Leiden, January 22nd, 2014. Brazilian Bare Phrases: Plurality, and Kinds. Roberta PIRES DE OLIVEIRA CNPq/UFSC ropiolive@gmail.com Aims: 1. To have a better understanding of the nominal system in spoken Brazilian Portuguese; 2. To investigate the Bare Singular and the Bare Plural; 3. To account for agreement and the semantics of noun phrases; 4. To investigate the Bare singular and the Bare plural in the scope of quantifiers (comparatives). Hypotheses: 1. Plural morphology is highly significant in BrP; 2. Plurality is (atomic) sums; 3. The Bare Plural is highly specialized; 4. The Bare Singular denotes the kind (mass are kinds as well), whereas the Bare Plural denotes a plurality. 0. Preliminaries Spoken BrP distinguishes singular and plural. The plural must be marked in the leftmost element1: (1) a. Os menino(s) Thes boy(s) c. * O meninos the boys b. O menino the boy It distinguishes between mass and count nouns: (2) a. * duas areias / * Eu vi areias Two sands / I saw sands. b. duas menina(s) two girls With numerals there is no need of marking plurality in the noun, (2.b). Thus, the plural mark in the noun seems to be a concord phenomenon. Descriptively, it has both a singular indefinite article and a plural indefinite: (3) a. b. 1 Eu vi um menino. I saw a/one boy Eu vi uns menino(s). I saw some boys. See Schere & Naro (1997), among others. Leiden, January 22nd, 2014. Comments: (i) ‘um’ is an indefinite and a numeral; (ii) it is unclear whether ‘uns’ is an article. And a Bare Singular and a Bare Plural! This goes against the generalization that says that if a language has an indefinite article (singular and plural), it should not have bare phrases. BrP has all the alternatives in the nominal domain overtly given. 1. The Bare Singular and the Bare Plural The Bare Singular is not a disguised plural. (i) (ii) Plural morphology must be marked in the left most expression. Schmitt & Munn (2002). Plurality in some nouns is accompanied by vocalic alternations: ovo /ovo/ vs ovos /óvos/ If the Bare Singular were a disguised plural we would expect the plural form without the morpheme, but that is not what we get: (4) a. b. João comprou /ovo/ * João comprou /óvo/ João bought eggs. 1.1 Comparison Rothstein & Pires de Oliveira (2013) show that the bare singular and the bare plural behave differently in comparison sentences. The bare plural behaves as the bare plural in English: they only raise comparison by cardinality: (5) a. b. João tem muitas pedras. (card/*vol) John has many stones. (card/ *vol) English does not have the bare singular. In comparison, flexible nouns in English do not behave as the bare singular in BrP because flexible nouns in mass syntax only allow for a mass interpretation, whereas in such a context, the bare singular is ambiguous: (6) a. b. João tem mais pedra do que o Paulo. (card / vol) John has more stone than Peter (*card / vol) The bare noun ‘pedra’ allows for a counting interpretation, which is blocked with stone. Moreover, the flexible behavior in English is restricted to some nouns. The bare singular differs from atomic mass nouns, such as furniture, because they have a count counterpart – ‘pedra’/ ‘pedras’, but ‘mobília’/* ‘mobílias’. However, it behaves as furniture because it gives rise both to cardinal and non-cardinal interpretations, (6.a). It is not the case that the bare singular is a plural without morphology. The plural morpheme ‘-s’ must appear in the leftmost element of the phrase. (7.a) is ungrammatical: (7) a. * João tem muita pedras. João has much stones b. João tem muitas pedra. João has many stone Thus, (6.a) is not the result of dropping the plural morpheme. Moreover, ‘muito’ combines with mass nouns, whereas ‘muitos’ does not: (8) a. João comprou muita areia. João bought much sand b. * João comprou muitas areia(s). João bought many sands Leiden, January 22nd, 2014. Thus, ‘muito’ is a mass quantifier. (8) shows that it combines with the bare singular, engendering a cardinal and a volume interpretation as furniture type mass nouns do: (9) João tem muito livro pra carregar. (card/vol) João has much/many book to carry. 1.2 Pluralities As expected, plural phrases may be interpreted inclusively or exclusively; though most of the time they are interpreted exclusively (generalized implicature): (10) Você tem filhos? Do you have kids / kid ? Tenho um / # Não, tenho um. I have one / No, I have one. The same is true with the Bare Singular: (11) Você tem filho? Tenho um / # Não, tenho um. The Bare Plural implies plurality whereas the Bare Singular does not: (12) a. João comprou livros. João bought books b. João comprou livro.2 João bought book. (12.a) is infelicitous in a situation where the speaker knows that João has bought just one book. This is not the case with (12.b). Thus, the bare singular does not imply plurality. This asymmetry reappears in contexts where the bare plural shows sensitivity to pluralities; the bare singular is not sensitive to plurality, as shown in (13.b): (13) a. O João tem # narizes romano(s). João has Roman noses b. João tem nariz romano. João has Roman nose. Moreover, only the bare plural combines with ‘numeroso’ (numerous), a predicate that selects for plurality: (14) a. b. * Pernilongo é numeroso nessa época do ano. Mosquito is numerous in this time of the year. Pernilongos são numerosos nessa época do ano. Mosquitos are numerous in this time of the year. If “a plural is always odd when sum values are pragmatically excluded from its domain of reference” (Farkas & de Swart (2010)), then the bare singular is not plural. Farkas & de Swart’s proposal that the plural has to denote sums explains the Bare Plural, but not the Bare Singular. If it were a plural predicate, why it cannot combine with ‘numeroso’ (numerous)? Why is it insensitive to plurality? 1.3 2 Different grammars? Whether ‘livro’ in (11.b) is an argument or is incorporated will be left as an open issue here. Leiden, January 22nd, 2014. One could imagine that the Bare Singular and the Bare Plural belong to different grammars: the Bare Singular is spoken BrP whereas the Bare Plural is written. Perini (2010) claims that the sentences below have the same meaning and differ with respect to the variety: (15.b) is written language: (15) a. b. Menina é estudiosa. Meninas são estudiosas Girls are studious. Comment: Generic contexts. It is hard to find differences, but: (16) a. b. # Tio gosta de pizza. (lawlike) Tios gostam de pizza. (generalization) Moreover, some predicates only accept the Bare Plural (14.b). And there are contexts where the Bare Plural is more natural; the Bare Singular is marked: (17) a. b. Crianças estão brincando na rua. Children are playing on the street. # Criança está brincando na rua. There are interpretations which are achieved only with the Bare Plural, i.e. subkinds. The Bare Singular is never interpreted as sub-kinds: (18) a. b. Baleias estão acabando. (kind; subkind) Whales are on the verge of extinction Baleia está acabando. (only kind) Finally, even a brief search in corpora shows that both forms co-exist: Search on blogs, January 18: comprou Comprar Definite ‘o livro’ 982 2.160 Indefinite ‘um livro’ 1.180 20.500 Bare Singular ‘livro’ 118 6.630 Bare Plural ‘livros’ 25 3.670 2. Theoretical Explanation Descriptively, the absence of the morpheme ‘-s’ in the noun phrase may indicate a singular predicate (1.b), mass denotation, (6.a), (8) and (9), a singular denotation (13.b) or an inclusive plurality (11), whereas the plural always denotes plurality (inclusive or exclusive). The bare singular is not sensitive to plurality, (13.b), and it does not combine with a plural predicate (14). Thus, the absence of plural morphology has different interpretations. Our proposal is that plural morphology denotes (semantic) pluralities, i.e. sets of (atomic) sums, and triggers plural agreement. The absence of plural morphology triggers singular agreement, and semantically it denotes a singular individual (a kind), but the noun does not have to be a singular predicate. Mass phrases, for instance, trigger singular agreement because they denote the kind, but they are not singular predicates: (19) Água faz bem pra saúde. Water makes well to healthy. Leiden, January 22nd, 2014. Water is good for healthy. Horn’s maxim is about the presence versus absence of overt morphology. In this particular sense, the plural form is marked, because it is overtly signaled, as argued by Farkas and de Swart (2010), and it must denote plurality. In this particular sense, the absence of plural morphology is weak and the plural is semantically strong, because it covers less situations: the plural denotes plurality, so it can only be used if (atomic) sums are involved. However, a semantic singular predicate is stronger than the plural predicate, as claimed by Sauerland (2008). Compare: (20) a. b. O menino saiu. n = 1 Os menino(s) saíram. n 1 (20.b) may be true in a situation where just one boy left and in a situation where more than one boy left, whereas (20.a) can only be true in a situation where just one boy left. Thus, the singular is more informative than the plural. Our proposal relies on Rothstein & Pires de Oliveira (2013), and Rothstein (2010). Root nouns are unsorted lattice structures, that is, there are no semantic atoms. They are sum structures. To my mind this is not number neutrality. Farkas & de Swart understand number neutrality as “in principle, may denote in any subdomain of the lattice” (6:19), but this definition also covers the notion of inclusive pluralities. An inclusive plural phrase may denote atoms or sums of atoms. They also claim that number neutrality means “truly insensitive to the atom/sum divide” (6:22), thus they are mass like. Unsorted lattice structures are primitive structures in the sense that there is no semantic distinction between sums and atoms, the grammar does not see this distinction (but that does not mean that cognitively there are no atoms). Thus, it is neither mass nor count (yet). Counting is a recursive operation that applies to some root nouns. BrP distinguishes between count and mass nouns via morphology: mass nouns cannot be pluralized nor counted (2). Counting is an operation that pairs an individual and a cardinal: [[Pc]] = {<a, 1>, <b, 1>, <c, 1>, <a+b, 2>, <a+c, 2>, <b+c, 2>, <a+b+c, 3>} Count predicates denote sets of pairs of an individual (including a plural individual) and a cardinal. The plural morpheme denotes the plural count predicate; it does not matter whether it is attached to the determiner or to the noun. At the lowest level of a derivation, root nouns denote unsorted lattices (not singular nor plural, not mass nor count). If the count operation applies, resulting in atoms and pluralities, then the combination with numerals is allowed. Root nouns cannot combine with numerals, precisely because they do not denote semantic atoms. Numerals require sets of atoms (Ionin & ()). Thus, ‘menina’ must be a plural count predicate in (2.b): (21) 2 count menina(s)pl meninaroot The absence of plural morphology in the nominal phrase introduces a presupposition of singularity which is satisfied if the cardinality is one: (22) o meninos count meninoroot Leiden, January 22nd, 2014. Bare phrases are the result of a type shift (Chierchia 1998). The down operator only applies to cumulative predicates. Thus, it applies to root nouns and to plural predicates. In either case, it denotes the maximal sum in every world (an intensional individual). Thus, (12.a) and (12.b) have different structures; roughly represented in (23): (23) a. comprou count livrosk livropl comprou livroroot livrok livroroot Assume Landman & Rothstein’s (2011) view that kinds may be the theme of episodic predicates. In (23.a), the kind is derived from a count predicate, thus, it induces a plural interpretation; thus, it is not felicitous in a situation where (the speaker knows that) just was book was bought. In (23.b), the kind is derived from a root noun, thus, it is felicitous in a situation that involves just one book (it is felicitous even if the speaker knows that only one book was bought). Since the Bare Singular does not denote a plurality, it cannot combine with ‘numeroso’ (numerous). Some predicates are restricted to semantic plurality, others are not. ‘Comparar’ (to compare) is not; it cannot combine with singular predicates, and the Bare Singular is not a singular predicate either: (24) Carlos compara currículo. Carlos compares curricula. The Bare Singular denotes a kind, and kinds may have many instances, just one instance or no instances in a particular world (there are no dodos in our world). Thus, it is felicitous in contexts where more than one individual is involved, as well as in contexts where just one individual is involved. Thus the Bare Singular is not sensitive to plurality, and we explain (13.b). 2.1 Comparison and quantifiers In (5.a) the Bare Plural denotes a plural predicate, thus, it cannot but be comparing cardinalities, given the semantics of the count noun. 3 How do we explain (6.a) and (9)? How do we arrive at the cardinal reading? Barner and Snedeker’s (2005) show that furniture has a cardinal interpretation, though it cannot be counted: (25) a. John has more furniture than Bill. b. *How many more? c. *John has one more furniture/two more furnitures than Bill. Rothstein & Pires de Oliveira (2013) claim that the cardinal reading with the Bare Singular in BrP, (6.a), and (9), as well as with object mass nouns as furniture or mobília results from measuring. Thus, count syntax forces quantity evaluations in terms of cardinalities, while mass nouns do not require it, but allow it depending on the noun. The mechanism for counting is, however, not the same. Pires de Oliveira and Mendes de Souza (2013) distinguish bare singulars from singular predicates which in comparisons coerce a partitive readings as in: (i) João tem mais mesa que Pedro. (ii) John has more table than Peter. 3 Leiden, January 22nd, 2014. Rothstein (2010, 2012) claims that the fundamental contrast at the root of the mass/count contrast is the distinction between measuring and counting: counting is putting entities in one-to-one correspondence with the natural numbers, whereas measuring is assigning an overall quantity a value on a scale. Mass nouns denote sums which can be measured in this way. They do not allow a set of atoms to be identified grammatically, and thus the linguistic operation of counting is impossible - three pieces of furniture / *three furnitures – as well as pluralization - * furnitures. While mass nouns do not allow a set of atoms to be identified grammatically and counting is impossible, operations such as comparison may evaluate the relative sizes of sets in terms of their cardinality if the context makes the set of natural atoms salient. This explains the cardinal interpretation of (6.a), (9) and (25). When a mass predicate is naturally atomic - furniture and ‘livro’ (book) -, the cardinal scale is one option that can be made salient by context for comparisons and other measurements. Thus, the question in (26) can be translated as “which quantity of furniture has a higher value on the cardinal scale”: (26) Who has more furniture? Answering this question may involve implicit counting, but may also involve estimation of cardinalities without counting. Count nouns in BrP always have a mass counterpart, since all root nouns have a mass realization, but only some have a count realization. Predicates that denote naturally atomic individuals tend to be interpreted via cardinality if no other contextual clues are given, but allow for comparison along other dimensions. 2.2 Hungarian As shown by Schvarcz (in preparation), Hungarian has two question words: hány (how many), which goes with count nouns, and mennyi (how much), which goes with mass nouns: (27) a. Mennyi rizs-(e)-t vettél? How much rice--object marker buy- Past-2ndSG ‘How much rice did you buy?’ b. * Három / Három kiló-t. Three / Three kilo-obj. marker ‘Three kilos’ Count nouns in Hungarian are not marked plural when they occur with numbers, as shown in the examples (27.b) and (28.c) below: (28) a. Hány könyv van a táskádban? How many book is there the bag+possesive your+suffix in ‘How many books are there in your bag?’ b. Csak három. ‘Only three’. c. # Három kiló. Three kilo. ‘Three kilos.’ The measure answer in (28.c) is pragmatically odd, since hány (how many) asks for a cardinal answer. Mennyi (how much) combines with a bare noun ‘könyv’ (book). Crucially, in these cases, Leiden, January 22nd, 2014. the appropriate answer is either a measure expression (29.b) or a cardinal quantity judgment which makes use of a number: (29) a. Mennyi könyv-e-t tudsz cipelni? How much book–LV-obj.marker you are able to carry? ‘What quantity of book can you carry?’ b. Három kiló-t. Three kilo-obj. marker. ‘Three kilos.’ c. Három-nál nem többet Mert Three- comparative not more because különben elszakad a táskám. otherwise break the bag+possesive my ‘Not more than three, otherwise my bag would break.’ d. Hárm-(a)-t three-PL ‘Three’ Assuming that mennyi (how much) induces a mass usage, volume and cardinal answers are acceptable with the mass counterpart of the count noun in Hungarian too, as ‘muito’. Whereas hány (how much) induces a count interpretation, as the plural morphology in the quantifier (muito-S (how many)) does in BrP, only allowing for cardinal readings. 3. Cross-linguistic variations In Chinese, root nouns are always available for the down operator, and the same is true for BrP. But BrP has articles and plural morphology. Chinese only has classifiers. Thus, counting is morphologically marked in BrP, whereas in Chinese classifiers are obligatory. English has articles and plural morphology, but no Bare Singular. Pires de Oliveira & Rothstein (2011) suggest that English is an either/or language: either the root noun is available for the down operator or the count operation applies, but not both. Thus, English has the Bare Plural, but no Bare Singular. In French, the root noun is never available, thus there are no bare phrases. In Yudja (Lima (2011)), counting is always available. References Barner, D. & Snedeker, J. 2005. “Quantity judgements and individuation: Evidence that mass nouns count”. Cognition. 97: 41-66. Chierchia, Gennaro. 1998. “Plurality of mass nouns and the notion of 'semantic parameter'”. In S. Rothstein (ed.) Events and Grammar, Dordrecht: Kluwer. 53-103. Farkas, D. & de Swart, H. 2010. The semantics and pragmatics of plurals. Semantics & Pragmatics. Vol. 3, 1.24. Ionin, T. & Matushanksy, O. 2006. The Composition of Complex Cardinals. Journal of Semantics 23: 315-360. Lima, Suzi. 2012. Numerals and the universal packager in Yudja (Tupi). Proceedings of the Sula to appear. Pires de Oliveira, R. & Rothstein, S. 2011. Bare nouns in are mass in Brazilian Portuguese. Lingua. 121: 2153-2175. Pires de Oliveira, R. & Mendes de Souza, L. 2013. O singular nu e a comparação: uma proposta de derivação semântica. Revista LinguíStica. 9(1). Rothstein, S. & Pires de Oliveira, R. 2013. Comparatives in Brazilian Portuguese: Counting and measuring, in press. Leiden, January 22nd, 2014. Rothstein, S. 2010. Counting and the mass-count distinction. Journal of Semantics. 27(3): 343-397. Rothstein, S. 2012. Numbers: Counting, Measuring and Classifying. In Proceedings of Sinn und Bedeutung 16, Aguilar-Guevara, Ana, Chernilovskaya, Anna, and Nouwen, Rick (eds), 527-543. MIT Working Papers in Linguistics: Cambridge, MA. Sauerland, U. 2008. On the semantic markedeness of phi-features. In Harbour, D. et. Al. (ed.) Phi theory, 57-83. Oxford: Oxford University Press. http://www.zas.gwzberlin.de/home/sauerland/downloads.html Schere, M. & Naro, J. 1997. A concordância de número no português do Brasil: um caso típico de variação inerente . In: Dermeval da Hora. (Org.). Diversidade Lingüística no Brasil. João Pessoa: Idéia. 93-114. Schmitt, C. & Munn, A. 1999. Against the nominal mapping parameter: bare nouns in Brazilian Portuguese. Proceedings of NELS 29. Schvarcz, Brigitta. In preparation. Issues in numbers, counting and measuring in Hungarian. ms Bar-Ilan University.