1.1. Concrete denoting bare singulars in Brazilian Portuguese

Much literature has explored the countability of concrete object-denoting nouns in Brazilian Portuguese (BP). The class of concrete object-denoting nouns includes nouns that denote substances (such as água ‘water’ and sangue ‘blood’) and that in most languages are encoded as mass nouns and nouns that denote objects, animals, and people (such as bola ‘ball’, cachorro ‘dog’, criança ‘child’) that in most languages are encoded as count nouns. The typology of countability (Chierchia 1998a, 1998b, 2010) considers the morphological and syntactic cues – that is, the distribution of numerals, number morphology, and determiners – and the lexical features of nouns – that is, whether they denote objects or substances – to distinguish count nouns from mass nouns across languages. In some languages (labelled “number marking languages” in Chierchia 2010), only count nouns such as cachorro ‘dog’ can be freely pluralized and combined directly with numerals (eu vi três cachorros ‘I saw three dogs’). In those languages (including English and Portuguese), mass nouns are only marginally acceptable in constructions where they are directly combined with numerals and/or are pluralized. For instance, a Brazilian Portuguese speaker may say dois cafés, por favor ‘can I have two coffees, please?’ in a restaurant, but in a situation where someone sees two drops of coffee on the ground, it is unlikely that that person will say eu vi dois cafés ‘I saw two coffees’. That is, while count nouns can be unrestrictedly combined with numerals directly, mass nouns can only do that in scenarios where portions are clearly individuated and a container is conventionalized, known in the literature as coercion or, more specifically, the universal packager (Doetjes 1997; Pelletier 1975, among many others).

One particular characteristic of count nouns in Brazilian Portuguese as compared to other number marking languages is that they may be used as bare singular arguments. As previously mentioned, while other number marking languages only allow bare plurals (Mary bought bananas), in Brazilian Portuguese, as originally observed by Schmitt (1996), both bare singulars and bare plurals are possible (Maria comprou banana/Maria comprou bananas). Formal and experimental semantics studies have explored the interpretation of bare singulars (Beviláqua, Lima & Pires de Oliveira 2016; Beviláqua & Pires de Oliveira 2014, 2019; Ferreira 2010; Lima 2015; Lima 2019; Lima & Gomes 2016; Lima & Oliveira in press; Menuzzi, Silva & Doetjes 2014; Müller 2002; Müller & Oliveira 2004; Munn & Schmitt 2005; Paraguassu-Martins & Müller 2007; Pires de Oliveira & Rothstein 2011a, 2011b; Pires de Oliveira & de Swart 2015; Schmitt & Kester 2005; Schmitt & Munn 1999, 2002, among many others). One frequently discussed question in the literature about bare singulars in Brazilian Portuguese is whether these nouns necessarily favour a mass interpretation over a count interpretation due to the absence of number marking. Some proposals (Müller 2002; Müller & Oliveira 2004; Munn & Schmitt 2005; Paraguassu-Martins & Müller 2007; Schmitt & Munn 1999) have claimed that these nouns are number-neutral (2). Being number-neutral, those nouns allow both singular and plural interpretations (3, 4), as their denotation includes singularities and pluralities.

These proposals do not predict the possibility of a mass (volume) interpretation of such nouns. Under this view, the absence of number morphology in bare singulars does not entail that these nouns have the same denotation as mass nouns. Schmitt and Munn (1999) showed that bare singulars (5) and substance-denoting mass nouns (6) do not share the same distribution in constructions that involve individuation:³

Furthermore, Schmitt and Munn (1999) show that while bare singulars are acceptable in constructions with reciprocals (7) and reflexives (8), substance mass nouns do not occur in constructions that involve distribution over atomic individuals (9)–(10). As such, the authors concluded that bare singulars and mass nouns do not share the same denotation:

Pires de Oliveira and Rothstein (2011b) observed that when contrasting bare singulars with object-denoting mass nouns such as mobília ‘furniture’, a different picture emerges. Pires de Oliveira and Rothstein (2011a: 237) observed that object mass nouns share distributional similarities with bare singulars; for example, they are compatible with distributive predicates (11) and reciprocals (12):

Based on these pieces of evidence, Pires de Oliveira and Rothstein (2011b) claimed that bare singulars are kind-denoting terms⁴ that share similarities with object mass nouns, such as furniture, that denote individuated entities but are grammatically mass nouns.

From the observation that bare singulars are similar to object-denoting mass nouns, Pires de Oliveira and Rothstein (2011b), based on Rothstein (2010), build an analysis for bare singulars according to which count and mass nouns are derived from an abstract root noun N_root. The derivation of a mass noun is obtained by applying the operation MASS to N_root (13):

As presented in (13), “mass nouns denote the kind associated with the root noun, while the predicative use of a mass noun can be recovered by the ∪ function. The ∪ function, when applied to the kind, will give back the set of instantiations of the kind in w₀, i.e., the subset of entities in N_root which exist in w₀” (Pires de Oliveira & Rothstein 2011b: 2163).

As for singular count nouns,⁵ the operation COUNT_k applies to N_root and picks out “the set of ordered pairs {<d,k>: d ε N ∩ k}, i.e., the set of entities in N_root which count as one in context k” (Pires de Oliveira & Rothstein 2011b: 2164).⁶ The critical aspect of their analysis is that bare singular nouns allow both a mass and a count interpretation. This is possible under their analysis because Pires de Oliveira and Rothstein (2011b: 2164) postulated that languages vary in whether they follow or not the “either/or” principle (14):

The authors noted that in English, the principle in (14) “is the default situation, there is a limited set of cases in which both MASS and COUNT_k may apply to N_root” (Pires de Oliveira & Rothstein 2011b: 2164). On the other hand, in languages like Brazilian Portuguese, the principle described in (14) does not apply. As such, the authors proposed that all nouns allow a count and a “lexicalized mass” counterpart. The lexicalized mass counterpart will surface as bare singulars given that mass nouns are morphologically singular. Nouns on which the operation COUNT can apply will also allow the operation MASS. That is, in languages like Brazilian Portuguese, count nouns will always have a mass counterpart; nouns on which the operation COUNT cannot apply, on the other hand, will not have a count counterpart, unless additional operations apply (i.e., coercion).

Rothstein and Pires de Oliveira (2011b) presented two predictions that are borne out from their proposal. First, bare singulars may occur as complements of mass quantifiers (15):

In (15), as discussed by the authors, the mother refers to the volume/weight of the books being carried; so volume, not cardinality, is the feature that is salient in the context; we could have a few very heavy books. Second, in comparatives, bare singulars allow both a comparison in terms of volume and cardinality (16):

While in sentences with bare singulars (16), the bare singular noun (pedra ‘stone’) allows for a comparison in terms of volume (size of the stones) and in terms of cardinality (number of stones), when the sentence includes a bare plural (pedras ‘stones’), only a cardinal interpretation is available.

Comparatives have long been discussed in the mass/count literature as a reliable diagnostic for distinguishing count nouns from mass nouns (Bale & Barner 2009). In comparative structures, count nouns are more likely to be evaluated in terms of cardinality, while mass nouns are evaluated in terms of volume. Most of the experimental studies on this topic in Brazilian Portuguese involve quantity judgment tasks (see, e.g., Beviláqua & Pires de Oliveira 2014, 2019; Lima & Gomes 2016), which are characterized by comparison constructions.

Quantity judgments (see Barner & Snedeker 2005) consist of comparison tasks, where participants are presented with two characters: one (A) that has large quantities of X (e.g., two large portions of X or two large individual Xs) and another (B) that has a more numerous quantity of Xs (e.g., six portions of X, six individual Xs), where each portion/individual is of a smaller size. The participants are then asked, “Who has more X?” For example, in order to answer the question “Who has more ketchup?” participants are presented with two big portions of ketchup (A) and with six small portions of ketchup (B) (Figure 1).

Figure 1

Sample stimuli, quantity judgment tasks (Barner, Li & Snedeker 2010: 4).

Quantity judgment tasks have shown that English speakers (16 adults and 16 four-year-olds) choose A over B when answering questions that include mass nouns (such as ketchup) (0% cardinal answers) and that they choose B over A when answering questions that include count nouns (such as pens) (100% cardinal answers) (Barner & Snedeker 2005). It is important to note that in English a small set of count nouns (so-called “flexible” nouns; Barner & Snedeker 2005; Gillon 1992; Rothstein 1999, 2004, 2010, among many others) can be used both in their pluralized form (count syntax), as in who has more chocolates?, and in their bare form (mass syntax), as in who has more chocolate? That is the case for nouns such as stone, chocolate, and paper. In a study with English-speaking adults and children (16 adults and 12 four-year-old children), Barner and Snedeker (2005) found that a volume interpretation is favoured when count nouns are used with mass syntax (who has more chocolate?), while a cardinal interpretation is favoured when they are used with count syntax (who has more chocolates?) (count syntax: 97% of cardinal answers; mass syntax: 3% of cardinal answers).

In the same study, when asked a question that included an object mass noun (who has more furniture?), English speakers (16 adults and 16 four-year-olds) favour a cardinal interpretation over a volume interpretation (97% of cardinal answers) (Barner & Snedeker 2005). That is, despite being grammatically mass nouns, these nouns allow a cardinal interpretation. These results are important particularly when contrasted with the results for the same type of task with flexible nouns, like stone, that can be used bare or pluralized. Bare flexible count nouns trigger, in English, a mass interpretation, as summarized in Table 1.

In sum, the results of quantity judgment tasks in English illustrate that while pluralized nouns are only associated with quantification of individuals, mass nouns of different types (water/furniture⁷) and bare count nouns (stone) do not “force an unindividuated construal” (Barner & Snedeker 2006: 167). Barner and Snedeker (2006) called this the Number Asymmetry Hypothesis.

As previously discussed in this article, and following Pires de Oliveira and Rothstein (2011b), Brazilian Portuguese and English are not similar with respect to the distribution of bare singulars. In English, as discussed by several authors (e.g., Pires de Oliveira & Rothstein 2011b; Schmitt & Kester 2005), only a small class of count nouns allow for this alternation (bare/plural). As previously discussed, in English, the principle described in (14) applies to all nouns. That is, only COUNT_k or MASS will apply to each N_root noun. In Brazilian Portuguese, the availability of bare singulars is not restricted to a small class of nouns, suggesting that the principle described in (14) does not apply to this language. That is, in Brazilian Portuguese, but not in English, all count nouns seem to have bare and plural counterparts.

The difference in the status of bare singulars (flexible nouns) in English (used in “lexically restricted contexts”; Schmitt & Kester 2005: 237) and bare singulars in Brazilian Portuguese becomes clear when we contrast the interpretation of these nouns in these two languages. As previously observed, in English, the small class of nouns that allow such variation seems to be interpreted as always referring to volume (see Table 1) when bare. That is, the absence of number naturally triggers a mass interpretation for such nouns. In Brazilian Portuguese, the same does not hold: as shown by Pires de Oliveira and Beviláqua (2006) and Pires de Oliveira and Rothstein (2011b), bare singulars allow both cardinal and volume interpretations.

The prediction that bare singulars allow cardinal and volume readings is borne out in Pires de Oliveira and Rothstein’s (2011b) analysis. However, if both readings are possible for bare singulars, why the cardinal interpretation seems to be favoured over the volume interpretation in neutral contexts needs to explained. Lima and Gomes (2016: 206) proposed two factors to account for these facts: 1) a natural atomicity bias (“nouns that denote kinds whose canonical instances are individuals more likely to be grammaticalized as a count noun and to be interpreted as referring to cardinalities in Brazilian Portuguese”) and 2) lexical statistics, based on Samuelson and Schiffer (2011) (“a BS noun like livro is more likely to be interpreted as a count noun due to the high frequency of count interpretations of other forms of the same lexeme (such as the plural form livros)”). If a natural atomicity bias exists, one interesting question is whether other lexical features could impact the interpretation of bare singulars. While much work has been done on the interpretation of concrete object-denoting nouns in Brazilian Portuguese, the literature on Brazilian Portuguese has not yet addressed the interpretation of bare abstract nouns derived from verbs (henceforth “deverbal nouns”). The investigation of deverbal nouns seems critical for this debate for two reasons.

First, as previously mentioned, prior to this study, no experimental study has focused on non-concrete-object-denoting nouns in Brazilian Portuguese. Second, by investigating abstract nouns and, more specifically, deverbal nouns, we can explore the lexical features of the N_root and the potential impact of those features on the interpretation of bare singulars. In other words, while according to Pires de Oliveira and Rothstein (2011b), among others, bare singulars allow both count (cardinal) and mass (volume) interpretations, lexical information from the root might determine the more likely interpretation for different nouns.

As such, the main goal of this article is to investigate the interpretation of deverbal nouns. Deverbal nouns are relevant to this debate as the verbs they are derived from have inherent aspectual features and might be critical for showing that while bare singulars allow both readings, the ultimate interpretation of a noun might be impacted by the features encoded by the root. In sum, we want to explore whether aspectual features impact the interpretation of such nouns with respect to countability and how these aspect features interact with plurality (or lack thereof).

1.2. Studies on abstract nouns

Across languages, most studies on countability are based on concrete object-denoting nouns. Studies that explore the countability of abstract nouns mostly focus on English and include formal, quantitative approaches to the topic (see Grimm 2014 for a review of the literature and Grimm 2016 for a case study on the noun crime) and experimental studies that explore the interpretation of deverbal nouns (see Barner, Wagner & Snedeker 2008).

Barner, Wagner, and Snedeker (2008) explored the impact of the noun form (bare or pluralized) on the interpretation of English deverbal nouns. The authors explored this question by means of a quantity judgment task. The authors manipulated two factors: noun form (nouns that cannot be pluralized – dancing, jumping – and nouns that can be pluralized – dance(s), jump(s)) and event type (durative vs. punctual). Therefore, four conditions were manipulated: punctual mass, punctual count, durative mass, and durative count. Participants were exposed to a short narrative followed by a target question to which two answers were available, as illustrated in (18) and (19):

Barner, Wagner, and Snedeker (2008) found that non-pluralized durative deverbal nouns (driving) were more likely to have a mass interpretation than non-pluralized durative deverbal nouns (drives). Contrariwise, non-pluralized punctual deverbal nouns (jumping) and pluralized punctual deverbal nouns (jumps) are equally likely to be associated with a count interpretation. As such, the NP meaning was critical for the interpretation of deverbal punctual nouns as their interpretation was not solely dependant on the form of the noun (bare or pluralized).

As previously mentioned, in Brazilian Portuguese, a less explored field of research is the interpretation of deverbal nouns with respect to countability. Two aspects of Brazilian Portuguese make this language an interesting test case for the impact of the absence/presence of number morphology on the interpretation of nouns. First, Brazilian Portuguese, unlike English, is a language where bare singulars are not restricted to marked uses and are frequently used in the object position of episodic predicates (see Pires de Oliveira 2014 for a review of the literature on bare singulars in Brazilian Portuguese⁸). Second, as discussed in Section 1.1, studies have shown that (concrete object-denoting) bare singulars are not necessarily interpreted as mass nouns. Instead, some studies suggest that the lexical meaning (i.e., whether a noun denotes an object or a substance) drives the interpretation of concrete object-denoting count nouns in neutral contexts. In other words, while both cardinal and volume interpretations are licensed by bare singulars, a cardinal interpretation seems to be overwhelmingly preferred over a volume interpretation in neutral contexts (i.e., when a particular interpretation is not favoured by the context; see Lima & Gomes 2016).

In this paper, we discuss the interpretation of deverbal nouns with respect to countability in Brazilian Portuguese. More specifically, we investigate the interaction between the lexical meaning of the nouns (whether they refer to iterative or continuous events) and noun form (bare or pluralized). According to the Number Asymmetric Hypothesis (Barner & Snedeker 2006), mass syntax (non-pluralized nouns) “is unspecified and permits comparison based on an unbounded range of measuring dimensions including mass, volume, time and crucially number” (Barner, Wagner & Snedeker 2008: 807). Given that the lexical meaning of the noun impacts the interpretation of concrete bare singulars in neutral contexts, we predict that a similar pattern will be observed for deverbal nouns in Brazilian Portuguese. That is, bare singulars denoting continuous events (derived from durative verbs) should be less likely to license a count interpretation compared to nouns that denote iterative events (derived from punctual verbs). These predictions are explored based on a quantity judgment task in Brazilian Portuguese that replicates Barner, Wagner, and Snedeker’s (2008) studies in English.