1 Introduction
How does the syntax of a noun phrase impact its meaning? Many languages, including English, make a formal distinction between mass and count syntax. In English, nouns typically used in count syntax (e.g., a cat; a table) can be pluralized and modified by cardinal numbers (e.g., one cat; two cats) or quasi-cardinal determiners (e.g., several cats; many cats). In contrast, nouns typically used in mass syntax (e.g., some sand; some water) can’t be pluralized without a change in meaning, cannot be modified by cardinal or quasi-cardinal expressions, and instead are modified by terms like little or much (Jespersen 1909; see Grimm & Wahlang 2021 for a more recent analysis). While these grammatical properties of the mass–count distinction are widely agreed upon, its semantic interpretation remains less certain.
According to many prominent proposals, beginning with Quine (1960), but adopted with variations by many linguists (Link 1983; 1998; Bunt 1985; Barker 1998) and psycholinguists (Macnamara 1972; Gordon 1985; Bloom 1994a; 1999; Wisniewski et al. 1996), the mass–count distinction corresponds to a contrast in how nouns quantify over elements in their domain. According to Quine, count nouns, but not mass nouns, “possess built-in modes … of dividing their reference” (1960: 91). Consequently, the denotations of count nouns have atomic and non-divisible minimal parts, or individuals, i.e., members whose sub-portions do not count as instances of the noun (e.g., the leg of a table isn’t a table). These individuals provide a basis for counting and comparison (e.g., having more tables means having more of these individuals). On this type of account, only mass nouns should allow for “cumulativity of reference”, such that combining two referents of a mass noun results in an instance of the same mass noun (e.g., combining water results in water), whereas two referents of a count noun cannot be labeled by a singular noun. Relatedly, only mass nouns should permit “divisity of reference” (Cheng 1973), where any part or portion of a referent of a mass noun counts as an instance of the same mass noun (e.g., any sub-portion of sand is sand). As a result, there are no atomic minimal parts in the denotation of mass nouns, and therefore no individuals to count or compare.
Against this broad approach, others have argued that there isn’t a perfect correlation between cumulativity, divisity, and mass–count syntax (see Gillon 1992; 1999). While many counterexamples are discussed in the literature, most critical to the present study is the claim that some mass nouns, like furniture, resist divisity. For example, it is not the case that any part or portion of something labeled as furniture also counts as furniture (e.g., the leg of a sofa is not furniture, but is a part of a piece of furniture). Similarly, other so-called “object-mass” nouns (Barner & Snedeker 2005) like cutlery, luggage, clothing, etc. do not appear to allow arbitrary divisity (see also Mihatsch 2016; Erbach et al. 2021).
Observations such as these led Gillon (1992; 1999) to conclude that, while count nouns are grammatically specified to denote a set with atomic minimal parts, the denotations of mass nouns are linguistically unspecified. According to this hypothesis, mass nouns denote sets with minimal parts if there is evidence in the world that there are linguistically accessible atomic individuals in their denotations (Bunt 1985; Gillon 1992; 1999; see also Katz 1970). In a related proposal, Chierchia (1998; 2010) argued that all mass nouns have minimal parts, but that only object-mass nouns have “stable” atoms like count nouns, such that their denotations can also be compared via counting. On this view, differences in syntax across languages should not affect how a noun quantifies—“Pavarotti’s hair is Pavarotti’s hair, whether we talk about it in Italian or in English, i.e. whether we get at it through a mass noun or through a count noun. If we don’t want semantics to start looking like magic, we have to say that in the real world “hair” and “capello” obviously denote the same stuff.” (Chierchia 1998: 88). Meanwhile, whereas nouns like furniture denote stable atoms, the minimal parts of other mass nouns, like mustard, are only vaguely specified and vary from context to context, making it difficult to count these parts and evaluate them according to number (for related proposals that distinguish between object-mass and substance-mass nouns in terms of the presence or absence of (stable) atomic parts, see Sutton & Filip 2018; Rothstein 2021; Erbach & Schoenfeld 2022).
In an effort to evaluate these claims, Barner and Snedeker (2005) asked adults and 4-year-old children to make quantity judgments for items denoted by object-mass nouns (e.g., furniture, jewelry), count nouns (e.g., shoes, cups), and substance-mass nouns (e.g., toothpaste, ketchup) that differed by number of individuals/portions and summed volume. For example, participants were asked to judge who had “more furniture” between a character who had two large tables and two large chairs, and another who had a greater number of things, but that had a smaller summed volume—e.g., three small tables and three small chairs. For object-mass nouns, both children and adults preferred the character with the greater number of individuals over the character who had the greater volume. Critically, participants made similar judgments for count nouns, but based judgments on volume for substance-mass nouns. Further, these judgments were not based purely on the availability of discrete physical objects in the world (c.f., Gillon 1992; 1999). When presented with items like stones or strings, participants based quantity judgments on number when the questions were phrased using count syntax (e.g., “Who has more strings?”) but on mass or volume when phrased with mass syntax (e.g., “Who has more string?”), despite the fact that physical things were salient in both contexts, and the number of atomic particles of string, stone, etc. was constant. Subsequent studies have replicated these findings in English and found similar results in other languages (Barner et al. 2009; Inagaki & Barner 2009; Van Witteloostujin 2013; Inagaki 2014; Van Witteloostuijn & Schaeffer 2014; Hacohen & Schaeffer 2016; Scontras et al. 2017; Lin & Schaeffer 2018; Mohr & Agyepong 2022), with novel mass and count nouns (Barner & Snedeker 2006), and with event-denoting deverbal object-mass nouns (Barner et al. 2008; see Bale & Barner 2018, for review).
Such results suggest an asymmetry between mass and count nouns, such that count nouns receive a uniform semantics, whereas mass nouns do not—some mass nouns individuate while others do not (e.g., see Bale & Barner 2009). Given this, several proposals have sought to preserve symmetrical accounts by arguing that object-mass nouns may actually quantify continuously after all. On one such account, originally described by McCawley (1975), nouns like furniture do not denote sets of individuals, but instead represent the function that such things fulfill, such that the denotation of furniture is measured and quantified by the degree to which the function of furnishing is fulfilled. As an example, McCawley argues that, “if Fred has 4 chairs, 3 magazine racks, 2 coffee tables, and 1 lamp, and I have 2 chairs, 1 desk, 1 bed, 1 sofa, and 1 table, my 6 pieces of furniture would constitute more furniture than Fred’s 10 pieces do” (p. 319). This is because “his” furniture better fulfills the function of “furnishing a space” compared to Fred’s (see also Prasada 1999).
In a related proposal, Rothstein (2010; 2017) argued that although the individuated reading of object-mass noun comparisons is the most salient, this meaning is not lexically encoded and may shift contextually (see also Hampton & Winter 2024; Rothstein & de Oliveira 2020). For example, in a context where two people are moving their furniture into a new place and John needs to move a grand piano, a large sofa, a double bed and a heavy wardrobe (4 pieces), while Bill needs to move four folding chairs, a small table, and a rolled-up mattress (6 pieces), people should judge that John has more furniture, hence basing their judgment on volume of furniture rather than number (Rothstein 2017: 122). Thus, whereas the individuation of count nouns is grammatically specified, the individuation of mass nouns is not, and is more strongly determined by context.
Several studies have attempted to test these ideas by manipulating function fulfillment. To do so, they presented participants with scenes that varied both the number of sub-kinds (e.g., chairs, sofas, etc.) and the number of instances of those sub-kinds (i.e., individual chairs). They then asked whether participants said there was “more” when the number of sub-kinds was greater, and better fulfilled the desires function of, e.g., furniture, or if instead quantity judgments were based primarily on the number of discrete individuals. For example, in one study, Gordon and Rodman (2006) placed objects denoted by object-mass nouns in contexts that emphasized their functional role (e.g., items of furniture used to furnish two rooms), and asked English-speaking adults and children to make quantity judgments (e.g., “Which room has more furniture?”). Crucially, each context contrasted the number of sub-kinds with the number of individuals (e.g., one room had a greater number of sub-kinds, while the other had more items of furniture but fewer sub-kinds). They found that adults based judgments on function fulfillment (i.e., choosing the group with more variety over the one with more items) around 30% of the time, and that children did so 55% of the time. In a related study, Grimm and Levin (2012; 2017) presented participants with text descriptions of scenes similar to those used by Gordon and Rodman (2006). When asked to judge between 5 items of one kind (e.g., 5 chairs) and 5 items that varied in kind (e.g., a sofa, 2 chairs, a coffee table, and a bookcase) participants unsurprisingly based judgments on the number of sub-kinds (since the two sets were equal in the number of individuals; see Hampton & Winter 2024, who also reported that participants base judgments on sub-kind variety when number cannot be used as a basis for judgment – e.g., 4 small bags of the same shape vs. 4 large bags of different shapes). In two follow-up experiments, Grimm and Levin again presented participants with text descriptions of two sets, one that contained more sub-kinds but fewer items (e.g., a sofa, an easy chair, a coffee table and a small bookcase), and the other that contained more items but fewer sub-kinds (one table and 4 chairs). In these studies, participants again based most of their judgments on number. Importantly, there were more judgments based on sub-kind variety when the context placed a focus on function fulfillment compared to a neutral context. For example, in some cases, like furniture, the majority responded based on sub-kind variety when function fulfillment was emphasized. Finally, Huang et al. (2022) tested contextual effects on quantification by asking Mandarin speakers to make quantity judgments for Mandarin translations of English object-mass nouns (which appear as bare nouns without grammatical markers). In a substance-focused context (e.g., two monsters grinding up objects and eating them), Mandarin speakers made quantity judgments based on volume, but in an individual-focused context (e.g. two fairies competing to create sets of objects), they made quantity judgments based on the number of things.
One question raised by such studies is whether they demonstrate that mass nouns are especially vulnerable to context effects, as proposed by McCawley (1975) and Rothstein (2010). Alternatively, it is possible that object-mass nouns, like count nouns, specify individuation grammatically, but that all nouns can be coerced contextually, including count nouns. For example, although Gordon and Rodman (2006) found that a function-oriented context made some adults base quantity judgments on function fulfillment for object-mass nouns, they also found the identical result for count nouns (e.g., toys, buildings). Similarly, although Hampton and Winter (2024) found that nouns classified as “count” led to more quantity judgments based on number than nouns they classified as “object-mass,” they did not find that mass nouns were more susceptible to contextual cues. Thus, their study supported the conclusion that some mass nouns may not be object-mass (which is uncontroversial), but did not address the mechanism underlying individuation in those mass nouns that did quantify by number. Finally, although Huang et al. (2022) found that participants made judgments based on volume for translations of English nouns like furniture, they did not test whether the same was equally true of words that are count nouns in English. Given this, context effects might be general to both mass and count nouns, and may not weaken the claim that certain nouns (e.g., object-mass nouns or count nouns) specify individuation. This is important, because on most accounts, evidence that pragmatic context can coerce a non-individuated meaning from a count noun would not lead to the abandonment of an individuated semantics for count nouns.
Given these considerations, we conducted two studies that sought to provide a stronger test of the contextual flexibility hypothesis by directly contrasting object-mass nouns with two different kinds of count nouns. In Experiment 1, we asked whether adult participants base quantity judgments on number or variety in contexts that support function fulfillment, and contrasted object-mass nouns to superordinate count nouns (e.g., tools, weapons). In Experiment 2, we tested English and French speakers with words that are object-mass nouns in English, but whose French translations are count nouns (e.g., furniture vs. meubles). If individuation is specified for object-mass nouns, as it is for count nouns, then in both experiments there should be no significant differences due to syntax. In contrast, if mass syntax permits greater contextual flexibility than count syntax, then object-mass nouns should be more impacted by context than count nouns in both studies.
4 General Discussion
In two experiments, we asked whether individuation of object-mass nouns is best explained as the product of grammatical specification, or instead arises from contextual coercion that is specific to the mass category. In particular, we tested whether, for object-mass nouns like furniture, participants prefer to base quantity judgments on the degree to which items fulfill a desired function, like “furnishing” (e.g., McCawley 1975; Gordon & Rodman 2006; Grimm & Levin 2012) or instead make judgments on the basis of number, as they do for related count nouns (e.g., Barner & Snedeker 2005; Bale & Barner 2009). In both experiments, we manipulated function fulfillment by presenting participants with a choice between two sets: one that had a greater number of individuals, and another that had fewer individuals, but a greater number of sub-kinds that would afford greater function fulfillment. We also manipulated whether function fulfillment of the sets was emphasized or not. Using this approach, we found three main results. First, both English and French speakers based quantity judgments overwhelmingly on number. This was despite the fact that the ratio of sub-kinds across each pair of sets (5:1) was much greater than the ratio of individuals (2:1), and despite the fact that participants judged the set with the great number of sub-kinds to better fulfill the intended function than the set with more individuals. Second, syntax did not affect judgments when comparing English object-mass nouns with either English count superordinate nouns (Experiment 1) or French count nouns that were translations of English object-mass nouns (Experiment 2). Finally, while there was some evidence that participants were more likely to base judgments on sub-kind variety when nouns were presented in a function-oriented context, there was no significant difference in how context affected judgments for English object-mass nouns relative to their French count noun translations, or English superordinate count nouns. These results suggest that, although context may sometimes inform how language users make quantity judgments (as per McCawley 1975), participants nevertheless base judgments mostly on number for object-mass nouns and related count nouns. Also, to the extent that context does impact judgments, it does so equally for nouns used in mass and count syntax. Thus, we did not find evidence for the idea that object-mass nouns are more contextually flexible than equivalent count nouns (Rothstein 2017).
As noted in the Introduction, several previous accounts argue that mass nouns like furniture have denotations that include atomic individuals, much like count nouns (e.g., Katz 1970; Bunt 1985; Gillon 1992; 1999; Chierchia 1998; 2010; Barner & Snedeker 2005; 2006; Bale & Barner 2009; Rothstein 2010; 2017; Landman 2020). Most accounts agree that the comparative morpheme more must be underspecified with respect to its dimension of comparison to account for the wide range of possible measurements such as number, mass, length, and volume (e.g., more books, more dirt, more string, more water) as well as abstract dimensions relevant to actions and events (e.g., more walking, more hockey, more hope; see Barner et al. 2008). However, past accounts differ in their treatment of object-mass nouns. According to some, only count nouns strictly force a numerical comparison, and while object-mass nouns may be biased towards numerical comparison, they do not impose it grammatically and are highly sensitive to context (McCawley 1975; Rothstein 2017; Landman 2020). In contrast, others propose that both count nouns and object-mass nouns specify comparison by number grammatically (Barner & Snedeker 2005; Bale & Barner 2009; Wellwood 2015). Consistent with the latter class of accounts, our data reveal that object-mass nouns are no more sensitive to context than count nouns. The contextual sensitivity of measurement may therefore have more to do with the flexible nature of the comparative morpheme than it has to do with the grammar of individuation (see Schwarzschild 2006; Wellwood 2015; Solt 2018; Bale & Schwarz 2020; Bale et al. 2021).
One important question that remains unanswered by this study and others in the literature is why some superordinate nouns that denote collections of individuals are expressed as mass nouns (e.g., furniture, clothing) while others are not (e.g., vehicles, weapons). Relevant to this question, previous studies (Gordon 1985; Barner & McKeown 2005) report that when a novel word is presented in ambiguous syntax, the number of objects referred to impacts whether an object-denoting noun is acquired as mass or count. For example, in a study by Barner and McKeown (2005), when children were told, “Look at the garn” while only one object was present, they often inferred that garn was a singular count noun and later pluralized it when referring to multiple objects of the same type. However, when the same utterance was used to refer to multiple objects, children interpreted it as a mass noun—since it lacked plural morphology—and did not pluralize it when referring to multiple objects. Given the fact that the referents of superordinates like furniture tend to appear together in groups (Wisniewski et al. 1996), when children hear an expression like “look at the furniture,” they might be more likely to interpret it as having plural reference, and therefore being a mass noun. Perhaps children make this inference even if the noun used is a count noun in the speaker’s dialect, and they intended to refer to only one piece of furniture (e.g., as might occur in French). In contrast, if the referents of superordinate terms like instrument are more likely to occur in isolation (e.g., a guitar), then children might be more likely to assume that these expressions are singular count nouns. In sum, our proposal is that syntactic ambiguity combined with referential ambiguity where children could assume that a singular noun refers to a plurality might result in misclassification of count nouns as mass nouns.
Related to this idea, future work should also explore why some languages feature more object-mass nouns than others. Because syntactic ambiguity should generally lead to classification errors, either towards mass or count nouns, cross-linguistic variability in syntactic ambiguity has also been investigated as a predictor of object-mass noun presence. One possibility, proposed by Erbach (2020), is that ambiguity might lead to more count nouns, and to fewer object-mass nouns. This idea is supported by findings from early corpus work (Bloom 1994b) that showed children misclassifying mass nouns as count when they denoted solid physical entities, resulting in errors like “broccolis,” “bacons,” and “furnitures.” However, it is also possible that syntactic ambiguity creates opportunities for children to classify individuating expressions as mass nouns. If this is the case, it may be that languages with richer number morphology are less likely to feature object-mass nouns. For example, in English, which features multiple object-mass nouns, differentiating between mass and count usages often hinges upon a single morpheme (e.g. “Look at the hair/hairs”), and some determiners (e.g., the) are mass–count neutral. By contrast, Italian marks plural information on both determiners and root nouns (e.g., “Guarda la sedia”/“Guarda le sedie”). Because plural information is more salient in Italian, count nouns may be less likely to be mis-classified by children as object-mass nouns, leading to fewer of them in Italian relative to English.10
Finally, in addition to understanding the question of how words might become object-mass nouns, future studies investigating the interface between mass–count syntax and semantics should broaden the scope of words included in research on this question. For example, whereas superordinate object-mass nouns like furniture are somewhat rare and idiosyncratic, in some languages, like English, mass nouns that refer to individuated phenomena can be freely generated via processes of nominalization from verbs that denote punctual events (e.g., too much kicking/jumping). Such cases are important because they allow additional ways to directly control lexical semantics while manipulating syntax (e.g., to compare jumps vs. jumping), and because they allow an abstraction away from the particularities of words like furniture. While work in English has found that deverbal nouns (e.g., a lot of jumping) quantify by number to the same extent as their count counterparts (e.g., a lot of jumps; Barner et al. 2008), future studies should explore this phenomena from a cross-linguistic perspective, and in language acquisition.
In summary, we presented two studies measuring English and French speakers’ quantity judgments of English object-mass nouns and collective count nouns. We found that in both languages, participants made quantity judgments mostly by number of individuals as opposed to number of kinds, and that there was no effect of syntax on quantity judgments (within English and cross-linguistically). These findings are compatible with the hypothesis that some mass nouns lexically specify individuation, and are no more susceptible to effects of context than similar count nouns. Therefore, while all count nouns individuate, mass nouns can refer to either individuated (object-mass nouns) or non-individuated (substance-mass nouns) phenomena.