The interpretation of animate nouns in child and adult Mandarin: from the Universal Grinder to syntactic structure

1 Introduction

Animate nouns are used to refer to entities in the world. Although we use these nouns effortlessly, their use involves a complex interaction of ontology, cognition and language. The present study investigates how animate nouns are interpreted by Mandarin-speaking children and adults. Our focus is their status of animate nouns as mass nouns versus count nouns. Animacy plays an important role in determining the meaning of nouns, such that nouns with higher animacy status are more likely to be assigned a count interpretation than ones with lower animacy status (Corbett 2000; Grimm 2018; Smith-Stark 1974). In two seminal papers, Pelletier (1975, 2012 proposed that even highly animate nouns nevertheless convey a mass interpretation when they are used in contexts that are felicitous for this interpretation. Introducing a thought experiment, Pelletier (1975) asks us to imagine a device, called a Universal Grinder, which can grind individual animate entities (e.g., a man) into substances (e.g., human flesh). He concludes from the thought experiment that the mass interpretation is readily available for these nouns, even if they are more typically assigned count interpretations in ordinary contexts. Pelletier (1975, 2012 concludes that (most) nouns are encoded in the lexicon with both a mass sense and a count sense.

In contrast to Pelletier’s conception of a Universal Grinder, this imaginary device has been cited in the literature as a way of illustrating how a mass interpretation can be “coerced” from a count noun (see e.g., Cheng et al. 2008; Chierchia 2010; Rothstein 2017: Ch. 7). The present study focuses mainly on an account proposed in Cheng et al. (2008). On that account, animate nouns are lexically specified as count nouns in both English and Mandarin, based on their ontological properties (Cheng and Sybesma 1998, 1999; Doetjes 1997). These authors contend that, in English, a mass interpretation of animate nouns can arise via “morpho-syntactic coercion” as witnessed in sentences like “There is dog all over the floor.” In Mandarin, however, nouns are “immune to morpho-syntactic coercion” according to Cheng et al. (2008: 53), because of the lack of number marking. On their account, a possible mass interpretation of Mandarin count nouns must be coerced pragmatically. In the case of animate nouns, however, pragmatic coercion is not operative, due to a lexical blocking effect. Therefore, Cheng et al. argue that Mandarin speakers only access the basic count interpretation of Mandarin animate nouns.

The present study investigates the role of morphosyntax and the role of context in the interpretation of animate nouns, to adjudicate between the competing accounts of the interpretation of animate nouns in Mandarin. In Mandarin, animate nouns can be used in a ‘bare’ nominal form, i.e., with no grammatical marker indicating their count/mass interpretation. The present study uses bare nouns to examine the effects of contextual information on interpreting animate nouns. In addition, animate nouns in Mandarin can appear with an individual classifier, allowing us to investigate the influence of the morphosyntax on their interpretation. To assess the possible role of coercion, we assess Mandarin-speaking children’s interpretation of animate nouns. It has been well documented that children younger than seven years old are less sensitive than adults to certain pragmatic norms that govern adult usage (see e.g., Chierchia et al. 2001; Noveck 2001). Preschool children, therefore, are less likely than adults to derive mass interpretations of animate nouns, if pragmatic coercion is required, as proposed by Cheng et al. (2008). Putting it the other way around, the finding that preschool children readily access mass interpretations of animate nouns would be circumstantial evidence that mass interpretations are encoded as part of a noun’s meaning in children’s lexicons and, by parity of reasoning, in the lexicons of adults.

The previous literature includes several studies investigating the interpretations assigned to nouns by Mandarin-speaking children and adults, including mass versus count interpretations (e.g., Cheung et al. 2009, 2012; Huang et al. 2021; Lin and Schaeffer 2018; Liu 2014). However, few previous studies included animate nouns. For example, the study reported in Lin and Schaeffer (2018) included only one animate noun, viz., gou ‘dog’. In their ‘count noun condition’, Lin and Schaeffer tested gou ‘dog’ and six inanimate nouns (e.g., qiu ‘ball’, qianbi ‘pencil’) in their bare forms (what they called ‘unclassified nouns’). They presented these nouns in neutral contexts. They report that the participants assigned a mass interpretation to these bare nouns less than 25 % of the time. It remains an open question, therefore, how morphosyntax and contextual information affect the interpretation of animate nouns in Mandarin, either by children or adults.

The present study implements a real-world version of Pelletier’s imaginary Universal Grinder. We propose a straightforward extension of the account proposed by Pelletier (1975, 2012, applied to Mandarin. Following Pelletier, we suggest that (most) animate nouns in Mandarin are encoded in the mental lexicon with both a count sense and a mass sense. Moreover, we propose that the interpretation of Mandarin animate nouns depends on the syntactic structure of the sentences in which they are used. To be more specific, the count interpretation of Mandarin animate nouns is directly established by the morphosyntax (i.e., the presence of an individual classifier). The specific interpretation that is assigned is also influenced by contextual factors, but contextual effects are not expected to override the interpretations that are derived through grammatical processes.

The present study reports the findings of experiments that were designed to empirically test between the account advanced in Pelletier (1975, 2012, extended to Mandarin, and the account proposed by Cheng et al. (2008). Four experiments investigate how Mandarin-speaking adults and 4-to-6-year-old Mandarin-speaking children interpret animate nouns. We assessed participants’ interpretation of these animate nouns using a Truth Value Judgment task (Crain and Thornton 1998). To anticipate the experimental findings, the patterns of responses by Mandarin-speaking children and adults are consistent with the account we propose, and resist explanation on the account proposed by Cheng et al. (2008). The findings indicate that both a mass interpretation and a count interpretation are readily assigned to animate nouns by Mandarin-speaking children, as well as by adult speakers of Mandarin. The mass interpretation of Mandarin animate nouns does not appear to require assistance from a Universal Grinder, either real or imagined. The findings also confirmed our proposal that both morphosyntax and contextual information affect the interpretation of Mandarin animate nouns, depending on the specific syntactic structure of the sentences that contain them. However, the findings reveal that the morphosyntax (the individual classifier) is the more decisive factor for both children and adults. The manipulation of the nonlinguistic context did not suffice to override the morphosyntax for either group of participants.

The remainder of the paper is structured as follows. Sections 2 and 3 review the concept of the Universal Grinder as characterized in Pelletier (1975) and Cheng et al. (2008), Section 4 introduces our alternative account of Mandarin animate nouns, Sections 5 and 6 describe the experimental investigations, Section 7 contains the general discussion, and Section 8 concludes the paper.

2 The Universal Grinder (Pelletier 1975)

Pelletier’s (1975) hypothetical Universal Grinder was devised to represent a felicitous context to facilitate the mass interpretation of nouns that are typically assigned count interpretations in English. The storyline is this: an individual entity is inserted into one end of the Universal Grinder, and a substance emerges at the other end. Suppose, for example, that a man is inserted into one end of the Universal Grinder, and a pile of human flesh emerges at the other end. According to Pelletier, when the transformation is complete, it is fully natural for someone to utter the sentence “There’s man all over the floor.” In Pelletier’s account, the Universal Grinder represents just “one of the possible normal answers to use the mass sense of our ‘normal’ count noun” (p. 456, his emphasis), and “there will be various occurrences of ‘man’ that one has to understand in a mass sense” (p. 458). According to Pelletier (1975), the upshot of this thought experiment is that nouns are not classified as either count or mass. Rather, he states: “we need to distinguish not between mass and count nouns but between mass and count senses of nouns” (p. 452, italics original). He continues: “…a mass or count sense of a word exist if one can describe a circumstance or set of circumstances in which that word with the requisite sense can (or would) be normally employed (p. 456)”. From this perspective, the count sense of a noun arises when it refers to “a naturally-constituted object (i.e., structured according to biological, geological, or cultural norms, etc.)”, whereas the mass sense arises when it refers to “the matter of which it (the object) is made” (p. 456).

In Pelletier’s (1975) account, the intended sense of a noun can also be made explicit using morphosyntactic properties of the phrase that contains the noun. This is illustrated by the minimal pair of sentences, “I like chickens” (the count sense of chicken) versus “I like chicken” (the mass sense) (p. 456). Even nouns that do not denote anything in the real world have both a count sense and a mass sense. To show this, Pelletier asks us to (counterfactually) apply the hypothetical Universal Grinder to the noun unicorn: “If there were any unicorns and if we were to put one into the grinder, there would be unicorn all over the floor.” Pelletier concludes that “…nothing is immune from the grinder treatment (p. 457)”. He takes his account to be the same as that proposed in Gleason (1965: 136–137), which states that “every noun, given the right context, can occur in either types of usage, count or mass.”^[1] As both Pelletier and Gleason acknowledge, one of the senses of a noun may be much more frequent than the other, but such frequency effects are overridden in “sufficiently unusual situations” such as the Universal Grinder.^[2]

According to Pelletier (2012), lexical items have both count and mass senses in all languages. A specific sense is “introduced in the construction of larger phrases” (p. 19). Typologically different languages use different morphosyntactic devices to specify count or mass senses of nouns. In number-marking languages like English, the count versus mass sense of nouns is indicated using plural morphology (-s) and by determiners (several), for example. In classifier languages like Mandarin, classifiers specify the syntactic count and mass status of nouns at the level of classifier noun phrases. Languages that lack both count-mass syntax and classifiers (e.g., Yudja) rely solely on contextual information to determine the interpretation of nouns.

3 The Universal Grinder (Cheng et al. 2008)

In contrast to Pelletier’s use of the Universal Grinder, this device has been appropriated in the literature to illustrate one way in which a mass interpretation can be “coerced” for a noun whose basic meaning is that of a count noun (Cheng et al. 2008; Chierchia 2010; Rothstein 2017: Ch. 7). In particular, Cheng et al. (2008) invoke the concept of the Universal Grinder to describe how the mass interpretation of count nouns can be coerced in typologically distinct languages, including English and Mandarin. They postulate two different “triggers for the Universal Grinder.” One of these factors, the morphosyntax, is operative in English, but not in Mandarin. As Cheng et al. note, English marks its count nouns, for example, using plural morphology and the indefinite determiner (Cf. Cheng and Sybesma 1998, 1999). Without these markers, the use of a count noun like dog is unacceptable, unless it is interpreted as a mass noun. The mass interpretation is taken to be a last-resort strategy, which is invoked to avoid producing an anomalous utterance (cf. Doetjes 1997). This strategy is used to explain the mass use of the count noun dog in English sentence (1) below: dog in this sentence lacks a count grammatical marker; hence it is forced to be mass to avoid ungrammaticality.

Cheng et al. (2008: 53) argue that, in contrast to English, Mandarin is “immune to morpho-syntactic coercion”, since count nouns in this language do not necessarily require a grammatical element to mark their count status. Instead, pragmatic coercion via the Universal Grinder is argued to apply in Mandarin. This concerns the second trigger for the Universal Grinder in Cheng et al.’s account. They contend, moreover, that pragmatic coercion is only possible for some count nouns. To cite one of their examples, pingguo ‘apple’ is a count noun denoting individual apples, as in example (2) (Example (9) in Cheng et al. 2008), but the mass interpretation of pingguo can be coerced in a context in which chopped apple is a key ingredient in a salad, as in Example (3) (= Example (14) in Cheng et al. 2008).

However, on their account, pragmatic coercion fails to trigger the mass interpretations of some animate nouns, including gou ‘dog.’ They claim that this animate noun can only be used to convey a count interpretation. They offer example (4) to illustrate the absence of coercion for gou ‘dog.’ In the context for (4), there is wall-paper covered with pictures of dogs (example 1a in Cheng et al. 2008).

To express a mass interpretation of animate nouns such as gou ‘dog’, the compound noun gou-rou ‘dog meat’ must be used, as in (5) (Example (1b) in Cheng et al. 2008). The availability of both gou and gou-rou in the Mandarin lexicon creates a blocking effect that pre-empts the mass interpretation of gou ‘dog’ in sentences like (4).

To recap, we have reviewed two accounts of the interpretation of nouns. One account, by Pelletier (1975, 2012, argues that (most) nouns are encoded in the lexicon with both a mass sense and a count sense. The mass interpretations of nouns that are typically used with their count sense can be revealed in an “appropriate context” that makes the mass sense “highly salient,” including the Universal Grinder. In languages like Mandarin, classifiers determine the count and mass status of nouns. The second account, by Cheng et al. (2008), argues that languages like English use morphosyntactic markers to coerce a mass interpretation of count nouns. In Mandarin, morphosyntactic coercion is not available, so only pragmatic context can coerce mass interpretations from count nouns. However, the role of pragmatic context is limited, due to blocking effects. Most animate nouns unambiguously designate a count interpretation, due to the availability of an explicit lexical alternative, a compound noun, that designates the mass interpretation.

4 Grammatical and contextual factors affecting the interpretation of Mandarin animate nouns

In this section, we discuss the availability of count and mass interpretations of Mandarin animate nouns. We agree with Pelletier on two things. First, lexical items have both count and mass senses (Pelletier 1975, 2012). Second, a specific sense is introduced in the construction of larger phrases. In Mandarin, classifiers specify the syntactic count and mass status of nouns at the level of classifier noun phrases (Pelletier 2012). Moreover, we will consider the role played by contextual information for the specification of countability in the interpretation of Mandarin nouns, a factor that is not taken into account in Pelletier’s (2012) characterization of Mandarin nouns. We also explore how morphosyntax and contextual information interact to affect the interpretation of Mandarin animate nouns. In this regard, we aim to provide an extension of the account proposed by Pelletier (1975, 2012. Let us elaborate on the details below.

We examine the countability of Mandarin animate nouns in a comparative structure (A has more N[s] than B). According to Bale and Barner (2009: Section 3.3), the comparative structure is “the most reliable way of assessing the count-mass semantics” (see also Bale and Barner 2018; Barner and Snedeker 2005, 2006; Gathercole 1985; Jespersen 1924). To illustrate, consider Example (6) (Example (13) in Bale and Barner 2009: 225). In (6a), the quantity judgment is based on the cardinality of the noun referent, i.e., the number of rocks. By contrast, the quantity judgment is based on a non-cardinality dimension in (6b), i.e., the overall amount of rock.

To see the distinct interpretations of rock in (6), let us imagine a context that should, in principle, allow both a mass interpretation and a count interpretation of rock. Suppose the context has two characters, Seymour and Esme. Suppose further that Seymour has four tiny rocks and Esme has two large rocks. Sentence (6a) is judged to be a true description of this context, whereas (6b) is judged to be false. The examples in (6) indicate that the same real-world entities can be designated via distinct dimensions, corresponding to their perceptual/conceptual properties. These properties include their cardinality, volume, weight, substance, and amount (Schwarzschild 2006). When a cardinal dimension is specified in one kind of comparative structure, more Ns, a count interpretation is accessed. By contrast, when a non-cardinal dimension is specified in the other form (more N), a mass interpretation is accessed (cf. Bale and Barner 2009; Beviláqua et al. 2016; Rothstein 2017). Nouns with dual (count and mass) uses are widespread in English (Borer 2005: Section 4.2; Pelletier 2012; Ware 1979). Indeed, as Ware (1979: p. 390) puts it, “Everything can be individuated, divided into parts, and everything can be measured en masse. Linguistic conventions can establish the appropriateness of count and mass occurrences, depending on our interests and practice”.

Ways of specifying ontological dimensions vary across languages. In English, nouns are marked with either count or mass syntax. In Mandarin, however, nouns can appear in a bare nominal form, without any grammatical marker indicating countability. In such linguistic contexts, Mandarin nouns are inherently ambiguous. Consider sentence (7), where the animate noun gou ‘dog’ appears with the comparative expression gengduo ‘more’, a comparative structure in Mandarin.

In (7), the bare noun gou ‘dog’ can have either a count reading (‘Big Bird Monster ate more dogs’) or a mass reading (‘Big Bird Monster ate more dog’). When gou conveys the count reading, the sentence indicates a quantity judgment based on number (a cardinal dimension). When gou conveys the mass reading, the quantity judgment is based on the overall amount of dogs (a non-cardinal dimension). These alternative readings represent two distinct dimensions of measurement in the same entities. The alternative (count and mass) interpretations of Mandarin bare nouns are triggered by distinct contexts, where the particular interpretation that is assigned depends on the selection of a dimension: e.g., the number of dogs (a cardinal dimension) versus a measure of their weight (a non-cardinal dimension). The mass interpretation of animate nouns, such as gou ‘dog’, is not limited to circumstances in which the noun referent is a substance, i.e., dog meat. On our account, there is no role for blocking effects in the interpretation of Mandarin animate nouns. Our account differs, therefore, from the account proposed by Cheng et al. (2008).

In addition to use as bare nouns, Mandarin nouns can also co-occur with classifiers (e.g., Borer 2005; Huang 2009; Huang and Lee 2009; Li 2013; Zhang 2013). The semantic function of individual classifiers is similar to that of plural morphology in English (Borer 2005; Doetjes 2021a, 2021b).^[3] In addition, classifiers indicate particular dimensions of noun referents and exclude other dimensions. As a consequence, the meaning of classifier-noun combinations cannot be changed by contextual manipulation (cf. Borer 2005: Section 4.2 on the relationship between morphosyntax and contextual information). Consider Example (8), which differs from (7) only by the addition of the individual classifier zhi. The presence of zhi entails that only the cardinal dimension is relevant for interpretation. Consequently, only the count reading (‘Big Bird Monster ate more dogs’) is possible (e.g., Bale and Barner 2009; Borer 2005; Huang 2009; Huang and Lee 2009; Li 2013; Pelletier 2012; Sharvy 1978).

So, our account of animate nouns in Mandarin is directly in line with the proposal by Pelletier (2012), which supposes that nouns are lexically specified with both a mass interpretation and a count interpretation. This feature of Pelletier’s proposal can be directly extended to Mandarin bare nominals, which lack a grammatical marker to specify countability. We contend that Mandarin bare nouns can be readily assigned both count and mass interpretations in neutral contexts. That is, Mandarin bare animate nouns are inherently ambiguous. In specific contexts, bare nouns are assigned count interpretations when they are used to refer to the cardinal dimensions of the entities they designate, and they are assigned mass interpretations when they are used to refer to the non-cardinal dimensions of the entities they designate. By contrast, when an individual classifier accompanies an animate noun, it is assigned a count interpretation.

To summarize, our account does not need to resort to the Universal Grinder to trigger the mass interpretation of Mandarin animate nouns. We show that the interpretation of Mandarin animate nouns depends on the specific syntactic structure of the sentences in which they appear. We also show that both the morphosyntax and the pragmatic context exert an influence on the interpretation that is assigned to sentences with animate nouns in Mandarin. Adopting a modular model of the language apparatus, first advocated by Fodor (1983), we propose that when both contextual and morphosyntactic features are present, the morphosyntax prevails. Because syntactic structure is input to the semantic interpretation, and not vice versa, contextual effects are not expected to hold sway over interpretations that have been previously derived by grammatical processes. The modular architecture of the language apparatus extends to child language learners, as well as to adult speakers, so morphosyntactic properties are expected to preempt contextual information for children, as well as for adults (Crain and Thornton 1998; Crain and Wexler 1999). With our account and its predictions set out, we now turn to the experiments.

5 Research issues

The present study has two aims. The first is to determine whether or not Mandarin animate nouns allow a mass interpretation. In this regard, some recent empirical research indicates that Mandarin-speaking children and adults assign both count and mass interpretations to bare nouns, despite the preferences for one interpretation or the other for particular nouns (Huang et al. 2022; Lin and Schaeffer 2018). For example, research indicates that the mass interpretation is assigned to bare nouns like jiaju ‘furniture’ and qiu ‘ball’ in appropriate contexts, even though these nouns have been taken to be count nouns in the literature (e.g., Liu 2014). The present study empirically investigates the accessibility of mass interpretations for Mandarin animate nouns, by both children and adults.

Assuming that mass interpretations of Mandarin animate nouns are accessible to children and adults, the second aim of the present study is to determine whether the mass interpretation is also encoded in the lexicon, alongside the count interpretation. To address these research questions, we investigated the interpretation assigned to Mandarin animate nouns by children and adults using two different contexts. One context introduced a real-world Universal Grinder, a scenario in which we introduce a device that grinds individuals into substances. A second context lacked a Universal Grinder scenario. Experiments were designed to determine whether such contextual support for a mass interpretation is a necessary condition for child and adult Mandarin speakers to access this interpretation of animate nouns. Evidence that mass interpretations of animate nouns are accessed in the absence of a Universal Grinder scenario would invite the conclusion that the mass interpretation is part of the lexical specification of animate nouns.

Why test young Mandarin-speaking children? We contend that children offer an ideal vehicle for adjudicating between the alternative theories about the basic meanings of animate nouns. Based on findings from recent research, it is plausible to suppose that preschool children are less likely than adults to use coercion to access an interpretation that is not immediately available to them; they may lack the cognitive resources to compute meanings for animate nouns that take them beyond those that are basic/literal (cf. Chierchia et al. 2001; Noveck 2001). The proposal by Cheng et al. (2008) argues that the basic meaning of animate nouns is a count meaning, such that a mass reading requires coercion. Therefore, this proposal leads to the prediction that young children will be unable to access mass readings or will access them significantly less often than adults do. On the other hand, if both count and mass readings are available for animate nouns, as we and others propose, then this predicts that child participants will be the equals of adults in accessing mass readings for animate nouns, since mass and count meanings should both be accessible to child language learners, all else being equal. In summary, child language offers a window to the basic meanings of lexical expressions, without the additional nuances in meaning that are contributed by a speaker’s real-world knowledge and conversational norms.

6 Experiments

The present study consists of four experiments. Experiments 1 and 2 investigate how morphosyntactic and contextual factors affect the interpretations assigned to animate nouns by Mandarin-speaking adults and by Mandarin-speaking 4-to-6-year-old children. To assess the influence of morphosyntactic information, two kinds of test sentences were presented to participants. The test sentences formed minimal pairs. One sentence type contained an individual classifier (Experiment 2), whereas the classifier was absent in the other sentence type (i.e., bare animate nouns in Experiment 1). To assess the effects of contextual information, both sentence types were presented to participants in different contexts. One context favored the count interpretation of the noun. We call these individual-oriented contexts. The second context favored the mass interpretation. We call these amount-oriented contexts.

Experiments 3 and 4 investigate whether a Universal Grinder scenario is necessary for the derivation of a mass interpretation for bare animate nouns in Mandarin. In Experiment 3, individual entities were physically transformed into substances by a Universal Grinder. Experiment 4 was a control experiment, with no physical transformation of entities associated with the Universal Grinder. A comparison of the mass interpretation assigned by participants in these two experiments and that in Experiment 1 (with no Universal Grinder) should enable us to determine whether a mass interpretation is encoded in the lexicon, as part of the basic meaning of animate nouns in Mandarin.

Table 1 outlines the four experiments, including a summary of the predictions of the alternative theoretical proposals under consideration.

As Table 1 indicates, our account predicts that bare animate nouns will be assigned either the count interpretation or the mass interpretation and, depending on the context, without requiring assistance from a Universal Grinder (Experiments 1, 3 and 4). In the presence of an individual classifier, only the count interpretation is expected to be assigned to animate nouns on our account (Experiment 2). By contrast, the account proposed by Cheng et al. (2008) supposes that Mandarin animate nouns have basic count meanings, so the derivation of a mass interpretation requires pragmatic coercion. Due to the effects of lexical blocking, participants are expected to only assign a count interpretation to bare nouns; the addition of an individual classifier should not make any difference in the quantity judgments assigned by the participants.

6.1 Experiment 1

Experiment 1 examined whether Mandarin-speaking children and adults assign count and mass interpretations to bare animate nouns in two distinct contexts: individual-oriented contexts and amount-oriented contexts.

6.1.1 Participants

The child participants were twenty-four 4-to-5-year-old Mandarin-speaking children. The child participants ranged in age from 4;7 to 5;6, with an average age of 5;2. This is the youngest age that can be tested with a Truth Value Judgment Task we adopt in the present study, as will be introduced shortly. The child participants were recruited from a university kindergarten in Jiangsu Province, P.R. China. There was also a control group of twenty-four adult participants. The adults ranged in age from 18 to 24 years old and were undergraduate and postgraduate students from the same university.

6.1.2 Experimental method and procedures

We used a Truth Value Judgment task (TVJT, Crain and McKee 1985; Crain and Thornton 1998) that involved judgments about quantity (Barner and Snedeker 2005). The task had two experimenters. One experimenter narrated a series of stories using toys and props. The other experimenter played the role of a puppet who watched the stories alongside the child participant. Following the completion of each story, the puppet was asked to explain to the child what happened in the story. The puppet’s explanation concluded with a test sentence. The test sentences were statements like ‘A ate more N than B’. The child participants’ task was to judge whether or not the puppet’s statement was right or wrong. The child was instructed by the experimenter to reward the puppet with a strawberry if it was right. The experimenter told the child that sometimes the puppet wouldn’t pay close attention to the stories, and would make a false assertion about the events that took place in the story. In this case, the child participants were instructed to give the puppet a red pepper, to remind it to pay closer attention to the stories that followed.

The TVJT differed in one important aspect from the question-answering judgment task used in previous research on quantity judgments (see e.g., Barner and Snedeker 2005). The question-answering task asks participants to choose one character or the other in response to questions like “Who has more rock/rocks?”. As discussed in Huang et al. (2022), the results of this experimental procedure indicate a preference for one interpretation of potentially ambiguous sentences but do not suffice to show that a test sentence does not permit further interpretations, beyond the preferred interpretation (see Crain and McKee 1985). The TVJT was devised in the 1980s to enable researchers to directly test for the availability of specific interpretations of sentences, by presenting the same test sentences in different contexts (Crain and Thornton 1998).

Experiment 1 used a within-subject design, so each participant was tested in both conditions. Before the testing session, each participant was given two practice trials to familiarize themselves with the task. Only those participants who correctly responded to both practice trials proceeded to the main testing session. The same experimental methods and procedures were used in all four experiments. Both the child and adult participants were tested individually, using the same experimental procedures and test materials.

6.1.3 Test sentences, test conditions and test materials

The test sentences contained a bare animate noun. The participants were requested to make a comparative judgment about quantity. That is, the participants were asked to judge the truth or falsity of sentences that contained the determiner gengduo ‘more’ followed by a bare animate noun, as illustrated in (7), repeated here as (9).

(9)

Daniao	yaoguai	chi	le	gengduo	gou
Big-Bird	Monster	eat	Asp	more	dog
(i) ‘Big Bird Monster ate more dog.’
(ii) ‘Big Bird Monster ate more dogs.’

There were two conditions. Condition 1 presented test sentences in an amount-oriented context. This context consisted of eating scenarios that compared the amount of food that was consumed by the animal characters. This condition was inspired, in part, by Gleason’s (1965) eating example (see Section 2, Footnote 1). On a typical trial, there were two monsters: Frog Monster and Bird Monster. In the story, both monsters will eat just about anything. One day, Bird Monster caught two big dogs weighing 100 jin in total (jin is a unit of measure used in China, One jin equals 500 g). Bird Monster ate them both and became very full. On the same day, Frog Monster caught four small dogs weighing a total of 10 jin. He also ate them all, but he was still hungry. The amount of dogs is the most significant dimension of measurement in the story. This result is achieved by highlighting the distinct effects caused by the difference in the amount eaten by the two monsters. That is, the large amount of dogs made Bird Monster full, and the small amount did not help satisfy Frog Monster’s hunger. By doing this, we aimed to draw the participants’ attention to the number of dogs. The English translation of the story script is given below. The last scenario of the amount-oriented context is shown in Figure 1.

Figure 1:

The last scenario of Experiment 1 (the amount-oriented context).

Two monsters were living in a forest, Bird Monster and Frog Monster. One day, they were playing in the forest. After a while, they were hungry. So they decided to go find something to eat. Then, they found a group of dogs. Bird monster caught two big dogs weighing 100 jin in total. Frog Monster caught four small dogs weighing 10 jin in total. Bird Monster ate the two big dogs and he became full and his stomach was bulging. Frog Monster ate the four small dogs but he was still hungry.

Following the story, the experimenter asked the puppet who ate gengduo gou ‘more dog(s).’ The puppet answered the question using sentence (9). Sentence (9) asserts that Bird Monster (who ate two big dogs) ate gengduo gou than Frog Monster (who ate four small dogs). See the Appendix for the arrangement of the test sentences and filler sentences.

The participant’s task was to judge whether the puppet’s statement was a true or false description of the story. Participants who accepted the sentence were taken to base their quantity judgment on the amount that was consumed, i.e., the mass interpretation. Participants who rejected the sentence were taken to base their judgment on the number of dogs that were consumed, i.e., the count interpretation. Thus, we make both the count reading and the mass reading available in our experimental design, and the acceptance or rejection of the test sentences can tell which reading is preferred in this specific context.

Condition 2 introduced the individual-oriented contexts by highlighting the cardinal quantity of entities. On a typical trial, Butterfly Fairy and Bird Fairy had a magic competition. Using her magic skills, Butterfly Fairy made two big dogs (weighing 100 jin in total) appear from nowhere. Showing off her own magical prowess, Bird Fairy made four small dogs (weighing a total of 10 jin) appear from nowhere. Bird Fairy won the competition and was awarded a gold medal. Butterfly Fairy received a black cross because she lost the magic competition. The English translation of the story’s script is given below. The last scenario of the individual-oriented context is shown in Figure 2.

Figure 2:

The last scenario of Experiment 1 (the individual-oriented context).

Two fairies were living in a magic forest, Butterfly Fairy and Bird Fairy. One day, they had a magic competition. Using their magic, Butterfly Fairy made two big dogs weighing 100 jin, and Bird Fairy made four small dogs weighing 10 jin. Bird Fairy won the competition for a gold medal, and Butterfly Fairy only got a black cross.

In this scenario, the number of dogs created by the two fairies determines which fairy can get a reward, and cardinality is thus the most important information for the quantity judgment. The salient information on cardinal quantity is reinforced using a gold medal versus a dark cross.

Following the story, the puppet asserted the test sentence in (10), which asserts that Butterfly Fairy created gengduo gou than Bird Fairy.

(10)

Hudie	xiannv	bian	le	gengduo	gou
Butterfly	Fairy	create	Asp	more	dog
(i) ‘Butterfly Fairy created more dogs.’
(ii) ‘Butterfly Fairy created more dog.’

If participants made their quantity judgment based on cardinality, then they were expected to reject sentence (10) in Condition 2, because Bird Fairy created more dogs than Butterfly Fairy. If participants made their quantity judgment based on amount, then they were expected to accept (10), because the overall amount created by Butterfly Fairy was greater than that created by Bird Fairy. Again, both the count reading and the mass reading were made available in our experimental design, and the acceptance or the rejection of the test sentences could indicate which reading was preferred in this context.

In the account we offered, the mass interpretation is expected to be preferred in the amount-oriented condition and the count interpretation is expected to be preferred in the individual-oriented condition. By contrast, in the account by Cheng et al. (2008), the count interpretation is expected in both conditions.

The test session consisted of two stories, and each story contained three target sentences. One was an amount-oriented context, and the other was an individual-oriented context. The same test sentences were presented in both conditions, including the bare animate nouns gou ‘dog’, niu ‘cow’, and yang ‘sheep’. In one story, Frog Monster and Bird Monster found gou ‘dog’ for breakfast, niu ‘cow’ for lunch, and yang ‘sheep’ for dinner. In the other story, Butterfly Fairy and Bird Fairy had a magic competition that consisted of three trials, where they tried through magic to create gou ‘dog’, niu ‘cow’, and yang ‘sheep’. Altogether, then, each participant heard 6 test sentences. There were 24 participants, making a total of 72 responses in each condition. The order of narration of the two stories was counterbalanced, and a typical testing session lasted about 12–15 min.^[4]

In addition to the test sentences, the puppet also produced two filler sentences in each trial. The filler sentences were presented before and after each test sentence. The filler sentences were clearly true or false and served to obscure the purpose of the study to ensure that the participating children enjoyed themselves (Crain and Thornton 1998) (see the Appendix for the arrangement of the test material).

6.1.4 Findings

The adults and the children exhibited the same pattern of responses in both conditions. In the amount-oriented condition, the child participants accepted the test sentences and assigned the mass interpretation 75 % of the time (54/72 trials) and the adult participants accepted them 71 % of the time (51/72 trials). A Mann-Whitney test revealed no significant difference between groups (Z = 0.205; p = 0.838). In the individual-oriented condition, there were 100 % rejections by both groups of participants, assigning the count interpretation. Whenever participants rejected the test sentences, they were asked to explain to the puppet “what really happened in the story?” These justifications clearly indicated that the participants had been assigned a count interpretation. For instance, in justifying their rejections of Example (10) Hudie xiannv bian le gengduo gou ‘Butterfly Fairy created more dog(s)’, the child participants said that (i) Butterfly Fairy only created two dogs while Bird Fairy created four dogs, or (ii) they said that Bird Fairy created more dogs. In the first type of justification, participants compared the distinct numbers (four versus two) of animal characters that were created by the two fairies. In the second type of justification, participants offered a correct way of describing the story. In addition, some adults offered a third type of justification, i.e., they explicitly pointed out that their quantity judgment was based on the cardinal dimension.^[5] The adult and child data are summarized in Figure 3.

Figure 3:

The experimental results in Experiment 1.

The experimental findings indicate that the majority of both the child and adult participants accessed both count and mass interpretations of Mandarin bare animate nouns. Therefore, the experimental data provide compelling empirical support for our hypothesis that both count and mass interpretations are part of the basic meaning of bare animate nouns in Mandarin. The data are not consistent with the account by Cheng et al. (2008), which supposes that only the count interpretation is possible for Mandarin animate nouns.

Note that, despite the presentations of the test sentences in an amount-oriented context (Condition 1), specific subgroups of participants consistently assigned a count interpretation to the bare animate nouns in the test sentences (29 % in adults, and 25 % in children). These subgroups consisted of 6 of the 24 child participants and 7 of the 24 adult participants. The justifications offered by these participants made it clear that they had attended to the cardinality of the nouns. Because this pattern of responses also appeared in Experiments 3 and 4, we will postpone a discussion of this response pattern to the General Discussion in Section 6.

6.4 Experiment 3

Experiment 3 investigated the interpretation of bare animate nouns assigned by Mandarin-speaking children and adults in another amount-oriented context, one involving a Universal Grinder. The experimental design of Experiment 3 differed from that of the amount-oriented condition of Experiment 1 in only one aspect: the addition of a Universal Grinder scenario. In a typical trial, Bird Monster and Frog Monster will eat anything, as before. But, in the context of this experiment, Bird Monster and Frog Monster have no teeth, so they have to grind their food before eating it. One day, Bird Monster caught two big dogs weighing 100 jin total. He ground the dogs into a big pile of food and ate it up. When he was done eating, he was very full. On the same day, Frog Monster caught four small dogs weighing 10 jin total. He ground his dogs into a small pile of food and ate it up, but he was still hungry. As in Experiment 1, the amount of ground food was salient in the story, because eating a large amount of food made Bird Monster full, whereas Frog Monster was still hungry after eating a small amount of food. The last scenario of the individual-oriented context is shown in Figure 4.

Figure 4:

The last scenario of Experiment 3.

After the story ended, the experimenter asked the puppet who ate gengduo gou, and the puppet answered by asserting that Bird Monster ate gengduo gou than Frog Monster, as in (13).

(13)

Daniao	yaoguai	chi	le	gengduo	gou
Big-Bird	Monster	eat	Asp	more	dog
(i) ‘Big Bird Monster ate more dog.’
(ii) ‘Big Bird Monster ate more dogs.’

The task was to judge whether the puppet’s statement was a true or false description of the story. The participants were twenty-four Mandarin-speaking children who ranged in age from 4;8;4 to 5;7;27, with an average age of 5;2;4. There was also a control group of twenty-four adults. Both the child and adult participants were from the same kindergarten and university as those who had participated in Experiments 1 and 2, but none of them participated in the earlier experiments. The same three animate nouns, gou ‘dog’, niu ‘cow’, and yang ‘sheep’ were used in the test story, and each participant was presented with 3 test sentences, making a total of 72 responses. As in Experiment 1, filler sentences were interspersed among the test sentences, but since there was only one test condition, a testing session only lasted about 7 min for each participant.

In Experiment 3, the child participants accepted the test sentences and assigned the mass interpretation 95 % of the time (69/72). The adult participants accepted them less often, 75 % of the time (54/72). In this experiment, only 1 of the 24 child participants consistently rejected the test sentences whereas, similar to Experiment 1, 6 of the 24 adult participants produced this pattern of responses. In short, Mandarin-speaking children consistently accessed the mass interpretation in the Universal Grinder scenario, whereas a subgroup of adult participants continued to access the count interpretation. Figure 5 provides a summary of the experimental findings.

Figure 5:

The experimental results in Experiment 3.

Comparing the findings of Experiment 3 and Experiment 1, it appears that the presence of ground substances enhanced the proportion of mass interpretation for the child participants, from 75 % (Experiment 1) to 95 % (Experiment 3), but not for the adult participants (70 % Exp. 1; 75 % Exp. 3). However, Mann-Whitney tests indicated no significant difference in the acceptance rates for either group, across experiments (child: p = 0.301, z = 0.96, adult: Z = 0.321, p = 0.748).^[6] The findings conform to our account. That is, the findings clearly indicate that the mass interpretation is readily available for both child and adult speakers of Mandarin, with minimal contextual support. Apparently, all that is needed to highlight the mass interpretation is to make it clear that the question under consideration is about the amount that was acted upon by the characters in a story. By contrast, Cheng et al. (2008) cannot explain why participants assigned a mass interpretation to animate nouns in Experiment 3.

Although the findings are consistent with the account we proposed, an alternative account of the findings is worth considering. The alternative account supposes that both of the contexts in Experiments 1 and 3 involved grinding objects into substances, although the actual ‘grinding’ in Experiment 1 was invisible, i.e., the result of eating. Experiment 4 was conducted to assess this possibility, by investigating the availability of mass interpretations in contexts that do not involve grinding. This is investigated in Experiment 4.

6.5 Experiment 4

Experiment 4 investigates the interpretation of Mandarin animate nouns co-occurring with the verb juqi ‘lift-up’. The study asks whether or not Mandarin-speaking children and adults assign a mass interpretation to bare animate nouns in sentences with juqi. The study also replicates Experiment 2 and sees if the insertion of an individual classifier suffices to override the context, inducing a count interpretation of the target sentences.

To make the context felicitous for a mass interpretation (without a grinding scenario), the stories involved a determination about the relative weight of the animal characters, in a lifting competition. On a typical trial, Bird Monster and Frog Monster had a lifting competition. First, both monsters tried to lift baskets that contained animals. There were two baskets. One basket had two big dogs weighing 100 jins, and the other one had two small dogs weighing only 10 jin. Bird Monster easily lifted the basket with the two big dogs. Frog Monster tried to lift the same basket, but he could not. However, he did manage to lift the basket that contained the two small dogs. Bird Monster was awarded a gold medal, and Frog Monster received a black cross. The last scenario of the individual-oriented context is shown in Figure 6.

Figure 6:

The last scenario of Experiment 4.

There were two sentence types. Type 1 sentences contained bare animate nouns, as illustrated in (14). Type 2 sentences had classifier-bearing animate nouns, as illustrated by (15). The sentences are a minimal pair, differing only in the presence/absence of the individual classifier zhi.

(14)

Daniaoyaoguai	juqi	le	duoshao	gou,
Bird Monster	lift-up	Asp	quantity-of	dog,
qingwayaoguai	jiu	juqi	le	duoshao	gou,
Frog Monster	then	lift-up	Asp	quantity-of	gou
(i) (lit.) ‘If Bird Monster lifted X amount of dogs, then Frog Monster also lifted up X amount of dogs.’
(Big Bird and Frog Monster lifted the same amount of dogs).
(ii) (lit.) ‘If Bird Monster lifted up X number of dogs, then Frog Monster also lifted X number of dogs.’
(Bird Monster and Frog Monster lifted the same number of dogs).

(15)

Daniaoyaoguai	juqi	le	duoshao	zhi	gou,
Bird Monster	lift-up	Asp	quantity-of	dog,
qingwayaoguai	jiu	juqi	le	duoshao	zhi	gou,
Frog Monster	then	lift-up	Asp	quantity-of	CL	gou
(lit.) ‘If Bird Monster lifted up X number of dogs, then Frog Monster also lifted up X number of dogs.’
(Bird Monster and Frog Monster lifted up the same number of dogs).

In sentences (14) and (15), the animate noun gou ‘dog’ appears in a Mandarin conditional structure, called a WH-conditional. This kind of conditional structure requires an identical pair of non-interrogative wh-pronouns, one in the antecedent clause and one in the consequent clause. The pair of wh-pronouns designates the same quantificational information, regardless of whether the interpretation is count or mass (e.g., ‘A and B have the same number of entities’ or ‘A and B have the same amount of entities’) (Cheng and Huang 1996; Chierchia 2000; Lin 1996). The wh-pronoun duoshao in (14) commonly appears in this kind of conditional structure. Duoshao is underspecified in countability, such that it can stand for either a non-specified amount or a non-specified number of entities (roughly equivalent in meaning to ‘the quantity of something’). Therefore, when the animate noun gou ‘dog’ co-occurs with duoshao, as shown in (14), this sentence is ambiguous. The count interpretation can be paraphrased as ‘Big Bird and Frog Monster lifted the same number of dogs’ and the mass interpretation can be paraphrased as ‘Big Bird and Frog Monster lifted the same amount of dogs’. In the scenario under discussion, the participants’ acceptance or rejection of the test sentences can be used to indicate which interpretation they have accessed in the context. In particular, if participants reject a test sentence, they can be taken to have made a quantity judgment based on weight (i.e., a mass interpretation). On the other hand, if participants accept a test sentence, they can be taken to have made a quantity judgment based on cardinality (i.e., a count interpretation).

When the non-interrogative wh-pronoun duoshao is combined with the individual classifier zhi, as in (15), the compound duoshao zhi functions as a count determiner, so the sentence conveys only the count reading ‘Big Bird and Frog Monster lifted up the same number of dogs’. We expect the participants to accept the test sentence in this context, as the numbers of dogs lifted by Bird Monster and Frog Monster were the same.

Experiment 4 did not use the same sentence structure with gengduo ‘more’ as was used in the first three experiments. This change was the result of a specific lexical blocking effect we discovered in pilot work. In a pilot study, we presented gengduo sentences with the verb juqi ‘lift-up’ to describe a lifting competition, as illustrated with a sentence in (16). Some of our adult participants in the pilot study remarked that this scenario encouraged them to use geng-zhong (lit. ‘more-heavy’) (see (17)) rather than gengduo if they were asked to compare the weight of the animal characters.^[7] According to these participants, the use of gengduo is preferred when one is asked to compare the number of animals. Based on their comments, it appears that the use of gengduo may block access to the mass interpretation of bare animate nouns.

(16)

Daniao	yaoguai	juqi	le	gengduo	gou
Bird	Monster	lift-up	Asp	more	dog
‘Bird Monster lifted more dog(s).’

(17)

Daniao	yaoguai	juqi	le	gengzhong	gou
Bird	Monster	lift-up	Asp	more (weight)	dog
‘Bird Monster lifted more dog.’

To avoid this lexical blocking effect, we proceeded to use a conditional structure containing the non-interrogative wh-pronoun duoshao (zhi), as shown in sentences (14) and (15). Previous research shows that Mandarin-speaking children as young as four years old are capable of comprehending this conditional structure (Huang et al. 2021). It is also worth noting that, because the mass interpretation in Experiment 4 is associated with the dimension of weight, rather than substance, the experiment circumvents any lexical blocking effect caused by the availability of Mandarin compound words like gou-rou ‘dog meat,’ as mentioned in the account proposed by Cheng et al. (2008) (see Section 3).

Experiment 4 was a between-subject design, and as mentioned earlier we recruited two different groups of participants for the testing of the two types of test sentences. In each group, we had twenty children and twenty adults. The child participants in the two groups are similar in age: Group 1 were presented with the Type 1 sentence in (14) (age range: 4;11;9–6;5;16; mean age: 5;10;5); Group 2 were presented with the Type 2 sentence in (15) (age range: 4;9;5–6;6;17; mean age: 5;10;15). The child and adult participants were not tested in the first three experiments, but they came from the same kindergarten and university. The three animate nouns, gou ‘dog’, niu ‘cow’, and yang ‘sheep’ were used in the test story. Altogether, each participant was presented with 3 test sentences. For each type of test sentence, there are 60 responses in total (20 participants * 3 test sentences). And we also provided simple filler sentences, in addition to the test sentences. The whole testing session lasted about 7 min for each participant.

Here are the main findings. In interpreting Type 1 sentences (with bare animate nouns), both children and adults rejected the test sentences 60 % of the time (36/60 trials), assigning the mass reading to the bare animate nouns in the test sentences. The child and adult participants gave appropriate justifications for their rejections. For instance, in rejecting Example (14), the participants stated that (i) the big dogs lifted by Bird Monster were heavier than the small dogs lifted by Frog Monster; or (ii) the weights of the dogs lifted by the two monsters are not the same. Therefore, both child and adult participants preferred to assign the mass reading to this weight-oriented context, just as we expected. However, we still found a small set of participants (8 out of 20 children and 8 out of 20 adults) who accepted all of the three test sentences and assigned the count reading in this weight-oriented context. Thus, these participants exhibited a strong preference for the count reading 40 % of the time, despite the felicitous context for the triggering of the mass interpretation. This concerns the individual variation, a phenomenon we observed earlier in Experiments 1 and 3, and we will discuss it briefly in the General Discussion section.

To compare, in interpreting Type 2 sentences (with the individual classifier + animate noun), both children and adults accepted the target sentences 100 % of the time, assigning the count reading to the individual classifier-bearing sentences. The experimental data is summarized in Figure 7.

Figure 7:

The experimental results in Experiment 4.

Taken together, the experimental data in Experiment 4 clearly show that mass readings are available for bare animate nouns used with the verb juqi ‘lift-up’. In the presence of an individual classifier, however, only the count interpretation is assigned. These experimental results are consistent with those in Experiments 1 and 2. Since no physical transformation is involved in the experimental scenarios, we assume that the attested mass reading is part of the lexical meaning of Mandarin animate nouns, not due to count-to-mass-shift coercion. Our experimental data provide empirical data to support our theoretical view that the Universal Grinder is not a necessary condition for the access of the mass interpretation. In the next section, we will discuss more about the triggering of mass readings by putting together all the data from our four experiments.

7 General discussion

The present study began by reviewing two accounts of the Universal Grinder, by Pelletier (1975) and by Cheng et al. (2008). In Pelletier’s account, the Universal Grinder is applied to nouns in English. It is argued as providing a felicitous context for the access of the less favored mass interpretation via a mental operation of grinding individuals to their substance. Since Pelletier assumes that both count and mass senses are part of the lexical specification, his account of the Universal Grinder does not invoke count-to-mass coercion. By contrast, Cheng et al. (2008) contend that the Universal Grinder is applied to count nouns in typologically distinct languages such as English and Mandarin, and the mass interpretation is triggered by count-to-mass coercion. For the interpretation of nouns in Mandarin, they argue for a pragmatic coercion account in which real-world knowledge triggered by the Universal Grinder is needed to coerce a mass interpretation of certain food-denoting nouns like pingguo ‘apple’ in restricted contexts, but not from animate nouns like gou ‘dog’, due to the lexical blocking effects. We agree with Pelletier (1975) that count and mass senses are part of the lexical specification of nouns (see also Pelletier 2012) and the Universal Grinder is not a necessary condition for the access of the mass interpretation. And then we proceed to explore a broader characterization of the countability of Mandarin animate nouns. This is how quantity judgments come into the picture in the present study.

Quantity judgments in comparative structures permit a comparison between distinct dimensions of noun referents (Bale and Barner 2009, 2018), and hence have become a popular methodology in recent years for the study of the count-mass issue across languages (e.g., Barner and Snedeker 2005, 2006; Lin and Schaeffer 2018). According to the common practice of quantity judgments, we characterize the count and mass interpretations with the mode of quantification: while the count interpretation is evaluated based on cardinal dimensions, the mass interpretation is assessed on various non-cardinal dimensions. Non-cardinal dimensions contain the dimension of substance as well as other dimensions, such as amount, volume, weight, etc. (Rothstein 2017). In four experiments, we found that the interpretation of Mandarin animate nouns depends on the syntactic structure of the sentences in which they appear, and both contextual information (cardinal versus non-cardinal dimensions) and morphosyntax (the presence/absence of a classifier) are involved in the interpretation of Mandarin animate nouns.

Let us discuss first the role played by contextual information. Contextual information is important for the interpretation of bare animate nouns, and their count and mass interpretations can be accessed in distinct contexts. Thus, we designed distinct felicitous contexts for triggering the count interpretation and the mass interpretation. As expected, we found that the individual-oriented context biased the child and adult participants to assign the count interpretation, whereas the amount-oriented and weight-oriented contexts biased them to assign the mass interpretation (Experiments 1, 3 and 4).

Contextual information is especially important for the triggering of the mass interpretation, which is a less preferred reading in participants’ interpretation of animate nouns, possibly due to the animacy effects we mentioned at the beginning of the present study (Corbett 2000; Grimm 2018; Smith-Stark 1974). This also concerns the lexical blocking effects mentioned in Cheng et al. (2008). In our view, the animacy effects and the lexical blocking effects may suppress the triggering of the mass interpretation, but this interpretation is nevertheless available when felicitous contexts are appropriately provided. Particularly, we think the lexical blocking effects are only restricted to the circumstance in which the dimension of substance is under consideration, and they can be avoided when other non-cardinal dimensions are highlighted.

Our experimental results confirmed our view. First, we designed in our experiments two contexts that highlighted the non-cardinal dimensions of amount and weight, namely, the amounted-oriented contexts in Experiments 1 and 3, and the weight-oriented contexts in Experiment 4. The implementation of these two contexts facilitated the access of the mass interpretation of bare animate nouns in both Mandarin-speaking children and adults. Since the animate nouns we tested are precisely the kind of nouns that should not have mass interpretation on the account of Cheng et al. (2008), our data are convincing and allow us to make a more appropriate statement about the interpretation of bare animate nouns: the animacy effects and the lexical blocking effects may affect the access of the mass interpretation, but they do not block the access of this interpretation. Felicitous contexts can facilitate access to mass interpretation. The importance of contextual information for access to mass interpretations is also attested in other studies, such as Grimm and Levin (2012), Beviláqua and Pires de Oliveira (2014) and Scontras et al. (2017). For lack of space we cannot discuss these studies in detail here.

Furthermore, we went a step further to argue that the mass interpretation is a basic lexical meaning, rather than a derived meaning triggered from a count-to-mass coercion. This generalization is based on two findings. First, the presence of a Universal Grinder scenario did not exert a significant effect on the access of the mass interpretation, when a felicitous context is already available (Experiment 1 and Experiment 3). Second, the mass interpretation is triggered in the event of lifting-up in which no Universal Grinder effect is possible (Experiment 4). These robust experimental data suggest that the triggering of the mass interpretation does not rely on the pragmatic coercion involved in the scenario of the Universal Grinder, and hence we assume that the mass interpretation is part of the lexical specification.

Now we move on to discuss the role played by morphosyntax. We tested this variable by comparing the interpretation of minimal pair sentences: sentences containing bare animate nouns versus those containing the additional individual classifier zhi. Both types of sentences were tested in two distinct contexts, i.e., the individual-oriented context and the amount-oriented context. Indeed, the presence/absence of an individual classifier caused fundamental interpretive differences in our experiments. As we discussed earlier, bare animate nouns were assigned the count interpretation and the mass interpretation in the individual-oriented context and the amount-oriented context, respectively. By contrast, participants assigned the exclusive count interpretation to classifier-bearing sentences, regardless of the non-linguistic contexts in which sentences were used. Thus, morphosyntax, or to be more specific, individual classifiers proved to be the decisive factor that determined the specific interpretation that was accessed. This is consistent with previous findings on young children’s sensitivity to the grammatical categories marking countability, such as the plural morphology in English (Barner and Snedeker 2005, 2006) and in Brazilian Portuguese (Beviláqua et al. 2016).

Now we are in a position to compare our study with Cheung et al. (2009). In a word extension task, Cheung et al. tested 3-to-6-year-old children by asking them to choose between a shape alternative (that matched in shape but not in substance) and a substance alternative (that matched in substance but not in shape) when they were presented with bare sentences (e.g., ‘point to blicket’) or sentences with a classifier (e.g., ‘point to one-CL blicket’). It is reported that the difference in the percentage of shape judgment between the bare noun condition and the classifier condition is low (between 3 % and 20 %) in children. Based on these results, Cheung et al. conclude that Mandarin-speaking children are insensitive to classifiers in learning words. As already shown in our experiments, children are sensitive to the presence of classifiers, and Cheung et al.’s claim is unwarranted. In our view, the different results can be largely attributed to the different experimental methods used. The quantity judgment task, the experimental task used in our experiments, is more sensitive to the count-mass semantics than the word extension task, the task used in Cheung et al. This is because the count-mass distinction is rooted in quantification, rather than about objects and substances. Due to space limitations, we cannot get into the details, but interested readers can refer to the relevant discussions in Bale and Barner (2009) and Srinivasan and Barner (2020).

So, we found that both morphosyntax and contextual information are involved in the interpretation of Mandarin animate nouns, but these factors play different roles as determined by the specific syntactic structure of the sentences in which this type of noun appears. Contextual information only affects the interpretation of animate nouns when they appear in a bare nominal form and does not override the morphosyntactic features that are associated with countability; classifiers unambiguously specify the countability of Mandarin animate nouns. Taking into account the factors of morphosyntax and context information in our experiments, the present study shows the complexity involved in the interpretation of Mandarin animate nouns. We believe our experimental design can be extended to the testing of other types of nouns in Mandarin, and help us understand better the countability of Mandarin nouns in general.

Let us discuss the remaining issues before we conclude the paper. Recall that in Experiments 1, 3 and 4, 30 %–40 % of individual participants (children and adults) assigned the count readings to bare animate nouns in the contexts that are felicitous for the triggering of the mass interpretation. Thus, the mass interpretation is assigned at a level ranging between 60 % and 75 % in these contexts. We need to explain why the mass interpretation did not reach the ceiling level in these experiments.

To explain the individual variation, we need first to consider a division of labor between grammar and world knowledge. On one hand, we think both count and mass interpretations should be available for the interpretation of Mandarin animate nouns, as part of our grammatical knowledge. On the other hand, we also expect a certain degree of preference for the count interpretation over the mass interpretation, due to factors like the animacy effect (Corbett 2000; Grimm 2018; Smith-Stark 1974). So, the crucial thing for us is that world knowledge cannot override our grammar (Borer 2005), despite world knowledge causing some bias in our perception and conceptualization of entities in the world. In this regard, the attested preference of the count interpretation in some of the participants merely indicates “the reflection of salient world knowledge” (Borer 2005: 107).

In terms of the participant’s performance in the testing, since the test sentences containing bare animate nouns are ambiguous with the count reading and the mass reading, the assignment of the alternative readings largely depends on how sensitive participants are to the specific contexts we designed. As explained in Section 6, even though both the cardinal information (the number of animal characters) and the non-cardinal information (the number of animal characters) were provided in the two distinct contexts, each context was designed to give prominence to one dimension of measurement (the cardinal dimension in the individual-oriented context vs. the non-cardinal dimension in the amount-oriented context). Those participants sensitive to this salience of information can obtain the two distinct readings in two distinct contexts. Otherwise, they may only get the count interpretation, due to the above-mentioned factors like the animacy effect. Therefore, even though we aimed to trigger the mass reading in the felicitous contexts, we did not expect the percentage to be at the ceiling. The test sentences remain ambiguous and the preference for the count reading can be hard to override. To us, the most crucial thing is, that as long as the participants exhibit a preference of the mass interpretation in the felicitous contexts, their behavior fits with our proposal. This is exactly what we found in Experiments 1, 3 and 4, as shown earlier. Thomas Hun-tak Lee (p.c.) also mentioned that gengduo ‘more’ tends to refer to the cardinal dimension and that this may contribute one possible reason to explaining why some participants assigned the count interpretation to bare animate nouns even in the amount-oriented context in our Experiments 1 and 3.

Interestingly, we notice that a similar percentage (60 %–75 %) is also attested in the triggering of the less favored mass interpretation in other languages. For example, Beviláqua and Pires de Oliveira (2014) report that in a context that highlights the volume of entities (e.g. filling a basket), bare singular nouns such as bola ‘ball’ and livro ‘book’ in Brazilian Portuguese are assigned a volume-denoting reading (a mass reading) by adults, over 75 % of the time. In another study, Grimm and Levin (2012) report that in a function-oriented context (e.g., furnishing a room), adult English speakers assign a mass reading 65 % of the time to furniture-type superordinate collectives. In a nutshell, when a felicitous context is provided, the less favored mass interpretation can reach a level ranging between 60 % and 75 %. This seems to indicate the average accessibility of the less favored mass interpretation across languages. In this regard, our data are consistent with the previous research from the crosslinguistic perspective.

8 Conclusions

From the Universal Grinder to syntactic structure, the present study looks deep into the countability of Mandarin animate nouns. We make strong use of the truth-value task-based quantity judgments to uncover the complex picture involved in Mandarin-speaking children’s and adults’ interpretation of Mandarin animate nouns. The main findings of the present study are two-fold. First, the experimental findings suggest that in addition to count readings, mass readings are an inherent part of the lexical specification of Mandarin animate nouns, rather than being the result of count-to-mass coercion. The mass interpretation of Mandarin animate nouns did not require assistance from a Universal Grinder, either real or imagined. However, without morphosyntactic cues, contextual information did influence the participants’ quantity judgments about bare Mandarin animate nouns. Second, the morphosyntax, rather than the non-linguistic context, proved to be the critical factor in how both Mandarin-speaking children and adults decided on which interpretation to assign animate nouns in comparative structures. When both contextual and morphosyntactic features are present, the morphosyntax prevails.

From the crosslinguistic perspective, it is worthwhile to extend our experimental design to the study of animate nouns or other types of nouns in other languages. For instance, taking into account the distinct morphosyntax of marking countability between Mandarin and English (Section 4), one can examine (i) whether the mass interpretation is also available for animate nouns or other typical count nouns in English; (ii) whether the access of mass interpretation relies on the operation of the Universal Grinder; (ii) how the morphosyntax and contextual information interact to affect the interpretation of animate nouns in English. Hopefully, this line of research may shed new light on the study of the Universal Grinder and the countability issue in general.