are 3 elevators)Interpretation: Take any one of these elevators to the third floor.(32) mooraneethirdmaaDifloortankatillliftelevatortogoLitake‘Take the elevator to the third floor.’• Weak definites are a class of definites that do not seem to be subject to unique-ness restrictions (Poesio, 1994; Carlson and Sussman, 2005; Aguilar Gue-vara, 2014; Aguilar-Guevara and Zwarts, 2010; Schwarz, 2014):• Weak defs also receive non-specific readings in episodic contexts, e.g. (31).• This proposal involves a Kind Lifting Rule (KLR), according to which theverb is associated with a systematic ambiguity between ordinary object-levelmeanings where it accepts a regular object as its internal Theme argument, aswell as a kind-level meaning in which it accepts an internal kind argument.• With object-level predicates in episodic sentences, instantiations of the kindare obtained through the Realization relation R (Carlson, 1977).• Challenge of the account for our data: Given the nature of KLR, this proposaldoes not explain weak definites in subject positions like in (33) in English and(25) in Kannada (because Agent is not an internal argument of the event).(33) The mailman didn’t bring anything today.• The analysis also contradicts Dayal’s (2004) conclusion that singular definitegenerics in English do not have access to instantiations. Instead, it takesinstantiations to be available without further comment.Takeaway: Tantalizing connection to overtly marked weak definites; kind inter-pretation may be involved (again). However, analysis can’t apply as-is, because offrequent subject position bare nominals.5 Our proposalBasic idea: We propose an alternative analysis for bare singulars in Kannada wherethe ‘true’ existential reading arises neither through the kind reading (as in Aguilar-Guevara on weak definites or Dayal #1) nor through IOTA (cf. Dayal #2), butthrough the compositional operation of Predicate Restriction (Chung and Ladusaw(2003); Restrict).• However, definite and kind readings can still also arise via typeshifts (regularIOTA, NOM).• Re type-shifting analyses of definite readings for bare nominal, see also Jenks(2015, 2018) on Mandarin, Thai (won’t do a full comparison here, as thiswould require exploring demonstratives as well).• As commonly assumed of bare nominals, we take the bare singular in Kan-nada to be of type 〈e,t〉.Page 5Sadhwi Srinivas and Kyle Rawlins (Johns Hopkins University)On the seemingly indefinite readings of Kannada bare singulars March 8, 2020• Bare nominal undergoes type-shifting to e in appropriate contexts using oneof two operators: NOM (Chierchia 1998 via Dayal; like Dayal we adopt thissimply for concreteness) and a fairly standard IOTA (Partee, 1986), definedin (34)-(35).(34) (NOM / ∩) : λP . λs . ιx[Ps(x)](35) (IOTA) : λP . ιx[P (x)]• With kind-level predicates like in (8), NOM applies to pick out the maximalgroup of entities that instantiate the kind in that situation. In (1)-(3) wherethe bare singular is interpreted definitely, IOTA applies.• Where neither the kind nor definite reading is available, like (25) uttered outof the blue in the absence of a hearer-old dog, the non-specific reading isobtained through Chung & Ladusaw’s Restrict operation, defined as in (36)(following Rawlins 2013’s formalization).• (Can be thought of as a general version of Predicate Modification.)(36) Where β = 〈β1〈...〈βn, t〉〉〉 such that n ≥ 0Restrict(A〈e,β〉, B〈e,t〉)= λx ∈ De . λy1 ∈ Dβ1 . ... . λyn ∈ Dβn . A(x)(y1)...(yn) ∧B(x)• Informally, this amounts to the 〈e,t〉 property restricting the domain of thepredicate it combines with to only those entities that share that property. Afew key cases:– (A is a VP) Restrict(A〈e,t〉, B〈e,t〉) = λxe . A(x) ∧B(x)– (A is a V) Restrict(A〈e,〈e,t〉〉, B〈e,t〉) = λxe . λye . A(x)(y) ∧B(x)– (A is an event-semantics-style vP)Restrict(A〈e,〈v,t〉〉, B〈e,t〉) = λxe . λyv . A(x)(y) ∧B(x)• We assume that type e argument slots can be closed by an existential closureoperation (Chung and Ladusaw 2003; building on Matthewson 1999 a.o. onEC for choice function variables). In Kannada, we view this as a last-resortoption, rather than a free option.• A narrow-scope reading is assured: existential closure of the unsaturatedargument is forced to occur at the event level (at the latest), while nega-tion/other quantifiers are interpreted outside the event domain. (See Cham-pollion 2015 for a recent thorough discussion.)• This proposal avoids problems identified in the other accounts: narrow-scopecomes for free with Restrict, non-specific readings are no longer position-dependent, and it is more parsimonious as it avoids additional ambiguity anddoes not preclude a unified analysis of definiteness across languages.• Generalizations about which operation is preferred (not formally analyzed):– All things equal, IOTA is preferred.– All things equal, subject position/sentence-initial position disprefersnon-IOTA readings.– NOM is forced by kind-level lexical predicates.– Constraints on anaphoric definite readings not developed here. (WIP)– Side-note: we take distribution of ACC in Kannada to correlate with theabove factors, rather than providing an interpretive constraint directly.Different from Hindi (as described in the literature).– Other, murky constraints...5.1 Example derivationConsider the sentence in (6), repeated below in (37):(37) a. (Discourse initial) There is something wrong with today. The sun isn’tshining, the flowers aren’t blooming, and...b. HoragaDe,OutsidehakkibirdjooraagiloudlyhaaDtasingingilla.not“It is not the case that there is a bird singing loudly outside.”We assume the following structure at LF (ignoring tense and modal projections) forthis sentence, with entries for the individual lexical items defined as in (38)-(43).haaDtajooraagiAGENThakki∃ehoragaDe∃villa(38) JhaaDtaKc,g= λev.sing′(e)(39) JjooraagiKc,g= λev.loud′(e)(40) JAGENTKc,g= λxe.λev.Ag(e, x) (Kratzer, 1996)(41) JhakkiKc,g= λxe.bird′(x)(42) JhoragDeKc,g= λev.outside′(e)(43) JillaKc,g= λP.¬PThe sentence is then interpreted through step-by-step composition as follows:(44) Jjooraagi haaDtaKc,g= λev.sing′(e) ∧ loud′(e)(45) JAGENT jooraagi haaDtaKc,g = λxe.λev.sing′(e) ∧ loud′(e) ∧ Ag(e, x)(by Event Identification)Page 6Sadhwi Srinivas and Kyle Rawlins (Johns Hopkins University)On the seemingly indefinite readings of Kannada bare singulars March 8, 2020(46) Jhakki AGENT jooraagi haaDtaKc,g =λxe.λev.sing′(e) ∧ loud′(e) ∧ Ag(e, x) ∧ bird′(x) (by Restrict)The entity arg xe must be immediately saturated by Existential Closure in order toavoid composition failure, giving us the intended non-specific meaning of the baresingular.(47) EC(Jhakki AGENT jooraagi haaDtaKc,g) =λev.∃xe.sing′(e) ∧ loud′(e) ∧ Ag(e, x) ∧ bird′(x)(48) JhoragaDe [hakki AGENT jooraagi haaDta]ECKc,g =λev.∃xe.sing′(e) ∧ loud′(e) ∧ Ag(e, x) ∧ bird′(x) ∧ outside′(e)Negation is assumed to always scope over the event level (Champollion 2015 a.o.),ensuring narrow scope of the bare nominal in negated sentences.(49) EC(JhoragaDe [hakki AGENT jooraagi haaDta]ECKc,g) =∃ev.∃xe.sing′(e) ∧ loud′(e) ∧ Ag(e, x) ∧ bird′(x) ∧ outside′(e)(50) Jilla [horagaDe [hakki AGENT jooraagi haaDta]EC ]ECKc,g =¬(∃ev.∃xe.sing′(e) ∧ loud′(e) ∧ Ag(e, x) ∧ bird′(x) ∧ outside′(e))It should be noted that if the context is modified so that hearer-oldness of the bird isgiven in the discourse, as in (51), then this leads to the definite reading of the baresingular hakki in (51-b).(51) a. (Context: The discourse participants have a pet bird who usually singsvery noisily.)b. HoragaDe,OutsidehakkijbirdjooraagiloudlyhaaDtasingingilla.not(Adujithushaaraagiwellide-aa?)COP-Q“The bird isn’t singing loudly outside. (Is it feeling well?)”In this case, composition proceeds not via Restrict, but by type-shifting with IOTA.Steps (46)-(47) above are replaced with (52):(52) J [[hakki]IOTA [AGENT jooraagi haaDta]]Kc,g =λev.sing′(e) ∧ loud′(e) ∧ Ag(e, ιx.bird′(x))(53) Jilla [horagaDe [[hakki]IOTA [AGENT jooraagi haaDta]]]ECKc,g =¬∃ev.sing′(e) ∧ loud′(e) ∧ Ag(e, ιx.bird′(x))6 Implications / Future work• One main implication of our proposal is that it does not take for grantedthat bare singulars in determinerless languages like Kannada are similar toEnglish singular definite generics (unlike Dayal 2004).• Instead, bare singulars may be closely related to English weak defs, given theready availability of kind readings with the bare singular in object position:(54) Nana-geI-DAThulitigerkanDretowardsbhaya.fear“I am afraid of tigers.” (kind readingX)• This has the further implication that weak definites in English may not beso similar to the usual analysis of English regular definites, contra Aguilar-Guevara’s (2014) view.– Further connection to develop: Coppock and Beaver (2015) argue fora type-shifting view even in the presence of overt definiteness markingon a D! (Maybe what we should have been studying are bare nominallanguages in the first place...)• In this talk, we have focused only on the bare singular while implicitly ac-cepting Dayal’s analysis for how non-specific readings of bare plurals arise(i.e., via access to plural instantiation sets). But it is worth thinking aboutwhether such a distinction must be maintained in Kannada.• Our proposal raises the question of why Restrict is not always preferred overIOTA and NOM, as might be expected to be under a “last resort” view oftype-shifting. One possibility is that this is due to IOTA and NOM beingassociated with stronger presuppositions, such that the stronger meaning ispreferred when it holds (by Maximize Presupposition!)• A crucial missing piece of our analysis is to characterize precisely those caseswhere definite readings are unable to arise. This is part of a larger project thatis WIP.ReferencesAguilar Guevara, A. (2014). Weak definites. Semantics, lexicon and pragmatics. LOT.Aguilar-Guevara, A. and J. Zwarts (2010). Weak definites and reference to kinds. In Semantics andLinguistic Theory, Volume 20, pp. 179–196.Ahn, D. (2019). THAT thesis: A competition mechanism for anaphoric expressions. Ph.D. dissertation,Harvard University.Barker, C. (2004). Possessive weak definites. University of Massachusetts Occasional Papers (29), 89.Carlson, G. (1977). Reference to kinds in English. Ph. D. thesis, University of Massachussetts, Amherst.Carlson, G. and R. Sussman (2005). Seemingly indefinite definites. Linguistic evidence: Empirical,theoretical, and computational perspectives 85, 71–85.Champollion, L. (2015). The interaction of compositional semantics and event semantics. Linguisticsand Philosophy 38, 31–66.Chierchia, G. (1998). Reference to kinds across language. Natural language semantics 6(4), 339–405.Christophersen, P. (1939). The Articles: A Study of Their Theory and Use in English. Einar Munksgaard.Chung, S. and W. A. Ladusaw (2003). Restriction and saturation, Volume 42. MIT press.Coppock, E. and D. Beaver (2015). Definiteness and determinacy. Linguistics and Philosophy 38(5),377–435.Page 7Sadhwi Srinivas and Kyle Rawlins (Johns Hopkins University)On the seemingly indefinite readings of Kannada bare singulars March 8, 2020Dayal, V. (1992). The singular-plural distinction in Hindi generics. In Semantics and Linguistic Theory,Volume 2, pp. 39–58.Dayal, V. (1999). Bare NPs, reference to kinds, and incorporation. In Semantics and Linguistic Theory,Volume 9, pp. 34–51.Dayal, V. (2004). Number marking and (in)definiteness in kind terms. Linguistics and Philosophy 27(4),393–450.Dayal, V. (2011). Hindi pseudo-incorporation. Natural Language & Linguistic Theory 29(1), 123–167.Dayal, V. (2015). Incorporation: Morpho-Syntactic vs. Semantic Considerations. In The Syntax andSemantics of Pseudo-Incorporation. Leiden, The Netherlands: Brill.Deal, A. R. and J. Nee (2018). Bare nouns, number, and definiteness in Teotitl´ an del Valle Zapotec. InProceedings of Sinn und Bedeutung, Volume 21, pp. 317–334.Diesing, M. (1992). Indefinites. MIT Press.Evans, G. (1977). Pronouns, quantifiers, and relative clauses (i). Canadian journal of philosophy 7(3),467–536.Gerstner, C. and M. Krifka (1987). Genericity, an introduction. Ms., Universita¨t Tu¨bingen.Heim, I. (1982). The Semantics of Definite and Indefinite Noun Phrases. Ph. D. thesis, UMass Amherst.Jenks, P. (2015). Two kinds of definites in numeral classifier languages. In Semantics and LinguisticTheory, Volume 25, pp. 103–124.Jenks, P. J. (2018). Articulated definiteness without articles. Linguistic Inquiry 49, 501–536.Kadmon, N. (1990). Uniqueness. Linguistics and Philosophy 13(3), 273–324.Kamp, H. (1981). A theory of truth and semantic representation. Formal semantics-the essential read-ings, 189–222.Kratzer, A. (1996). Severing the external argument from its verb. In Phrase structure and the lexicon,pp. 109–137. Springer.Lidz, J. (2006). The grammar of accusative case in Kannada. Language 82, 10 – 32.Lo¨bner, S. (1985). Definites. Journal of semantics 4(4), 279–326.Matthewson, L. (1999). On the interpretation of wide-scope indefinites. Natural Language Seman-tics 7(1), 79–134.Partee, B. H. (1986). Noun phrase interpretation and type-shifting principles. In M. J.Groenendijk,D.de Jongh (Ed.), Studies in discourse representation theory and the theory of generalized quantifiers,pp. 115–143. Foris.Poesio, M. (1994). Weak definites. In Semantics and Linguistic Theory, Volume 4, pp. 282–299.Rawlins, K. (2013). About ‘about‘. In Semantics and Linguistic Theory, Volume 23, pp. 336–357.Roberts, C. (2003). Uniqueness in definite noun phrases. Linguistics and Philosophy 26(3), 287–350.Russell, B. (2005). On denoting. Mind 114(456), 873–887.Schwarz, F. (2014). How weak and how definite are weak definites. Weak referentiality, 213–235.Wilkinson, K. (1991). Studies in the semantics of generic Noun Phrases. Ph. D. thesis, University ofMassachussetts, Amherst.Page 8


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