Bare singular noun phrases in Rioplatense Spanish
Carolina Oggiani- Universidad de la República, Uruguay/ SNI
Bare singular noun phrases –also called bare singulars (BS)– are nominal structures that lack a determiner. Even though Spanish belongs to the languages with a developed system of articles, BS have been well-attested in the literature (Espinal & McNally, 2001, among others). This presentation seeks to show that Rioplatense Spanish (RPS) displays a broader picture of argumental BS, which have not yet been properly explored. Based on different tests (the nature of their modifiers, their number interpretation, their discourse properties, their adjacency to their governor and their lexical restrictions), we claim that there are three classes of BS in RPS.
Assuming that all nouns are predicative expressiones (Matushansky, 2008), we propose the following nominal system, in which nouns enter the derivation with one of the following lexical features (similar to Gomeshi & Massam, 2009):
(1) a. Noun[common/weak]: {xk/o: property of x)}
b. Noun[strong]: {xk/o: property of x ∧ is-named x, (Nstrong)}
Within the first group, named weak definites, BS behave as complements of locative prepositions en (in) and a (to).
(2) a. Los jugadores están en cancha.
the players are in field
‘The players are in the football field.’
b. Los niños van a piscina de 11 a 12.
the kids go to pool from 11 to 12
‘The kids go to the pool from 11 to 12.’
We claim that group1 BS have a weak definite reading. They combine with the iota operator and they get their [ WEAK] feature when the locative preposition slects for an abstract entity of the type <e k> (as in de Swart, 2015). Via a Realization Relation, this abstract entity is instantiated by a particular object and via a Usage Relation (Aguilar-Guevara, 2014) we account for their enriched meaning, associated with a prototypical activity. Syntactically, we argue that they are numberless DPs, whose D hosts a null determiner (Longobardi, 1994).
The second group, called strong definites, comprises nouns that freely appear in any syntactic position.
(3) a. Rectorado queda en primer piso rectorate is in first floor
‘The Rectorate (my rectorate) is in the first floor.’
b. Renovaron secretaría el año pasado.
refurbished-3.p secretariat the year last
‘The secretariat (my secretariat) was refurbished last year.’
We argue that group 2 BS enter the derivation with a [ STRONG] feature, that enables their strong definite reading (<e o> type) (Longobardi, 1997; Matushansky, 2008). We claim that they combine with the iota operator after the semantic operation Possesion Relation (Barker, 2011) is applied, which limits the referent to local institutions,
“owned” (in broad terms) by the speaker or hearer of the sentence. As for their syntax, they are DPs –with a null determiner– that select for a NumP.
There is a third group, whose BS are restricted to object position of presentational and creational verbs. We name them weak indefinites.
(4) a. Estrenamos obra el sábado premiered-3.p play the Saturday
‘We premiered a play last Saturday’.
b. Terminé monografía de Semántica finished-1.sg. paper of semantics
‘I finished a paper on Semantics’.
We argue that they are pseudo-incorporated structures (Dayal, 2011) and that they get their [WEAK] feature (type <eo,t>) once they combine with creational and presentational verbs. Group 3 verb+BS combinations also add some semantic enrichment, to be explained by the appropriately classificatory notion (Dayal, 2003). That is to say, the
subject is interpreted as the agent of a prototypical activity. Furthermore, we claim that they project up to NumP– given their singular interpretation– and that they pseudoincorporate to the verb as a predicate modifier.
Overall, this paper provides sufficient empirical evidence that shows that, even though BS in RPS are superficially alike, they are syntactically and semantically different. Moreover, we seek to show that it is possible to derive their compositional meaning and that their semantic enrichmeant arises as a result of article omission.
References Aguilar-Guevara, A. (2014) Weak definites. Semantics, lexicon and pragmatics. LOT.// Barker, C. (2011) Possessives and relational nouns. In Von heusinger, Maienborn, and Portner (eds). Semantics: An international Handbook of Natural Language Meaning. Hsk 33.2.// Dayal, V. (2003). A semantics for pseudoincorporation. Ms. Rutgers University. / Dayal, V. (2011) Hindi pseudo-incorporation. Natural Language & Linguistic Theory, 29(1):123–167.// de Swart, H. (2015) Constructions with and without articles. In Borik, O. & Gehrke, B., editors, T he Syntax and Semantics of Pseudo-Incorporation , 126–156. Brill.// Espinal, M.T. and McNally, L. (2011) Bare nominals and incorporating verbs in Spanish and Catalan. Journal of Linguistics, 47(01):87–128.// Ghomeshi, J. & Massam, D. (2009). The proper D connection. In Ghomeshi, J., Paul, I., & Wiltschko, M., editors, Determiners: Universals and variation , volumen 147, pp. 67–96. John Benjamins.// Longobardi, G. (1994) Reference and proper names: A theory of N-movement in Syntax and Logical
form. Linguistic Inquiry , 609–665.// Matushansky, O. (2008) On the linguistic complexity of proper names. Linguistics and Philosophy, 31(5):573–627.//
The negation of cualquier in a non-argumental position
Ana Clara Polakof (Universidad de la República, Uruguay/ SNI)
Etxepare and Uribe-Etxebarria (2018) assume that cualquier sometimes has an existential value, and sometimes a universal in a negated environment. Thus, they explain the dierences between 1 and 2 by assuming that 1a does not have the universal reading available, while 2a does have it. They argue that in 1, it is not just the existential value which renders 1b ungrammatical. It is the fact that the PP de estos introduces a non-scalar reading of days that forces them to be interpreted as similar.
1. (a) Cualquier da de estos me suicido
any day of these CL.reex commit.suicide(I)
`One of these days I just might kill myself.'
(b) *No cualquier da de estos me suicido.
neg any day of these CL.reex commit.suicide(I)
`I won't kill myself one of these days.'
2. (a) Cualquier da es bueno para suicidarse.
any day is good for committing suicide
`Any day is a good one to commit suicide.'
(b) No cualquier da es bueno para suicidarse.
not any day is good to commit suicide
`Not just any day is a good one to commit suicide.'
In this presentation, we will sketch a solution to the problem at hand by assuming that cualquier is a universal indeterminate pronoun, as in Menendez-Benito (2010). To explain why 1a is grammatical, and 1b is ungrammatical, we assume that what makes 1a grammatical is the presence of a PP which involves an indexical which allows us to refer to instances of times in the world. Most importantly, we analyse it as a case of subtrigging because the same constituent unsubtrigged is ungrammatical, as in 3. If we replace de estos with another modier, instead, the sentence is still grammatical when non-negated and ungrammatical when negated, as in 4. Thus,
subtrigging, as well as its apparition in a non-argumental postion in 1b, plays a fundamental role.
3. (a) # Cualquier da me suicido
any day CL.reex commit.suicide(I)
# `Any day I kill myself.'
(b) # No cualquier da me suicido.
neg any day of these CL.reex commit.suicide(I)
`Not just any day I kill myself.'
4. (a) Cualquier da que quiera me suicido
1
any day that (I) want CL.reex commit.suicide(I)
`Any day that I want I just might kill myself.'
(b) *No cualquier da que quiera me suicido.
neg any day that (I) want CL.reex commit.suicide(I)
`Not just any day that I want I might kill myself.'
If we take Kratzer (2005) into account, we may defend that cualquier NP would always trigger sentential quantication. Thus, we would also expect 1 to involve a universal propositional operator, as Menendez-Benito (2010). Following Aloni (2019), exh applies to the subtrigged cualquier NP, and a type shift e allows it to
refer to the maximal collection of these days in w0. Next, # avoids vacuous quantication with regards to the propositional quantication triggered by the universal indeterminate pronoun. We assume, in addition, that the phrase is introduced by an ommited preposition that allows cualquier NP to introduce instances of times as
in 5b:
5. (a) Cualquier da de estos me suicido
`Any day of these I just might kill myself.'
(b) [8] ((at # (SHIFTe (exh[cualquier da, de estos])) me suicido )
The path to explain the ungrammaticality of the negated cualquier NP in 1b is clearer. Since it is constituent negation, the negative operator remains within the temporal DP, and, following Collins (2017), we can defend that a DP of type e cannot be modied by negation, and that only generalized quantiers which involve a t may be negated. Though the neeed for generalized quantiers may be questioned (Kratzer (2005), among others), what matters is that negation cannot interact with a type e phrase. Thus, we can defend that, since cualquier NP has been typed shifted to e, a representation such as 6b should be generated. However, since negation cannot modify cualquier NP directly, the derivation fails:
6. (a) *No cualquier da de estos me suicido
`Not Any day of these I just might kill myself.'
(b) [8] (Me suicido at : (SHIFTe (exh[cualquier da, de estos]))
References Aloni, M. (2019). Indenites as fossils: The case of wh-based free choice. Collins, C. (2017). The distribution of negated quantier phrases in English. Ms., NYU. Etxepare, R. and Uribe-Etxebarria, M. (2018). Context-Sensitive Aspects of Constituent Negation. In Martin, R. and Gallego, A. J., editors, Language, Syntax,
and the Natural Sciences, pages 77{98. Cambridge University Press, Cambridge. Kratzer, A. (2005). Indenites and the operators they depend on: From Japanese to Salish. In Carlson, G. and Pelletier, F., editors, Reference and Quantication: The Partee Eect, volume 173, pages 113{142. CSLI Publications Stanford, CA. Menendez-Benito, P. (2010). On universal free choice items. Natural Language Semantics, 18(1):33{64.
A comparative study of English and Spanish vulgar minimizers
Angel Luis Jiménez-Fernández (University of Seville)
Susagna Tubau (Universitat Autònoma de Barcelona)
In order to shed light on vulgar minimizers (VM) as a cross-linguistic class, we compare the behaviour
of VM and non-vulgar minimizers (NVM) in English (E) and Spanish (S). The relevant facts are the
following:
[1] while NVM must co-occur with negation in negative contexts, (1) & (2), VM optionally co-occur
with negation in negative contexts, (3) & (4).
(1) They have(*-n’t) said a word about it. [E]
(2) (*No) probó (ni) un bocado durante la comida. [S]
not tried not even a bite during the lunch
‘S/he didn’t have a bite during lunch.’
(3) They (don’t) know jack shit about linguistics. [E]
(4) (No) sé una mierda de lingü.stica [S]
not know.1SG a shit of linguistics
‘I (don’t) know jack shit about linguistics’
Following Postal (2004), we assume that VM correspond to two distinct lexical entries: (i) a DP with a zero-incorporated numeral, (5), and (ii) a polarity item (PI), (6).
(5) [DP [D zero] + [N jack shit]] (6) [DP [D Æ] + [N jack shit]]
The entry in (5) yields a negative reading but is not diagnosed as sentential negation by Klima’s (1964) tests, (7) & (8). NVM, by contrast, are PIs corresponding to the lexical entry in (9).
(7) They know jack shit about linguistics {and so do you / *and neither do you}. [E]
(8) Sé una mierda de lingü.stica {y tu también / *y tu tampoco}. [S]
know.1SG a shit of linguistics {and you too / and you neither}
‘I know jack shit about linguistics {and so do you}.
(9) [DP [D a] + [N word]]
[2] In fragment answers, VM are optionally preceded by ‘not’ in E, (10A.a), and by ni ‘not even’ in S, (10A.b), whereas ‘not’/ni are compulsory with NVM, (11) (Vallduví 1994, Tubau 2016).
(10) [Question: What does she know about linguistics?]
A: a. (Not) jack shit. [E]
b. (Ni) una mierda. [S]
not even a shit
‘(Not) jack shit.’
(11) [Question: What did she say about it?]
A: a. (*Not) a word. [E]
b. (*Ni) una palabra. [S]
not even a word
‘Not a word.’
Such behaviour is consistent with NVM only having the structure in (9), and VM having either the structure in (5) or (6). Assuming that fragment answers involve ellipsis after Focus-fronting of the fragment (Merchant 2004), (11) shows that a PI such as (6)/(9) can only be Focused if preceded by a negative licensor.
[3] In E only VM can be preposed without an overt ‘not’, (12) vs. (13). This is the same for S VM and the particle ni, (14) vs. (15).
(12) Mary knows a lot about Maths but (not) jack shit she knows about linguistics!
(13) (*Not) a word did Mary say about it.
(14) (Ni) una mierda sabe Pedro de lingü.stica.
not even a shit knows Pedro of linguistics
‘(Not) jack shit Peter knows about linguistics.’
(15) (*Ni) un bocado probó María durante la comida.
not even a bite tried Mary during the lunch
‘Not a bite did Mary eat during lunch.’
[4] In E, only NVM trigger Subject-Auxiliary inversion (SAI), (13) (vs. (12)). In S, both VM and NVM trigger SAI when preposed, (14) & (15). Let us assume that languages divide into three groups depending on whether they require SAI in regular Focalisation: in some languages (e.g. S) SAI is obligatory, in others optional, and in others banned (e.g. E), (16).
(16) THIS BOOK I don’t need (but that one I do). (Haegeman 2012: 8)
Hence, the grammaticality of (12) is predicted to the extent that it involves regular Focalisation, rather than Focalisation of a negative expression (also known as Negative Preposing –Radford 2009, Haegeman 2012), (17), with SAI.
(17) Never in my life have I been to Africa. Lack of SAI with VM, therefore, indicates that these are not negative. This is expected for VM with a zero-incorporated numeral, (5), and also confirmed to be the case for VM preceded by ‘not’ if Jackendoff’s (1969) It is not so that test is used, (18). According to Jackendoff, only sentences that
can be paraphrased with It is not so that [S X-Y] express sentential negation. As shown in (18b), the sentence in (18a) cannot be paraphrased as (18b).
(18) a. Not jackshit she knows about linguistics.
b. #It is not so that she knows jackshit about linguistics. We argue that while VM can be Focused in E, they cannot undergo Negative Preposing (triggering SAI) because they are not negative. The not that optionally precedes the VM is, unlike what is the case for NVM, an instance of constituent negation. In S, if preceded by ni, VM and NVM behave similarly to negative indefinites such as nadie ‘n-body’ and nada ‘n-thing’ (cf. Tubau 2020): when
VM/NVM are moved to the left, sentential negation cannot be overt (which is typical of Non-Strict Negative Concord languages), (19), but, unlike in E, negation is diagnosed as sentential, as shown not only by Klima’s ‘either/too’ test, (20), but also by Jackendoff’s It is not so that X-Y test (translated into S as ‘No es el caso que X-Y’ and with the verb in the subjunctive mood).
(19) a. No veo (ni) un alma / (ni) un pijo. [NVM / VM]
not see.1SG (not even) a soul / (not even) a dick
‘I don’t see a soul / dick.’
b. (*Ni) un alma / (*Ni) un pijo *(no) veo. [NVM / VM]
(not even) a soul (not even) a dick not see
‘Not a soul did I see.’ / ‘Not dick I saw.’
(20) {Ni un alma / ni un pijo} veo y tú tampoco / *y tú también.
not even a soul not even a dick see.1SG and you either / and you too
‘Not a soul do I see and neither do you’ / ‘Not dick I see and so do you.’
(21) No es el caso que vea {(ni) un alma / (ni) un pijo}.
not is the case that see.1SG.SUBJ not even a soul not even a dick
In short, E and S show many similarities, but also a crucial difference: while VM in E are not negative,
they can be in S thanks to ni. In E, the not that can optionally precede VM when these are preposed
is an instance of constituent negation.
References: Haegeman, L. 2012. Adverbial clauses, Main Clause Phenomena, and the Composition of the Left Periphery. Oxford: OUP. Jackendoff, R. 1969. An interpretive theory of negation. Foundations of Language 5: 218-241. Klima, E.S. 1964. Negation in English. In J.A. Fodor & J. J. Katz, eds. The structure of language. Readings in the philosophy of language, 246-323. New Jersey: Prentice Hall. Merchant, J. 2004. Fragments and ellipsis. Linguistics and Philosophy 27:661-738. Postal, P. 2004. The structure of one type of American English vulgar minimizer. In P. Postal, ed. Skeptical linguistic essays, 159-172. Oxford/New York: OUP. Radford, A. 2009. Analysing English Sentences: A Minimalist Approach. Cambridge: CUP. Tubau, S. 2016. On the syntax of English
minimizers. NLLT 32(1): 739-760. Tubau, S. 2020. Maximizers and minimizers as different types of polarity items. In M.T. Espinal & V. Déprez, eds. The Oxford Handbook of Negation, 407-425. Oxford: OUP. Vallduví, E. 1994. Polarity items, n-words and minimizers in Catalan and Spanish. Probus 6: 263-294.
Flavors of intensity and emphasis: the case of muito in Brazilian Portuguese
Luciana Sanchez-Mendes (Universidade Federal Fluminense UFF)
Data This paper discusses a topic that has been recently drawing attention of semanticists: the fact that some lexical items can convey different flavors of intensity and emphasis (see, e.g. the overview presented in Beltrama and Trotzke, 2019). More specifically, this work addresses the issue investigating muito in Brazilian Portuguese in examples such as the one in (1) bellow (spontaneous data from Twitter) in contrast with its prototypical use as an intensifier in a wide range of domains (2):
(1) Foi muito golpe. ‘It was really a coup’
(2) Quadros Gomes (2011)
a. Tem muito aluno no pátio. ‘There are many students at the courtyard’ COUNT-SG
b. Tem muitos alunos no pátio. ‘There are many students at the courtyard’ COUNT-PL
c. Tem muito leite na geladeira. ‘There is a lot of milk at the fridge’ MASS
d. João é um aluno muito inteligente. ‘João is a very smart student’ ADJECTIVE
e. João correu muito. ‘João ran a lot’ VERB
f. João correu muito rápido. ‘João ran very fast’ ADVERB
One or two muitos? This behavior of muito is similar to totally in English (Beltrama, 2018), suffix -issimo in Italian and šému in Washo (Beltrama & Bochnak, 2015) in the sense that they all are intensifiers that can be applied under certain circumstances to predicates that are not grammatically gradable displaying some sort of emphatic content. These data rise the question if: (i) there are two related but different meanings for these items (one for emphasis (1) and one for intensive readings (2)) or if they have the same semantic core that can derive these variants. Rainer (1983) pursued the first option by positing two lexical entries to the suffix -issimo in Italian. The second option can either assume one lexical entry for the modifiers and consider that the flavors of modification are due the modified expression (as Beltrama, 2018 for totally in English) or postulate one single compositional analysis for all the uses (as Beltrama & Bochnak, 2015 for -issimo in Italian and šému in Washo).
Proposal Is this paper I argue in favor of an analysis for muito based on one lexical entry following some assumptions of Bertrama’s (2018) proposal for totally. Nevertheless, I posit a slightly different pragmatic content to muito. In Bertrama’s (2018) analysis, in contexts like (3), totally is an attitude level modifier that does not affect the truth conditions of the sentence.
(3) You should totally click on that link! It’s awesome.
In order to analyze totally in (3) in accordance with its use as an intensifier of predicates that are associated to upper-bounded scales (e.g. totally full, cf. Kennedy & McNally, 2005), Beltrama (2018) proposes that the modifier can be applied to a scale based on the speaker’s commitment that is equally upper-bounded. According to Beltrama (2018), the contribution of emphatic totally is to strengthen the speaker’s commitment by adding the proposition to the common ground.
I follow Beltrama’s proposal to totally in the sense that the emphatic use of muito is also pragmatic and does not contribute to the at-issue content of the sentence. Nevertheless although emphatic muito (pt) and totally (eng) have a similar content that can be summarized as “unquestionably” they differ in the sense that totally is used in contexts in which the speaker is adding an information that s/he usually doesn’t take for granted including cases of outlandish statements. That’s why (4a) is odd but (4b) is adequate.
(4) a. Luke totally got married at 25. b. Luke totally got married at 12.
Muito in turn is used when the speaker wants to emphasize its certainty about the proposition as in examples in (5) and (6) (spontaneous data). These examples show that emphatic muito has a similar content of Verum focus expressions (cf. Romero & Han, 2004 for really).
(5) Vai ter muita Copa, gente! Meaning: The World Cup will really happen.
(6) Tá tendo muito greve sim! Meaning: It is really happening a strike.
Nevertheless, muito is not entirely like Verum focus expressed by really. Really is used only in epistemically biased contexts such as Are you sure that “p”?, whereas totally and muito can also be
used in unbiased contexts without questioning how certain the speaker is (7) (see Beltrama, 2018 for discussion for totally).
(7) Vai ter Copa? Meaning: Will the World Cup happen?
Vai ter muita Copa! Meaning: The World Cup will really happen.
So muito has a similar contribution as Verum focus but a distribution comparable with an intensifier such as totally. Then its adequate analysis might combine these two characteristics. I will engage in a proposal that considerer muito as a degree modifier whose interpretation is dependent on the scale of the modified phrase (see Quadros Gomes & Sanchez-Mendes, 2015 for discussion). Formally, I propose that muito is a degree modifier that is context-sensitive in the same sense as the pos morpheme is dependent on the scale encoded by the modified expression (cf. Kennedy, 2007). This property is capture by s that is a context-sensitive function that returns a standard of comparison based on properties of both the modified phrase g and on the context of assertion.
(8) [[ muito ]] = λgλx. g(x) s(g)
In its intensifier use, muito is applied to gradable predicates and boosts its property application (e.g muito bonito means very beautiful). In its emphatic use it is applied to a propositional level scale that encodes an epistemic content. Crucially its emphatic use is in accordance with the general idea presented in McNally (2016) that modifiers that have pragmatic uses (not affecting the at-issue content) are applied to saturated expressions (individuals and propositions) Than it makes sense to posit a propositional level pragmatic scale available to be attached by degree modifiers. Additionally, according to Doetjes (2008) saturated predicates are precisely the expressions associated to closed/bounded scales. Then I suggest that the propositional level modification of muito involves a closed scale like the totally one.
Then since this pragmatic level scale is closed, the interpretation available for s is its maximal degree/endpoint that represents the maximal commitment of the speaker. This maximal engagement can be presented in words in (9).
(9) ASSERT(muito(p)) = The speaker is certain about the content of p and believes that the linguistic description of p is the best way to add it to the CG
One way to formalize this notion is given in (10). Asserting muito p means that in all words w that belongs to the Projected Common Grounds PCG that are part of PS (the set of common grounds projected by the assertion of p) are p-worlds, i.e. are worlds in which p is true.
(10) ASSERT(muito(p)) = ∀PCG ∈ PS(Assert(p))[∀w ∈ PCG]: p(w)=1
A further question left for this proposal (and also for the totally one) is how muito gets its interpretation on the proposition level and appears linearized inside the verb phrase. Covert movement like the one proposed to quantifiers would be one way of analysis (May, 1985). For now, I leave this question open.
References Beltrama, A. (2018). Totally between subjectivity and discourse: Exploring the pragmatic side of intensification. Journal of Semantics, 35, 219–261. Beltrama, A., & Bochnak, M. R. (2015). Intensification without degrees cross-linguistically. Natural Language & Linguistic Theory, 33, 843–879. Beltrama, A.; Trotzke, A. (2019). Conveying emphasis for intensity: Lexical and syntactic strategies. Language and Linguistics Compass, 13(7), 1–13. Doetjes, J. (2008). Adjectives and degree modification. In: L. MacNally & C. Kennedy (eds.) Adjectives and adverbs: syntax, semantics and discourse. Oxford: Oxford University press, 123-155. Kennedy, C. (2007). Vagueness and grammar: The semantics of relative and absolute gradable adjectives. Linguistics and Philosophy, 30, 1–45. Kennedy, C., & McNally, L. (2005). Scale structure, degree modification and the semantics of gradable predicates. Language, 81, 345–381. May, R. (1985). Logical form. MIT Press. McNally, L. (2016). Modification. M. Aloni & P. Dekker (eds.) Cambridge Handbook of Formal Semantics. Cambridge University Press, 442-466. Quadros Gomes A.P. (2011). Uma proposta de distinção semântica para os intensificadores ‘muito’ e ‘bem'. Estudos Linguísticos, 40, 379-394. Quadros Gomes, A. P. & Sanchez-Mendes, L. (2015) Degree modification in Brazilian Portuguese and in Karitiana. Revista Virtual de Estudos da Linguagem, 13, p. 5-32. Rainer, Franz. (1983). L’intensificazione di aggettivi mediante -issimo. In: D. Maurizio, W. Dressler & G. Held (Eds.) Akten des 2. Österreichischitalienischen Linguistentreffens, Tübingen: Narr, 94–102. Romero, M. & Han, C. (2004). ‘On yes/no questions’. Linguistics and Philosophy, 27:5, 609–658.
Universal Concord as Syntactic Agreement
Ka Fai Yip (Yale University)
Introduction: Concord, a phenomenon where two or more linguistic materials give the same semantic contribution, is extensively studied from negative concord (NC). Its nature is howeversubject to debate, e.g. NC has beentreated as a semantic licensing of NPIs (Giannakidou 2000)or syntactic Agree (Zeijlstra 2004).While impenetrability (e.g. islands) is oftenused in argumentation, intervention effects (IEs)has been paid little attention on.On the other hand, interrogative and existential concord are recognized(Kratzer & Shimoyama 2002), butuniversal concord(UC)seems notto be well-attested. Following Zeijlstra (2004)'s definition for NC, I define UC as two or more universal elements yielding one semantic universal quantification.
Goals:(i) Fill the empirical gap by showing Cantoneseverbal suffix-canis aUCelement; (ii) Providenovel evidence from IEsto argue for a syntactic agreement approach to concord.
Data: The first set of data presents a paradigm of UC. Cantonese verbal suffix -canin (1), claimed to be a universal quantifier (UQ) (Tang 2015, Lee 2017), may express universal quantification over events on itsown, evidenced by the incompatibility with an existential quantifier. Crucially, -cancan occur with an overt UQ like mui-ci'every time' while giving the same semantics(=(2)). Fitting UC's definition, this constitutes a problem to compositionality: how do two UQs give one universal quantification? Note that the co-occurrence of genuine UQsin (3)gives rise to two universal quantification. Since the event variable is already bound by one of the UQs, another UQ cannot bind any variable and results in vacuous quantification (=(4)).
(1) [Aaming(mui-ci/*jaujatci)jam-cannaai]go-touzautung.(∀/ *∃)
Mingevery-timehaveonetimedrink-CANmilkCL-stomachthenache
‘Every time (/ *there was once) Ming drank milk, his tummy felt odd.’
(2) ∀e[DRINK(milk)(Ming)(e)∃e'[ACHE(Ming's stomach)(e')]](where e, e' stand for event variables)
(3) *ZijiuAamingmui-cijamnaai,go-touzautung.
only.ifMingevery.timedrinkmilkCL-stomachthenache
(4) *∀e2∀e1[ DRINK(milk)(e1)(Ming)∃e'[ACHE(e')(Ming's stomach)] ](vacuous quantification)
Another set of data shows IEs in UC.Q(uantificational)-elements cannot occur in between a UQ and -can, e.g.negation, focus, modals, quantifiers. Crucially, this set of interveners coincidewith interveners toother syntactic operators, e.g. ‘why’ andA-not-A (Law 2001, Soh 2005).
(5) mui-cimoudaai(*-can)syu, …
every-timenot.havebring-CANbook
'Every time (he) hadn't brought a book, …'
(6) mui-cidakkeoijung(*-can)gaan-fong, …
e.-t.only3SGuse-CANCL-room
'Every time that only he used the room, …'
(7) mui-cijiuzou(*-can)jego-zan
e.-t.shoulddo-CANstuffthat-time
'Every time when he should work, …'
(8) Zijiujaujanman(*-can)je, …
only.ifhavepersonask-CANstuff
'Once someone asks him for something, …'
Hyper-raising, evidentiality, and phase deactivation
Tommy Tsz-Ming Lee (University of Southern California)
Ka Fai Yip (Yale University)
Introduction. Raising-to-subject out of a finite clause (hereafter hyperraising) is attested in various
languages (Ura 1994), and it raises concerns over the Phase Theory (that suggests that CP is a phase,
Chomsky 2000, 2001) and the Ban on Improper Movement (Chomsky 1973, May 1979). The combined
effects overgeneralize to rule out hyperraising in any language. Existing proposals to hyperraising fall into
two groups: (i) the deficient-CP approach suggests that some CPs are inherently non-phasal, if these CPs
lack some relevant properties of ordinary ones (e.g. Firreira 2005 for Spanish; Zeller 2006 for Nguni;
Alexiadou & Anagnostopoulou 1999 for Greek; Uchibori 2000 for Japanese); (ii) the conditioned-phasedeactivation
approach suggests that a CP ceases to be a phase if it stands in a syntactic relation with a
higher head, as argued in Nunes (2008), Halpert (2016, 2019).
Goals. This study provides further evidence in support of the second approach and we further propose
that an evidential component encoded in the embedding predicate serves to deactivate a CP phase, hence
licensing hyperraising. Since hyperraising is only licensed by a subset of attitude verbs, this approach, by
making reference to evidentiality, captures the language-internal variation among embedding predicates.
The data. In Cantonese and Vietnamese, some attitude verbs (‘feel.like’/‘hear.say’) show a raising
pattern, where the surface subject is thematically related to the embedded predicate (i.e. surface subject ≠
attitude holder), as in (1), despite the presence of a CP boundary (marked by complementizers).
(1) a. Coeng-jyui gamgok/tengman waa ti m wui ting [Cantonese]
CL-rain feel.like hear.say C NEG will stop
b. Cơn-mưa nàyi cảm giác/nghe nói rằng ti sẽ không dừng [Vietnamese]
CL-rain this feel.like hear.say C will NEG stop
Both: ‘(lit.) The rain is felt/heard that (it) won’t stop.’
Yet, the same raising pattern is not observed with other attitude verbs (‘feel-RESULT’/‘know’), as in (2).
(2) a. *Coeng-yui gamgok-dou/zidou waa ti m wui ting [C]
CL-rain feel-RESULT know C NEG will stop
b. *Cơn-mưa nàyi cảm-thấy/ biết rằng ti sẽ không dừng [V]
CL-rain this feel-RESULT know C will NEG stop
Both: ‘(lit.) The rain is felt/known that (it) won’t stop.’
Evidence for A-movement in (1). The embedded subject in (1) lands on matrix Spec TP. First, (3)
suggests that the movement privileges subjects over objects, an asymmetry attributable to a locality
condition for A-movement: the subject is a closer Goal to the matrix Probe than the object. This also
suggests that the subject is not raised by topicalization (which does not show the same asymmetry).
(3) a. go-toifungsubj gamgok [CP tsubj wui ceoiwai jat-zo-singsi] Subj-obj asymmetry [C]
CL-hurricane feel.like will detroy one-CL-city
b. *jat-zo-singsiobj gamgok [CP go-toifung wui ceoiwai tobj ]
one-CL-city feel.like CL-hurricane will detroy
Both intended: ‘(lit.) The hurricane is felt that (it) will destroy a city.’
Second, (4b) shows that the moved subject can bind a pronominal variable that it fails to bind before
movement (=4a), instantiating a typical property of A-movement. Data in Cantonese:
(4) a. *[on keoii caandei ge m-tung] tengman [CP mui-lap-zyunseki dou jau m-tung gwongzaak]
accord it origin MOD different hear.say every-CL-diamond all have different luster
b. mui-lap-zyunseki [ on keoii caandei ge m-tung] tengman [CP ti dou jau m-tung gwongzaak]
every-CL-diamond accord it origin MOD different hear.say all have different luster
‘Every piece of diamondi, according to itsi origin is heard that (it) will have different lusters.’
An evidential component. We propose that attitude verbs can be classified into two groups based on
whether the attitude report is indirect-evidence-based. (5a) shows that verbs in (1b) are compatible with
direct evidence, whereas those in (1a) show the opposite in (5b). The latter ones are only compatible with
indirect evidence. The distinction in evidentiality correlates with their raising possibilities: only the
indirect-evidence-based attitude verbs license raising.
(for a talk or a poster)
(5) Scenario: It is 10°C. Aaming went out without wearing a coat. Shivering, he said:
a. ngo gamgok-dou/zidou ceotmin hou dung [C, same for V cảm-thấy/ biết]
1SG feel-result know outside very cold ‘I feel/ think/ know it is cold outside.’
b. #ngo gamgok/tengman ceotmin hou dung [C, same for V cảm giác/ nghe nói]
1SG feel.like/hear outside very cold ‘I feel like/ hear that it is cold outside.’
The evidential component proposed here can be seen as an extension of von Fintel & Gillies’ (2010)
proposal on English must, which encodes an evidential component in its lexical semantics. The
correlation between raising possibility and evidentiality is further supported by evidence in Romanian
(Alboiu & Hill 2016): when perception verbs take indicative clauses, raising(-to-object) is allowed only if
the sentence comes with an indirect evidential reading.
Analysis. Our proposal consists of three components: (i) CP may carry an [EV] feature, an interpretable
feature that marks an indirect evidence-based proposition (see Kratzer 2016). While [EV] has no
phonological realization in Cantonese and Vietnamese embedded clauses, it may be realized overtly in
other languages, such as C-T agreement in Quechua (Faller 2002, Sánchez 2004). (ii) Attitude verbs in (2),
but not in (3), carry an uninterpretable counterpart [uEV], which agrees with the embedded CP. Again,
Cantonese and Vietnamese generally lack morphological inflection. However, in other morphologically
rich languages like Lithuanian, verbal inflection may be used to mark evidentiality (Gronemeyer 1997).
(iii) With this featural setup, we adopt the idea of “phase unlocking” advocated in Rackowski & Richards
(2005), which suggests that the locality condition imposed by CPs is obviated if a higher head first agrees
with a phase and then goes on to agree with a phase-internal element. Substantially, the [uEV] feature on
attitude verbs in (2) agrees with the [EV] feature on CP (=6a). This Agree relation ‘unlocks’ the CP phase.
Consequently, it enables a second Agree relation (e.g. for EPP or Case) between the attitude verb and the
embedded subject (=6b). A-movement across a CP boundary is achieved with no intermediate touchdown
in Spec CP, hence no Improper Movement. The derivation goes on as (6c) depicted.
(6) a. An attitude verb c-selects and agrees with a CP phase:
[vP v[uEV][EPP] [CP[EV] C [TP S VP ] ] ]
b. It further agrees with S and triggers movement:
[vP Si v[uEV][EPP] [CP[EV] C [TP ti VP ] ] ] (if there is no DP/pro in the numeration)
c. The S is further raised to the matrix Spec TP (= raising-to-subject constructions).
It follows that verbs in (3) and in English lack an [uEV] feature, and the step in (6b) is thus impossible
without the mediation of Spec CP and subsequent A-movement from which instantiates Improper
Movement. Note that if there is a DP or pro in the numeration, the EPP feature on the verb in (6b) can be
alternatively satisfied by Merge (rather than Move), deriving (5b).
Discussions. (i) Our proposal attributed some ‘exceptional’ raising behaviors to the lexical semantics of
predicates. That is, only indirect-evidence-based attitude verbs carry an [uEV] feature which enables
hyperraising. This echoes Wurmbrand’s (2019) recent claim that raising is more restricted than previously
thought. Besides phase, raising may also be constrained by topichood of the raised DP (Sener 2011 for
Turkish), predicative properties (Yoon 2007 for Korean; Horn 2008 for Japanese) and thematic
configuration of the matrix predicates (Wurmbrand 2018 for English). With cross-linguistic support, we
showed that evidentiality also plays a significant role in raising. (ii) This paper supported the view that
phases may be deactivated during derivation, by reporting a novel ‘unlocking’ effect of Agree relation on
[EV] features between verbs and their complement clauses. The finding is consistent with Rackowski &
Richards’ (2005) claim that Agree relation on Case between v and a CP helps explain the pattern of
multiple wh-movement in Tagalog, and with Halpert’s (2019) proposal that T phi-agrees with a CP in
Zulu, which then allows for hyperraising of an embedded subject. That phasehood is conditioned
dynamically provides a way to explain why locality conditions may be ‘selective’.
Selected references. Halpert, C. 2019. Raising, unphased. NLLT, 37(1), 123-165. Rackowski, A., &
Richards, N. 2005. ‘Phase edge and extraction: A Tagalog case study’. LI, 36(4), 565-599. Wurmbrand,
Susi. 2019. “Cross-clausal A-dependencies.” In Proceedings of the 54th CLS, 585–604.
Past Imperatives in Brazilian Portuguese
Rerisson Cavalcante (Universidade Federal da Bahia)
Higor Monteiro Paiva (Universidade Federal da Bahia)
We discuss the existence of imperative sentences with past interpretation in human languages and argue that Brazilian Portuguese (BP) matrix past subjunctive sentences (MPS’s) have such status (cf. (1)). Most authors consider imperatives inherently incompatible with the past, although they differ on explanations for this. Beukema and Coopmans (1989), assuming the absence of temporal distinctions and past morphology in data such as (2), consider that INFL is marked as [-Tense] in the imperatives, which also would explain the absence of subjects. Zanuttini (1991, 1996) sees imperatives as lacking TP, which would explain the incompatibility between true imperatives and pre-verbal negative markers (since NegP selects TP in her account). Platzack and Rosengren (1998) argue that imperatives lack FinP, whose absence results in the morphological poverty of imperative verbs and in the absence of TP and MoodP; they recognize that imperatives with future tense morphology do exist in some languages, but they do not consider it as evidence against their analysis, arguing that there is no language with tense oppositons in imperatives. On the other hand, other authors assume the existence of IP/TP in imperatives. Rivero (1994) and Han (1999, 2001) deal with the incompatibility between true imperatives and pre-verbal negation in terms of (syntactic or semantic) restrictions on I-to-C verbal movement. Other authors give one more step and consider temporal distinctions in imperatives as possible, but they still reject the possibility of past tense morphology/interpretation. Sapir (1912 [1909]) pointed out the existence of a present imperative (for orders to be carried on immediately) and a future imperative (related to a more distant time) in Takelma language (cf. PALMER, 2001: 151, 157-158 , 180; MITHUN, 1999: 153-154). And Roorick (1995: 340-341), against Beukema and Coopmans, rejects the idea that imperatives have no tense features, pointing out that Latin has a morphological distinction between present and future imperatives (cf. (3)), but he still assumes a ban on past morphology in imperatives. He proposes that imperatives are defined by a [unrealized] feature which allow present and future imperatives, but not past ones. BP MPS’s present a challenge for these analyses. Four cases of possible past imperatives in other languages are noteworthy. The first is that of French reference grammars (cf. GREVISSE; GOOSSE, 2008: 892), which record an impératif passé alongside with the regular present imperative (cf. (4)), but a closer look reveals this characterization to be mistaken, since the real distinction is aspectual, not temporal: impératif passé has a perfective reading regarding a point in the future, while present imperative has an imperfective reading. (This distinction is also possible in English through the use of the present perfect, cf. Jary and Kissine (2014: 262), in (5)). We assume that the impératif passé is not a real case of past imperative. Furthermore, BP MPS’s do not have this type of perfective aspectual behavior. The second and third cases come from English. Davies (1986: 165) shows that past interpretation is not available in present imperatives such as (6a), unless they are coordinated with a declarative such as (6b). And Jary and Kissine (2014) show that imperatives with present perfect can be linked to a time before the speech (cf. (7)). This use differs from the present imperative as to the state of the world to which it refers: in present imperatives, the corresponding declarative sentence is taken to be false; the imperative then establishes that such declarative should become true. But in (7) the (past) imperative cannot affect the state of the world. The veracity or falsity of the declarative sentence is already determined, but such value is unknown to the speaker. Therefore, data in (7) do not fall under Roocyck’s analysis of imperatives as [unrealized]. We consider this English structure in (7) to be a real type of past imperatives, but it does not correspond to the properties of the BP data in (1), in which the truth value of the corresponding declarative is false and it is already known to the speaker. The fourth case of potential past imperatives is the closest to the BP data. Duinhover (1995) treats Dutch sentences with auxiliary hebben (‘have) or zijn, (‘be’) in the past followed by a main verb in the participle (cf. (8)) as imperatives irrealis (irreële imperatief). Dutch irreële imperatief and BP MPS’s differ morphologically (perfective indicative morphology versus subjunctive imperfective morphology), but both are used in the same contexts, as a speaker’s complaint on a action that the listener was supposed to carry on in the past (cf. also BOSQUE 1980 on retrospective imperatives). Below, more properties of BP MPS’s: (i) just as
present imperatives, MPS’s display null subjects, differing from the tendency for overt subjects in declaratives/interrogatives; (ii) MPS’s fit Roorick’s characterization of imperatives as [unrealized] even though they are directed to the past; (iii) MPS’s are sensitive to the dynamic-static distinction: for instance, (1c) has a dynamic reading; (iv) however, MPS’s differ from present imperatives (and from declarative/interrogative clauses) in being marginal with the adverb nunca (‘never’) (cf. (9-10)); this is somehow surprising since MPS’s exhibit subjunctive verbs and nunca forces subjunctive (instead of indicative) in present imperatives. In declaratives/interrogatives, both indicative and subjuntive verbs are allowed with nunca. All of this raises the question: are MPS’s really imperatives? Or do they fall under other mood/sentential type? Not being real orders/commands is not an obstacle for them to be considered imperatives. Making commands is a prominent function of the imperatives, but not its only one. Imperatives also express a number of non-directive functions such as requests, permissions, good (and bad) wishes, conditionals, and other cases in which the participants have no control over the events. MPS’s also cannot be considered reduced conditional sentences, as they differ from the conditionals in allowing coordination (including coordinations with rationale interpretation) as in (11). This behavior is parallel to that of the present imperatives (cf. (12)). Additionally, the fact that MPS’s are formed with (past) subjunctive verbs is also not an argument against, but actually an argument for, an imperative status, since subjunctives are dependent clauses and do not freely occur in matrix sentences without a licensor. The only exception to this in the BP is precisely the case of present imperatives, which reinforces the parallel between present imperatives and MPS/past imperatives. Furthermore, the imperative category is best characterized as a sentential type (opposed to declarative, interrogative, and exclamative types) and not as a morphological type (opposed to indicative and subjunctive verbs), since there are several languages in which imperative sentences have a distinct syntax, but not a specific verbal morphology. Finally, we would like to make two suggestions for future research on the topic: (i) despite not being conditional sentences, BP past imperatives may have risen from a reanalysis/grammaticalization of conditional structures with I-to-C movement as in (13) (which became ungrammatical in the current spoken BP); (ii) past imperatives could be incorporated into a formal pragmatic account on imperatives based on To Do List (TDL) if we consider the function of imperatives to be not only to add properties to the listener’s TDL, but to perform several types of operations on the items of the TDL. The past imperatives clearly refer to items that were previously present in TDL.
(1) a. Ligasse antes (de vir)! / b. Não corresse (tanto)! / c. Não fosse tão idiota!
call.SUBJ.PAST before of come not run.SUBJ.PAST that-much not be.SUBJ.PAST such idiot
(2) a. *Went to London, please. / b. *Behaved yourselves, for God’s sake.
(3) a. amate (love.IMP.PRES.2PL) b. Amatote (love.IMP.FUT.2PL)
(4) a. Rentre à six heures! / b. Sois rentré avant six heures!
retourn.IMP.PRES. to-the six hours be.SUBJ.PRES retourned before six hours
(5) Have finished your homework before dinner time!
(6) a. *Turn up yesterday! b. Turn up yesterday and you’d have got a real shock.
(7) a. Please, don’t have made things worse. / b. Please, don’t have said anything rude!
(8) a. Had gebeld! (lit.: ‘Tinha chamado!”) / b. Was gekomen! (lit.: ‘Estava/era lavado!’)
(9) a. Nunca corra muito na estrada! / b. Nunca ligue para mim! (present imperatives)
never run.SUBJ.PRES much in.the road never call.SUBJ.PRES to me
(10) a. #Nunca corresse tanto! / b. #Nunca ligasse antes de vir! (past imperatives)
never run.SUBJ.PAST much in.the street never call.PRES to me
(11) Ligasse antes... (a) ... e/que eu lhe avisava. (b) ... porque às vezes eu trabalho à noite.
call.SUBJ.PAST before... and/that I you let know because sometimes I work at night
(12) Ligue antes... (a) e/que eu lhe aviso (b) ... porque às vezes eu trabalho à noite.
call.IMP.PRES before... and/that I you let-know. because sometimes I work at night
(13) a. Tivesse ele se esforçado mais, teria conseguido.
had.SUBJ he himself tried.harder more, would-have succeeded
b. [CP [C’ tivesse [TP ele [T’ tivesse [vP ele [VP se esforçado mais ... ]
The syntax/semantics of noun phrase in a language with no DP:
an analysis of Karitiana
Luiz Fernando Ferreira (Universidade de São Paulo)
Karin Vivanco (Universidade Estadual de Campinas)
Maria Del Mar Bassa Vanrell (Wellesley College)
This work investigates the syntactic structure of noun phrases in an articleless language called Karitiana (Tupi Branch, Arikém Family) and its impacts for the semantic derivation at LF. NPs in this language are always bare and can have a definite, indefinite or generic reading as illustrated in (01) below.
(01) ‘y kinda’o
eat fruit
‘to eat fruits/a fruit/the fruit’
The works which investigated NPs in Karitiana (Müller, 2011; Müller & Sanchez-Mendes, 2016 among others) consider that the lack of articles is a strong evidence that this language does not have a DP and that NPs are all predicates. Nevertheless, those proposals do not clarify how the composition with NPs that are predicates work without DPs since verbs do not take predicates as arguments. Müller & Sanchez-Mendes (2016) suggest that a Heimiam approach to NPs might be a solution, but they do not explore this and leave it for future research.
So, our goal is to look at some of the solutions available at the literature of how NPs compose with verbs when there is no DP and also try to deriving the three readings that NPs can have exemplified in (01). One could assume that: (i) NPs are generalized quantifiers and type <<e,t>,t> following Montague (1974); (ii) the language has a null DP, (iii) NPs are kinds (type e) following Chierchia (1998); (iv) type-shifting operations following Partee (2002), (v) NPs introduce a variable and a predicate and their semantic type is t following Heim (1982). All truth conditions of a sentence can be derived by all those proposals can derive the readings in (01), so our choice is based either on economic or on syntactic reasons.
The generic reading could be derived assuming a phonologically null universal quantifier that binds the variable x in Montague and Heim’s proposals or assuming the nominalization type-shif which turns predicates into kinds (𝑃 →𝑃∩) in Chierchia, Partee and null DP’s proposals.
In the generalized quantifier approach (Montague, 1972), the definite/indefinite readings would be derived through different lexical entries. For a word like boroja (‘snake’), the definite lexical entry would be λP<e,t>.[∃𝑥[∀𝑦[𝑦 𝑖𝑠 𝑎 𝑠𝑛𝑎𝑘𝑒 ↔𝑦=𝑥] & 𝑃(𝑥)]] and the indefinite would have the entry λP<e,t>.[∃𝑥[𝑥 𝑖𝑠 𝑎 𝑠𝑛𝑎𝑘𝑒 & 𝑃(𝑥)]]. The problem with this analysis would be that it is not economical since all common nouns would be ambiguous having different lexical entries for the definite or indefinite reading.
One could argue that for two null DPs in syntax: one definite and another indefinite. This would also derive the meanings in (01). However, we argue that a null DP is not the right solution based on adjective extraction out of NPs (Boskovic, 2009). Boskovic (2009) proposes that DPs are phases that block the movement of adjectives out of it. Thus, if a language does not have DP, extraction of adjectives is allowed. Karitiana allows adjective extraction as illustrated below (see Vivanco 2018). Thus, we argue that any account that involves a null DP is not correct for Karitiana.
(02) a. ‘eemm1 ∅-na-m-potpora-j [ese t1]
dirty 3-decl-caus-boil-fut [water t1]
‘The man will boil the dirty water’
b. [Morã horot] i-amy pykyp-y-ty Karin?
qu comp 3-buy clothes-epen-obl Karin
“Like what did Karin buy the clothe?"
Chierchia (1998) assumes that bare nouns in English do not have a DP and denote kinds (type e). This cannot be the solution for Karitiana since it predicts that if all NPs are mass and that the language should have a classifier system, which is not the case.
Another possibility to derive the meanings in (01) is through type-shift operations proposed by Partee (Partee, 2002). The definite meaning in (01) could be derived using the iota type-shift (P → 𝜄𝑥[𝑃(𝑥)]), whereas the indefinite meaning could be derived using the a type-shift (𝜆𝑃[𝜆𝑄[∃𝑥[𝑄(𝑥)& 𝑃(𝑥)]]]). The problem with the type-shift operations would be too similar to a null DP analysis since: (i) they would occur at the same position as DP, between the VP and its arguments; and (ii) they would have the same semantics as DPs.
For last, one could assume a Heimiam analysis in which NPs are predicates and determiners introduce a variable and the semantic type of DPs is t. This is what Salles & Matthewson (2016) assume for Pirahã. If the role of the DP is introducing a variable, what introduces the variable in a language with no DP? Müller & Sanchez-Mendes (2016) suggest the Heimiam approach for NPs in Karitiana, but they leave open the matter if what introduces the variable.
They suggest that this might be done through a null DP, but we have already seen that there is no null DP in Karitiana. Thus, the variable in Karitiana would be introduced by the NP itself. Therefore, NPs are not predicates as originally assumed but they are type t. The NP moves out of the sentence and the verb composes which the trace left by the NP which is type e. After that, coordination rule composes the NP and the sentence which are both type t. The difference between definites and indefinites relies on a pragmatic condition according to which indefinites introduce a variable which must be new whereas definites introduce a variable that must already be salient in the context. Thus, the definite/indefinite distinction is possible in English because the D node makes it possible to introduce a variable in different ways. In Karitiana, variables are introduced by NPs. Thus, there is only one way the variable can be introduced and the definite/indefinite distinction does not arise in this language.
Therefore, the best options to describe the phenomenon of bare nouns phrases in Karitiana are proposals (iv) and (v), and the choice between theses may rely on economy factors.
Selected References
BOSKOVIC, Z. (2009). More on the no-dp analysis of article-less languages. Studia linguistica, 63(2), 187-203.
CHIERCHIA, G. (1998). Reference to kinds across language. Natural language semantics, 6(4), 339-405.
HEIM, I. (1982). The semantics of definite and indefinite noun phrases. PhD Dissertation. University of Massachusetts, Amherst.
MÜLLER, A. (2011). On the encoding of the definite/indefinite distinction in Karitiana. In Proceedings of sinn und bedeutung.
MONTAGUE, R. 1974. Formal Philosophy. Selected papers of Richard Montague, edited and with an introduction by Richmond H. Thomason. New Heaven, Conn.: Yale University Press.
MÜLLER, A. & SANCHEZ-MENDES, L. (2016). The semantics of bare nouns in Karitiana. O sintagma nominal em línguas amazônicas, 241-261.
PARTEE, B. (2002). Noun phrase interpretation and type shifting principles. Formal semantics. The essential readings. Oxford (UK), Malden (USA), p. 352-381.
GUTIÉRREZ, R. & MATTHEWSON, L. (2016). Existentials and (in)definiteness in Pirahã. SULA 9: Proceedings of the Ninth Conference on the Semantics of Under-Represented Languages in the Americas.
VIVANCO, K. (2018). Perguntas Qu-, Orações subordinadas e ordem de palavras em Karitiana. PhD dissertation. University of São Paulo, São Paulo.
DIFFERENTIAL OBJECT MARKING IN TUKANO
Bráulio Lopes (Universidade Federal de Minas Gerais)
Fábio Duarte (Universidade Federal de Minas Gerais)
INTRODUCTION: Tukano is a language spoken mainly in the region of the Uaup.s river basin,
located on the border between Brazil and Colombia. According to statistical findings, the Tukano
ethnic group has total population of 8167 people, 2016 of which inhabit Colombian territory
(DANE, 2005) and 6151 inhabit Brazilian territory (IBGE, 2010). The main objective of this
presentation is to investigate the phenomenon of Differential Object Marking (DOM) in Tukano. By
differential object marking we mean that Case systems may serve as a grammatical device to
encode certain semantic differences. In line with this, we follow Comrie’s (1981) and Croft’s (1988;
1990) assumptions that specificity, animacy and person-number features play a major role regarding
the activation of differential object marking across languages. Within the typological literature
(Givón 1976; Comrie 1981; Croft 1988; 1990; Bentley 1994), it has been assumed that the relevant
semantic features that trigger DOM are the ones that occupy a higher position in the hierarchies
below.
Definiteness Hierarchy:
definite > specific > indefinite > non-specific
Animacy Hierarchy:
human > animate > inanimate
The analysis presented here aims to show how DOM is connected to the scale of definiteness in
Tukano, as was shown in other languages of the same family (STENZEL, 2008). To support such
analysis, different contexts of occurrence of the suffix responsible for object marking, namely {-re},
will be taken into account. In Tukano, it is observed that DOM is triggered whenever the direct
object is definite, as follows:
(1) wi'i-r. we.-' (2) wi'. we.-'
house-OBJ build-PRES.VIS.1 house build-PRES.VIS.1
'(I) build the house' '(I) build house (I'm a house-builder)'
(3) numi. sĩ'i-r. ɨm.tihis. wa're-.mo
woman DEM.AN.MSG-OBJ perfume apply-REC.PAST.VIS.3FSG
'The woman applied perfume on that one'
THE PROPOSAL: Based on the above data, one may conclude that the suffix {-re} realizes a structural case, particularly when the object is high in the definiteness hierarchy. In this sense, there is a direct correlation between definiteness and the occurrence of the Case marker {-re} on D/NPs.
Thus, the absence of this suffix encodes that the object is indefinite. Based on this, we propose that the Case marker {-re} on DPs is structurally obligatory if the object is definite/specific. It is, however, forbidden if the object is indefinite/non- specific. In double object construction, the Goal receives dative when the theme and the goal are in the same local domain.
FINAL REMARKS: As a conclusion, we assume that the {-re} marker can then be interpreted as a clear example of Case syncretism in the sense that languages tend to use the same morpheme for encoding different grammatical relations. In double object constructions, the Case marker {-re} can only appear on the higher argument, usually the argument that receives the GOAL/BENEFICIARY theta role. In such contexts, the THEME/PATIENT tends to remain unmarked regardless of whether it is definite or indefinite. In terms of linear order, the Goal systematically precedes the direct object, thereby emerging the following word order.
Keywords: Tukano. Differential Object Marking. Definiteness.
References
AISSEN, Judith. Differential object marking: iconicity vs. Economy. In: Natural Language & Linguistica Theory, vol. 21, n.3, p 435-483, 2003.
COMRIE, Bernard. Language Universals and Linguistic Typology. Chicago: University of Chicago Press. 1989. 2. ed.
DANE, Censo Nacional de Poblacion de 2005. Dispon.vel em <https://www.dane.gov.co/index.php/estadisticas-por-tema/demografia-y-poblacion/censo-general-
2005-1>. Acesso em 12/12/2018.
IBGE, Censo demográfico 2010. Dispon.vel em: <https://indigenas.ibge.gov.br/images/pdf/indigenas/folder_indigenas_web.pdf>. Acesso em
12/12/2018.
RAMIREZ, Henri. A Fala Tukano dos Ye'pâ-Masa: Tomo I Gram.tica. Manaus: CEDEM. 1997.
STENZEL, Kristine. Koitiria 'differential object marking' in cross-linguistic perspective.
Amerindia. Rio de Janeiro: Museu Nacional. 2008, n. 32, p. 153-181.
THE SEMANTICS OF THE SIMPLE AND PROGRESSIVE
PRESENTS IN BRAZILIAN PORTUGUESE
Ana Muller (University of São Paulo)
Focus. This paper focusses on the semantics of the Simple (1a,2a) and Progressive (1b,2b) Present tense inflexions in Brazilian Portuguese (BP) in event-denoting matrix sentences.
Background. Both inflections yield SIMultaneous and HABitual readings, albeit SIM-readings of the Simple Present (SPres) have a special flavor, and have been called telescopic (SMITH 1991). SIM-readings describe events as co-occurring with Speech Time (1a,1b); whereas HAB-readings describe (potential or actual) patterns of iterated events (2a-2b) (SMITH 1991; BONEH & DORON 2008) within an interval that includes Speech Time. Based on Klein (1994), we assume that tense denotes a relation between Speech Time (SpT) and Topic Time (TopT), and that aspect denotes a relation between Event Time (EvT) and Topic Time.
(1) a. Pelé corre com a bola (in a soccer game broadcast).
‘Pelé runs with the ball.’
b. Neste momento, Pelé está correndo com a bola.
‘At this very moment, Pelé is running with the ball.’
(2) a. Pelé corre todos os dias.
‘Pelé runs every day.’
b. Ultimamente, Pelé está correndo todos os dias.
‘Lately, Pelé has been running every day.’
Claims. The difference between the Simple and the Progressive Presents lies in aspect. The (telescopic) SIM-readings of the SPres are PERFective; whereas SIM-readings of the ProgPres are IMPERFECtive (SMITH1991). This means that, albeit both morphologies demand that TopT be identical to SpT, the aspectual relation between their TopTs and EvTs differs. For SIM-readings the SPres demands EvT to be identical to TopT (PERFV); whereas the ProgPres demands EvT to include in TopT (IMPV), as evidenced by their respective entailments in (3).
(3) Neste exato momento, Madonna está cantando/canta. TopT=neste exato momento
‘At this very moment, Madonna is singing/sings.’
⇨ ProgPres: Madonna has to already be singing before SpT.
⇨ SPres: Madonna does not have to be singing before SpT.
Note that SIM-readings are singular, they refer to an individual event, which is individuated by its running time. HAB-readings, on the other hand, are plural. Many authors claim that HABituals involve iteration (FERREIRA2016; BONEH&DORON2008, among others). Nevertheless, they do not make any claims regarding the aspectual nature of the events that are iterated. Our claim is that HAB-readings of both forms result from the iteration of their respective SIM-readings. Thus, although both HABituals are IMPerfectiVe, SPres- HABs denote plural events made out of PERFectiVe subevents; whereas ProgPres- HABs denote plural events made out of IMPerfectiVe subevents. Consequently, the difference between SPres and ProgPres HABituals lies in the kind of events they iterate.
This may be illustrated by the well-known fact that, in a narrative, a sequence of sentences marked with PERFectiVe aspect is interpreted as a sequence of events; whereas, if marked for IMPerfectiVe, the events are interpreted as co-occurring. The perfectivity of SPres hab-sentences and the imperfectivity of ProgPres sentences are illustrated in (4-5). (4a) with the SPres is fine; but (4b) with the ProgPres is awkward. On the other hand, the sequence of sentences in (5a), with the SPres, expresses a sequence of events; whereas the sequence of sentences in (5b), with the ProgP, expresses co-occuring events.
(4) a. Todos os dias depois que João toma banho, Maria canta.
b. #??Todos os dias depois que João está tomando banho, Maria está cantando.
‘Every day after João takes a shower, Maria sings.’
(5) a. João toma banho. Maria canta.
‘João takes a shower. Maria sings.’
b. João está tomando banho. Maria está cantando.
‘João is taking a shower. Maria is singing.’
The properties of the SPres and ProgPres inflections and their corresponding readings are summarized in the Table below.
Another evidence for our claim that HAB-readings are generated by the iteration of the corresponding SPres and ProgPres SIM-predicates is the entailments of both forms of sentence (6).
(6) Ultimamente, às 6:00h da tarde, Madonna está correndo/corre no clube.
‘Lately, at 6:00 p.m., Madonna is running/runs at the club’.
⇨ ProgP: Madonna starts running at the club sometime before 6:00 p.m..’
⇨ SP: Madonna starts running at the club at 6:00 p.m..’
In (7-8), we sketch logical forms for Madonna está correndo/corre no clube, in order to make our claims about their aspect more transparent.
(6) Prog-SIM: ∃e [running (e) & at the club (e) & τ(e)⊃ at this very moment]
SPres-SIM: ∃e [running (e) & at the club (e) & τ(e)= at this very moment]
(7) Prog-HAB: ∃E [E= Σe1…en & running (ex) & at the club (ex) & τ(e) ⊃ TopT(ex)]
SPresent-HAB: ∃E [E= Σe1…en & running (ex) & at the club (ex) & τ(ex) = TopT(ex)]
Where: e: singular event; E: plural event; τ(e): the running time of e.
References.
BONEH, N. & E. DORON, 2008. Habituality and the Habitual Aspect. In: ROTHSTEIN, S. (Ed.). Theoretical and Crosslinguistic Approaches to the Semantics of Aspect. Amsterdam: John Benjamins, 2008.
COMRIE, B. 1985. Tense. Cambridge: University Press,1985.
FERREIRA, M. 2016. The semantic ingredients of imperfectivity in progressives, habituals, and counterfactuals. Natural Language Semantics.
KLEIN, W. 1994. Time in language. London: Routledge, 1994.
PARSONS, T. 1990. Events in the semantics of English: a study in subatomic semantics. Cambridge, Mass.: MIT Press.
SMITH, C. 1991. The parameter of Aspect. 2. ed. Dordrecht: Kluwer Academic Publishers.
What is happening to clitics in Brazilian Portuguese?
Janayna Carvalho (Universidade Federal de Minas Gerais)
Ana Regina Calindro (Universidade Federal do Rio de Janeiro)
1. Aims: Brazilian Portuguese (BP) shows an asymmetry between the distribution of 1st and 2nd person clitics (me/te) and 3rd person clitics (o/lhe/se). While the latter are no longer licensed in accusative and dative positions (o/lhe), 1st and 2nd person clitics still are (me/te).
Given this scenario, the aim of this paper is twofold. Firstly, we demonstrate that, compared to 1st and 2nd person clitics, 3rd person clitics have a fundamentally different featural constitution, which accounts for their ongoing disappearance. Secondly, we explore the fact that 3rd person clitics, usually se (in reflexive sentences) and lhe, has been reanalyzed as 2nd person in some varieties of BP. We suggest that the syncretism between 2nd and 3rd person clitics (lhe/se) is a byproduct of the loss of the features that distinguish 3rd person from 1st and 2nd.
2. Relevant Data: The pronominal system in BP has been undergoing several changes since the 18th century (cf. Cyrino 1994 et seq., Berlinck 1996 et seq., Galves, Torres Morais & Ribeiro 2005, a.m.o.). Accusative 3rd person clitics o/a(s) and its variants were the first ones to disappear (cf. 1, adapted from Kato et al. 2009:249). All sentences in (1) are well-suited for answering the question ‘Did you see Pedro yesterday?’. The answer in (1a) exemplifies the clitic form, (1b) the null object variant and (1c) overt full pronouns:
(1) a. Vi-o na biblioteca. (19th century)
(I) saw-CL.3rd in.the library
b. Vi Ø na biblioteca (20th century)
(I) saw – Ø in.the library
c. Vi ele na biblioteca (20th century)
(I) saw he in.the library
‘I saw him in the library’
BP has also lost the 3rd person dative lhe(s) (cf. 2a) since the 19th century, replacing it by other strategies as the prepositions a/para followed by full pronouns (cf. 2b):
(2) a. A Maria enviou uma carta ao João/enviou-lhe uma carta.
the Maria sent a letter a/to João/sent-CL.3rd.DAT a letter
b. Maria enviou uma carta para/ao João/ele.
Maria sent a letter a/to.the João.OBL/him
Finally, se - another 3rd person clitic - is being progressively lost in BP. This is a multifunctional element, used in impersonals, passives, middles, anticausatives and (inherent) reflexives. Putting (inherent) reflexives aside, all other aforementioned sentence types have an alternative construal without the clitic:
(3) a. No Brasil (se) vende caipirinha. (impersonal)
in.the Brazil (CL. 3rd) sell caipirinha
‘One sells caipirinha in Brazil.’
b. Roupa de algodão (se) lava facilmente. (middle)
cloth of cotton (CL.3rd) wash easily
‘Cotton clothes wash easily.’
c. A porta (se) fechou. (anticausative)
the door (CL.3rd) closed.
‘The door closed.’
3. Loss of 3rd person clitics mirrors a hierarchical featural structure: Harley & Ritter (2002) observe that crosslinguistically gender is found in 3rd but not in 1st and 2nd person pronouns, which suggests an implicational scale in phi-features bundling, as shown in (4). In this representation, 1st and 2nd person are subordinated to a participant node. 1st person realizes the ‘speaker’ feature; 2nd person realizes the ‘adressee’ feature. All features related to 3rd person are under an individuation node (‘group’, ‘minimal’, ‘class’ and the nodes they dominate). See the representation from Harley and Ritter (2002:486) below:
The hierarchy of individuation on (4) mirrors the losses of 3rd person clitics in BP, as described in section 2 (accusative clitic in 18th century; dative clitic in 19th century). The accusative clitic was the first one to disappear since it has gender features (o-CL.MASC/a-CL.FEM). The dative clitic was the next, because of its specification for animate entities. In colloquial BP, the multifunctional se is still available in several environments for most speakers, which shows that only a clitic non-specified for class features can survive in BP. The representation in (4) suggests that the (ongoing) loss of 3rd person clitics in BP is due to the loss of phi-features that are low in an individuation scale. In this paper, we assume only the features proposed for 3rd person by Harley and Ritter (2002). The syncretism between 2nd and 3rd person requires other assumptions about pronominal constituency, which we briefly discuss in section 4.
4. Consequences of the loss of 3rd person clitics: A consequence of the loss of 3rd person clitics is the syncretism between 2nd and 3rd person clitics in BP. With the loss of the distinctive features in 3rd person clitics, they become similar to 2nd and 1st person clitics, in the sense that they only refer to participants. However, the hierarchy proposed in (4), which rightly accounts for 3rd person clitics features, is not adequate for the massive syncretism between 2nd and 3rd person in BP. Under a nanosyntactic approach, Wyngaerd (2018) assumes that 1st, 2nd and 3rd person pronouns form a sole structure, with 3rd person being the most embedded node, followed by the 2nd person node. Each node contributes a feature: 1=speaker, 2= participant, 3= person. This means that 1st, 2nd and 3rd person pronouns stay in a containment relationship: if a 2nd pronoun is assembled in the syntax, it obligatorily contain features of the 3rd person pronoun. Hence, a second person pronoun has the features ‘participant’ and ‘person’. Likewise, a 1st person pronoun has ‘speaker’, ‘participant’ and ‘person’ features. The containment relationship readily accounts for the syncretisms between pronouns in the languages of the world: syncretism happens when pronouns share features (person, in this case) and do not have conflicting features such as gender, which BP lost for 3rd person.
Selected References:
BERLINCK, R. The portuguese dative. The dative, 1996.CYRINO, S. O objeto nulo no português do Brasil: um estudo sintático-diacrônico.Tese de Doutorado, 2004. GALVES, C.; MORAES, M. Aparecida T.; RIBEIRO, I. Syntax and morphology in the placement of clitics in European and Brazilian Portuguese. Journal of Portuguese Linguistics, 2005.HARLEY, H.; RITTER, E. Person and number in pronouns: a feature-geometric analysis. Language, 2002. KATO et al. Brazilian Portuguese and the recovery of lost clitics through schooling. In: PIRES, A.; ROTHMAN, J. (Ed.). Minimalist inquiries into child and adult language acquisition, 2009. WYNGAERD, G. The feature structure of pronouns: A probe into multidimensional paradigms. Exploring nanosyntax, 2018.
The perfect ingredient for a pancake mix: agreement failure
Janayna Carvalho (Universidade Federal de Minas Gerais)
Artemis Alexiadou (Humboldt University/ZAS)
1.Pancake sentences: Pancake sentences have two core characteristics: (i) they are copular constructions in which an adjective surfaces in its unmarked form for gender and number; (ii) the noun in subject position receives a special reading, sometimes referred to as ‘denotation widening’. (1), from Brazilian Portuguese, and (2), from Hebrew, exemplify two alternative modes to generate pancake sentences, which we call, in this order, type 1 and type 2 pancake sentences. In (1), the nominal element is in subject position (see Rodrigues and Foltran 2014; de Conto 2018); (2) exemplifies languages of the second type, where the nominal is dislocated to a peripheral position (Subj2, according to Roy & Schlonsky 2019), and the pronoun ze is in Spec,TP.
(1) Crianças é divertido. Brazilian Portuguese
Child-FEM-PL is-SG fun-SG. (Rodrigues and Foltran 2014)
(2) yeladim ze kaSe Hebrew
children Z hard (Shirtz 2014:77)
“(Having/raising) children is hard.”
2. Proposal: In this work, we explore properties of agreement in pancake sentences, exemplified in (1) and (2). We argue that in type 1 pancakes, the adjective selects for a covert proposition, which corresponds to a CP (see tests in (3) and (4)). With the selection of a CP, the adjective has nothing to agree with and surfaces with its default features, indicating agreement failure (cf. Preminger 2014). In type 2 pancakes, a demonstrative pronoun, with default phi-features, merges with the adjective.
In both types, the interpretation of a hidden event has a syntactic correlate: the presence of an element (a covert clause or demonstrative pronoun) precluding the agreement between the adjective and the noun. Since the covert clause (in type 1) and the demonstrative pronoun (in type 2) have default features, the adjective invariably surfaces with its default features too.
3. Type 1 languages: Based on tests in Dikken, Larson and Ludlow (2019), we show that type 1 pancake sentences have a covert proposition. That is, the nominal mulher in (3) is part of a clause. Sentence (3) has two readings, as demonstrated below.
(3) Mulher é perigoso amanhã.
woman is dangerous tomorrow
Reading 1: [[eventualities involving woman tomorrow] is dangerous].
Reading 2: [[eventualities involving woman] is dangerous tomorrow].
The adverbial tomorrow cannot directly modify a noun, therefore the ambiguity in (3) indicates that the complement of this adjective is not the noun mulher, but rather a clause. The second piece of evidence comes from the licensing of two temporally inconsistent adverbial modifiers, each of them modifying different clauses. Consider the following context in which (4) is acceptable : “ we thought that inviting/having women a week ago would be dangerous in our yesterday’s party, but it proved to be wrong.”
(4) há uma semana mulher era perigoso na festa ontem.
ago one week woman was dangerous in.the party yesterday
‘Inviting a woman a week ago was dangerous at the party yesterday.’
Josefsson (2014) discusses some properties of pancake sentences that also corroborates a hidden structure in Swedish as well, although her implementation differs from ours. Being a V2 language, Swedish only allows one argument to precede the verb. However, in (5), two constituents Två älskare (‘two lovers’) and varje kväll (‘each night’) apparently precede it.
(5) [Två älskare] [varje kväll] är omoralisk-t
two lover -PL each night be.PRS immoral-Neut-SG
“To have two lovers each night is immoral.” (Josefsson 2014: 66)
This receives a natural account if these two constituents are embedded in a larger clause, as discussed by the author. The covert clause is the constituent that really precedes the verb in (5). The constituent ‘varje kväll’ (each night) is modifying this covert clause, and does not count as a separate constituent. On its turn, the constituent ‘Två älskare’ (two lovers) is the complement of a null verb in this clause.
Note that (5) is parallel to BP data in (3): a temporal adverbial adjacent to the noun modifies a covert clause. This indicates, once more, that there is a covert full clause (with a TP node to which the adverbial phrase can adjoin to) in type 1 pancakes. Given these facts, we assume that languages with type 1 pancakes select for a hidden full clause - with all projections found in overt full clauses (CP, TP, VP and the like) – complementing the adjectives.
Full clauses are phases (Chomsky 2001), hence the adjective cannot cross C to agree with the nominal inside this clause as schematized in (6). That is the reason why the adjective does not agree in gender with the noun in pancake sentences, but does so in regular copular sentences shown in (7), in which a sole DP is present.
(6) é perigoso [CP [TP[VP.... [DP mulher]]]] (7) é perigosa [DP mulher]
is dangerous-MASC woman-FEM is dangerous-FEM woman-FEM
4. Type 2 pancakes: In type 2 pancakes, the adjective does not agree with the noun either. It is merged in a small clause with a demonstrative pronoun (see (2)). The adjective can only target the demonstrative pronoun to agree, since the noun is in a peripheral position (cf. Roy & Schlonsky 2019). In this paper, we take Hebrew to be representative of type 2 pancakes.
The agreement between the adjective and the pronoun fails in Hebrew because the pronoun has default gender features. Spector-Shirtz (2014) notes that ze does not have phi-features. Its inflected version (ele) is ungrammatical in pancake sentences.
(8)Sney orxim ze /*ele meacben. (Hebrew)
two guests ZMASC-SG ZMASC-PL. irritating
“(Having) two guests is irritating.”
4. Type 1 vs. type 2 pancakes: We have seen that the adjective surfaces with its default features in type 1 and type 2 pancakes. However, a key difference between them stems from the elements used to generate the denotation widening reading in these sentences. In type 2 languages, it suffices that a noun in a peripheral position enters an aboutness reading with the demonstrative pronoun in subject position. As a consequence, a specific type of adjective is not mandatory to generate a denotation widening reading in this case. In type 1 languages, covert propositions must be selected by adjectives for this reading to arise. Consequently, a sentence like (9) can only exist in type 2 languages, in which the interaction between the DP in the peripheral position and the pronoun suffices for generating this reading.
(9) ha-magavot ze ba-ambatiya
the-towels Z in-the-bathroom
‘The towels’ location is in the bathroom.’
5. Conclusion: A crucial component of pancake sentences is the ban on agreement between a noun and, generally, an adjective. In languages with type 1 pancakes, the noun is embedded in a covert CP selected by the adjective. Given the phasal status of CPs, the adjective cannot agree with such noun. In languages with type 2 pancakes, the adjective or other elements (like a PP, see (9)) does not agree with the noun either. The adjective forms a small clause with a pronoun with default features, and the noun is merged in a projection above TP. In this case, the denotation widening reading is obtained because of the noun position.
Our account differs from others that attribute the lack of agreement in pancake sentences to special properties of nominal elements (cf. Enger 2004). Crucially, the agreement mismatch stems from the fact that the adjective cannot agree with the noun in both types of pancake sentences.
Selected references: De Conto, Luana. (2018) Tese é complicado: a leitura de situação em sentenças copulares com concordância não marcada. Ph. D. dissertation. ● Dikken, M. et al. (2018) Intensional transitive verbs and abstract clausal complementation. In: Non-Propositional Intentionality. ● Josefsson, G. (2014). Pancake sentences and the semanticization of formal gender. Language Sciences. ● Rodrigues, P., & Foltran, M. J. (2014). Concordância em construções copulares do português brasileiro. Estudos Linguísticos ●Spector Shirtz, I. (2014). The syntax of non-verbal predication in Modern Hebrew: Predicate nominals, pseudoclefts and clefts. Ph. D dissertation.
The underlying identity of passives, impersonals, anticausatives, middles and deponents: an unified approach to Latin’s passive syncretism under Distributed Morphology
Lydsson Agostinho Gonçalves (Universidade Federal de Juiz de Fora)
Paula Roberta Gabbai Armelin (Universidade Federal de Juiz de Fora)
This study offers an analysis of the passive syncretism in Latin. The system proposed is developed under the Distributed Morphology framework (HALLE & MARANTZ, 1993; MARANTZ, 1997; and much subsequent work) and built upon the previous works of Schäfer (2008) and Lazzarini-Cyrino (2015). In Latin, the same morphology that constitutes the canonical passive voice is also found in other structures, as exemplified below:
(1) Different uses of passive morphology in Latin:
a. Primus locus (...) auo dicatur. (Seneca, Thyestes, 717-718) (passive)
“The first place (...) is dedicated to his grandfather.”
b. Dies noctesque estur, bibitur. (Plautus, Mostellaria, 235) (impersonal passive)
“Day and night, people eat and drink.”
c. Completur caede inter castra murosque uacui. (Tacitus, Historiae, III.29) (anticausative)
“The space between the camp and the walls becomes filled with death.”
d. Impiger occultis Poenus postquam abditur (...) (Silius Italicus, Punica, XV.488) (middle)
“After the fearless Poenus hides (...)”
e. Animus aetas uirtus uostra me hortantur. (Sallust, The Conspiracy of Catiline, 58, 119) (deponent)
“Your will, your age, your virtue urge me.”
That phenomenon is not exclusive to Latin, since similar paradigms are also found in several other languages, and occasionally involve the reflexive domain as well. In Brazilian Portuguese, for example, the clitic se displays a similar distribution. Approaches to these data regularly resort to the effects produced by different Voice heads (EMBICK, 2004; ALEXIADOU, 2013), but those approaches don’t properly predict why some languages are completely syncretic among all of those domains and why some are only partial, since the syncretic marker is seen as the exponent of one single functional category, Voice. Schäfer (2008) and Lazzarini-Cyrino (2015), however, propose that the syncretic marker found in those contexts is actually a syntactic argument of the verb, taking into account the fact that this element is always in complementary distribution with the internal or the external argument. Our proposal to account for the passive syncretism in Latin has this line of analysis as a starting point.
More specifically, we argue that the syncretic morpheme -r in Latin is a non-referential argument, that is, an anaphor. Anaphors are defined by Heinat (2006) as φ-defective DPs, which must find a referent in the structure in order to value them. Consequently, in the system we propose, when there is an anaphor in the internal argument position, it can be bound to the external argument and have its φ-features valued, which also creates an identity among the two elements. If the syncretic element is in the external argument position, however, there’s no other DP for it to be bound with; as a result, it cannot receive a referential interpretation in LF and it is read as an expletive.
Empirically, in Latin, the syncretic -r suffix is in complementary distribution with the external argument, which means it is originally introduced in that position. Following Lazzarini-Cyrino (2015), we propose that the introduction of an anaphor in that position is a device for fulfilling the syntactic needs of a transitive Voice head; when there’s no proper DP to occupy that position, the anaphor is introduced as a default in order to save the derivation, since it doesn’t add semantic content in that configuration.
Taking that into consideration, the basic derivation for all syncretic contexts in Latin will be essentially the same: an anaphor is introduced as the specifier of Voice and, since it cannot find a referent, it cannot be interpreted, becoming an expletive. Something must explain, however, why it is realized as an affix in the language. Lazzarini-Cyrino (2015) observes that this kind of syncretism always manifests itself in morphologically dependent elements, attributing this to a problem in case attribution. Since the anaphor is a DP, it has to be case-marked. However, since it doesn’t have φ-features, it cannot enter an Agree relationship with T and, as a consequence, it doesn’t get its case feature valued. That derivation can only go forward if the problematic element is moved out of a case position – that is, by incorporating it into the verb. Incorporation is usually barred for external arguments, but Levin (2015) shows that it is possible at PF, by means of local dislocation, with the only condition being that the incorporated DP is linearly adjacent to verb. Levin (2015) also argues that this is a mechanism for licensing DPs without case – if by some reason a DP remains without case in the syntax, PF fixes it by means of this operation. Following that approach, we propose that the marker incorporates into the T head above it at the PF branch.
Moreover, we propose that the different interpretations these structures have may be attributed not to the status of the syncretic marker itself, but to the structure as a whole, using a system based on different possible flavors of Voice (FOLLI & HARLEY, 2005): an anaphor occupying the specifier position of a DO Voice yields the passive/ impersonal interpretation, while the same configuration in a CAUSE Voice yields an anticausative. For middle contexts, we propose that there is an additional Appl head (PYKKKÄNEN, 2008) in the structure, which introduces the affected argument that will become the syntactic subject, while the external argument position is again filled by the default anaphor. Deponents are divided into three types: agentive deponents are reinterpreted as middle (ALEXIADOU, 2013), being derived in the same manner; experiencer deponents have a similar configuration, but, following Pesetsky’s (1996) considerations about experiencer arguments, we propose an EXP head in place of the Appl head; finally, unnacusative deponents have no Voice head, and the anaphor is in the complement position of the Appl or EXP head, from where it incorporates.
Our proposal correctly predicts the contexts in which the syncretic passive morpheme appear in Latin, while unifying its syntactic functioning. Also, our system dispenses with the need for a Passive Voice head, since it does not limit the marker to “non-active” contexts. Moreover, the system is able to derive the deponent verbs more straightforwardly. Finally, the different range of contexts in which the syncretic marker may appear in is reduced to the specification of the Vocabulary Item: any given language may have a more specified item for some configuration, preventing the syncretism from reaching all contexts.
SELECTED REFERENCES
ALEXIADOU, A. Where is non active morphology. In: Proceedings of the 20th conference on Head driven phrase structure grammar, p. 244-262, 2013.
EMBICK, D. Unaccusative syntax and verbal alternations. The unaccusativity puzzle: Explorations of the syntax-lexicon interface, p. 137-158, 2004.
FOLLI, R.; HARLEY, H. Flavors of v. In: Aspectual inquiries, p. 95-120. Springer, Dordrecht, 2005.
HALLE, M.; MARANTZ, A. Distributed Morphology and Pieces of Inflection. In: HALE, K.; KEYSER, S. J. (eds.). The View from Building 20: Essays in Linguistics in Honor of Sylvain Bromberger. Cambridge, Mass: MIT Press, 1993, p. 111-176.
HEINAT, F. Probes, pronouns, and binding in the minimalist program. The Department of English in Lund: Working Papers in Linguistics, v. 6, p. 19-37, 2006.
LAZZARINI-CYRINO, J. P. O sincretismo passivo-reflexivo: um estudo translinguístico. Tese de Doutorado. São Paulo: Universidade de São Paulo, 2015.
LEVIN, T. F.. Licensing without case. MIT, 2015.
MARANTZ, A. No Escape from Syntax: Don’t try morphological analysis in the privacy of your own Lexicon. University of Pennsylvania Working Papers in Linguistics, Philadelphia, v. 4, n. 2, p.221-225, 1997.
PESETSKY, D. M. Zero syntax: Experiencers and cascades. MIT press, 1996.
PYLKKÄNEN, L. Introducing arguments. MIT press, 2008.
SCHÄFER, F. The syntax of (anti-) causatives: External arguments in change-of-state contexts. John Benjamins Publishing, 2008.
Agreement in Brazilian Portuguese copular sentences
Bruna Karla (Universidade Federal dos Vales do Jequitinhonha e Mucuri)
This work analyzes nominal agreement in sentences such as (1), in Brazilian Portuguese (BP), where
the subject is feminine whereas the adjective, in the predicate position, is masculine. While current
proposals focus on the “disagreement” between the subject and the adjective, I hypothesize that there is
agreement between the adjective and a silent noun, in the internal structure of the post-copula DP.
(1) Moto é perigoso. (Family conversation, Belo Horizonte, 11/19/2020)
Motorcycle-FEM.SG is dangerous-MASC.SG
A motorcycle is dangerous.
1 DESCRIPTION
1.1 The predicate may refer to both a situation and an individual
According to Siqueira (2017), these sentences allow both an “event”/“situation” reading (1a) and a
“property of an individual” reading (1b):
(1a) (Andar de) Moto é perigoso.
(Riding of) Motorcycle-FEM.SG is dangerous-MASC.SG
(Riding on) a motorcycle is dangerous.
(1b) Moto é (um meio de transporte) perigoso.
Motorcycle-FEM.SG is (a-MASC.SG means-MASC.SG of transportation) dangerous-MASC.SG
A motorcycle is (a) dangerous (means of transportation).
1.2 Both readings accept a paraphrase with ‘algo’ in the predicate position
Additionally, a paraphrase with com ‘algo’ is appropriate both in the “event” reading (1a’) and in
the “property of an individual” reading (1b’).
(1a’) (Andar de) Moto é (ALGO) perigoso.
(Riding of) Motorcycle-FEM.SG is (SOMETHING) dangerous-MASC.SG
(Riding on) a motorcycle is (something) dangerous.
(1b’) Moto é (um meio de transporte/ALGO) perigoso.
Motorcycle-FEM.SG is (a-MASC.SG means-MASC.SG of transportation/SOMETHING)
dangerous-MASC.SG
A motorcycle is (a) dangerous (means of transportation)./ A motorcycle is (something) dangerous.
1.3 The “property of an individual” reading is the only one possible in some cases
Therefore, the “property of an individual” reading is not excluded in these sentences, against what
has been assumed by Rodrigues and Foltran (2013, 2014, 2015), Conto (2016), and Martin et al. (2020). In
fact, it is the only possible reading (2c) in some cases, as in (2a):
(2) a. “Mostarda é amarelo”. (SIQUEIRA, 2017, p. 97)
Mustard-FEM.SG is yellow-MASC.SG
A mustard is yellow.
b. *Comer mostarda é amarelo.
Eat mustard-FEM.SG is yellow-MASC.SG
*Eating mustard is yellow.
c. Mostarda é (um alimento) amarelo.
Mustard-FEM.SG is (a-MASC.SG food-MASC.SG) yellow-MASC.SG
A mustard is (a) yellow (food).
1.4 The predicate may have either descriptive or evaluative adjectives
Besides, Siqueira (2017) demonstrated that such structures may occur with either evaluative
adjectives (1) or descriptive adjectives (2a), while Rodrigues and Foltran (2013, 2014, 2015) and Martin et
al. (2020) exclude both the occurrence of descriptive adjectives and the “property of an individual” reading,
which should be reviewed.
1.5 The subject does not need to be a bare noun
Furthermore, Siqueira (2017) and Conto (2016) show that the subject may contain not only bare
nouns, beyond Rodrigues and Foltran’s (2013, 2014, 2015) prediction, but also nouns preceded by
demonstratives (3), possessives, cardinals, definite and indefinite articles, quantifiers, etc.
(3) Essa questão da vaga é sério mesmo. (Belo Horizonte, 10/23/2019)
This issue-FEM.SG of the slot is serious-MASC.SG indeed
This issue regarding the open position is indeed serious.
2 ANALYSIS
2.1 Null categories and agreement
Kayne (2005, 2019), Pesetsky (2013), and Höhn (2016) consider the existence of a null category to
account for the apparent mismatch of agreement, respectively, in number in Italian and French, in gender
in Russian, and in person in Spanish and Greek. Likewise, I assume that a pronominal null category, in the
DP predicate, triggers the gender agreement in sentences like (1), (2a), and (3).
2.2 Silent noun ALGO ‘something’
Therefore, the predicate of the copular sentence is not simply an adjective, but a DP made up with
an indefinite null pronoun (algo ‘something’) plus an adjective, as observed in (1’), (2’), and (3’).
(1’) Moto é [(ALGO) perigoso].
Motorcycle-FEM.SG is (SOMETHING) dangerous-MASC.SG
A motorcycle is (something) dangerous.
(2’) Mostarda é [(ALGO) amarelo].
Mustard-FEM.SG is (SOMETHING) yellow-MASC.SG
A mustard is (something) yellow.
(3’) Essa questão da vaga é [(ALGO) sério] mesmo.
This issue-FEM of the slot is (SOMETHING) serious-MASC indeed
This issue regarding the open position is indeed (something) serious.
2.3 DP-structure in the predicate position
Henceforth, the predicate APs, in the structures above, make up a DP with a covert pronoun algo,
as observed in (4) for the post-copula DP in (1).
(4) [DP D [AgrP (ALGOMASC.SG)i Agr [AP perigosoMASC.SG A [NP ti]]]].
2.4 Feature checking
This pronoun bears masculine gender features (and singular number) and triggers agreement in
gender with the adjective. In this operation (PESETSKY; TORREGO, 2007), the adjective (probe),
containing uninterpretable gender features, becomes valued [uF val] via agreement with the silent noun
algo (goal), containing interpretable and valued gender features [iF val].
2.5 Copular exclamatives
The same hypothesis applies to copular exclamatives (5) in which the copula may be covert and the
DP predicate precedes the subject.
(5) a. Que (ALGO) bom essa chuvinha! (Family conversation, Belo Horizonte, 11/13/2019)
What (SOMETHING) good-MASC.SG this rain-DIM-FEM.SG
How good is this little rain!
b. (ALGO) Super rico essa live! (Comment at a live interview chat, Belo Horizonte, 05/19/2020)
(SOMETHING) Super rich-MASC.SG this-FEM.SG live
This live is being super rich!
As for the analysis of the CP, I assume that, in (5a), the merge position of the DP ‘que bom’ is the
predicate of a small clause (KATO, 2007, SIBALDO, 2011) and its subject is the DP ‘essa chuvinha’. The
inverted position, in the sentential linear order, of the predicate results from its movement to the CP-domain,
as represented in (5a’). Likewise, in (5b), the DP ‘super rico’ moves to the CP, as seen in (5b’). Inside the
CP-domain, this DP moves probably to Spec, FocusP, according to Mioto’s (2001) analysis of the whmovement
in interrogatives. In (5), the movement is driven by exclamative illocutionary force.
(5) a’. [CP Que bomi C (que) [IP I (está)j [VP V (t)j [SC [DP essa chuvinha] [DP ti]]]]].
b’. [CP Super ricoi C (que) [IP I (está)j [VP V (t)j [SC [DP essa live] [DP ti]]]]].
3 CONCLUSION
This analysis demonstrates that there is no “disagreement”, in the copular sentences at stake, but
agreement between the adjective and a silent pronoun, in the internal structure of the post-copula DP. It
also shows that the agreement is not centered in the relation between the predicate and the subject. That is
why the subject form (whether clausal or nominal, with or without a determiner), the semantic type of the
adjective (whether descriptive or evaluative) or the predication (whether individual or situational) do not
matter. The focus on the relation between predicate and subject is changed into the relation between
adjective and noun, in the DP-internal structure of the predicate. Therefore, the analysis outlined in this
paper is aimed at accounting for the agreement, in the data referred, in a unified and parsimonious fashion.
This proposal is also coherent with a broader project on structures with apparent mismatch of agreement in
BP (XXXXX, 2016a, 2016b, 2017, and 2018).