Elsevier

Cognition

Volume 178, September 2018, Pages 178-192
Cognition

Original Articles
On children’s variable success with scalar inferences: Insights from disjunction in the scope of a universal quantifier

https://doi.org/10.1016/j.cognition.2018.04.020Get rights and content

Abstract

Previous developmental studies have revealed variation in children’s ability to compute scalar inferences. While children have been shown to struggle with standard scalar inferences (e.g., with scalar quantifiers like “some”) (Chierchia, Crain, Guasti, Gualmini, & Meroni, 2001; Guasti et al., 2005; Noveck, 2001; Papafragou & Musolino, 2003), there is also a growing handful of inferences that children have been reported to derive quite readily (Barner & Bachrach, 2010; Hochstein, Bale, Fox, & Barner, 2016; Papafragou & Musolino, 2003; Singh, Wexler, Astle-Rahim, Kamawar, & Fox, 2016; Stiller, Goodman, & Frank, 2015; Tieu, Romoli, Zhou, & Crain, 2016; Tieu et al., 2017). One recent approach, which we refer to as the Alternatives-based approach, attributes the variability in children’s performance to limitations in how children engage with the alternative sentences that are required to compute the relevant inferences. Specifically, if the alternative sentences can be generated by simplifying the assertion, rather than by lexically replacing one scalar term with another, children should be better able to compute the inference. In this paper, we investigated this prediction by assessing how children and adults interpret sentences that embed disjunction under a universal quantifier, such as “Every elephant caught a big butterfly or a small butterfly”. For adults, such sentences typically give rise to the distributive inference that some elephant caught a big butterfly and some elephant caught a small butterfly (Crnič, Chemla, & Fox, 2015; Fox, 2007; Gazdar, 1979). Another possible interpretation, though not one typically accessed by adults, is the conjunctive inference that every elephant caught a big butterfly and a small butterfly (Singh, Wexler, Astle-Rahim, Kamawar, & Fox, 2016). Crucially, for our purposes, it has been argued that both of these inferences can be derived using alternatives that are generated by deleting parts of the asserted sentence, rather than through lexical replacement, making these sentences an ideal test case for evaluating the predictions of the Alternatives-based approach. The findings of our experimental study reveal that children are indeed able to successfully compute this class of inferences, providing support for the Alternatives-based approach as a viable explanation of children’s variable success in computing scalar inferences.

Introduction

According to the standard semantics of disjunction, which is derived from classical propositional logic, a basic disjunctive sentence like (1) is true if at least one of its disjuncts is true. However, when a sentence like (1) is used in everyday conversation, it often gives rise to the additional inferences in (2) and (3), sometimes called scalar inferences.1 (Fox, 2007, Gazdar, 1979, Sauerland, 2004, among many others).

The traditional Gricean explanation for how such inferences are derived involves a combination of general reasoning and assumptions about how rational agents interact in conversation (Gamut, 1991, Grice, 1975, among many others). More specifically, the proposal is that the hearer assumes that at any given point during a conversation, the speaker will produce the most informative utterance that is relevant for the purposes of the conversation and that she believes to be true. For example, when the speaker utters (1), the hearer will reason that if the speaker could have uttered the more informative sentence (4), then she would have done so. The fact that the speaker uttered (1), rather than (4), leads the hearer to infer that the speaker does not have sufficient evidence that (4) is true. If the hearer takes the speaker to be well-informed, then the hearer is led to infer the negation of (4), which is the exclusivity inference (2).

The question of how such inferences are derived has been subject to much subsequent work since Grice and is the subject of ongoing debate. Approaches differ in whether the source of the phenomenon lies within the grammar or in a post-grammatical pragmatic domain. One aspect that most approaches agree on is that the hearer’s reasoning about what the speaker might have said should be constrained to a set of alternatives (Fox and Katzir, 2011, Gazdar, 1979, Horn, 1972, Katzir, 2007). A standard idea in the literature is that at least some of these alternative sentences are generated on the basis of certain words being associated with others in the language user’s mental lexicon. For example, Horn (1972) suggests that certain words occupy positions on lexical scales. The scales are ordered by informational strength (e.g., 〈or…and〉, 〈some…all〉, 〈might…must〉).2 When a speaker utters a sentence that includes one of the terms on the scale, alternative sentences are generated by replacing the relevant lexical item with one of its stronger scale-mates. For instance, the stronger alternative to the speaker’s utterance (1) is sentence (4). The alternative sentence, (4), is generated by replacing the lexical item ‘or’ in (1) with the stronger scale-mate ‘and’. Because the hearer supposes that the speaker is being cooperative and has produced the strongest statement that she was prepared to make, the hearer infers that sentence (1), with ‘or’, is the strongest statement the speaker felt she had evidence for. Therefore, the hearer infers the negation of the stronger alternative with ‘and.’ The negation of the alternative (4) yields the exclusivity inference (2).3

The relationship between the exclusivity inference and the ignorance inference (3) is controversial (for discussion, see Fox, 2007, Meyer, 2013, Sauerland, 2004). One approach invokes the same mechanism that underlies the derivation of scalar inferences. On this approach, the ignorance inference (3) involves a comparison of the speaker’s utterance (1) with the alternative sentences (5) and (6). These alternatives correspond to the individual disjuncts of the sentence in (1) (Gazdar, 1979, Sauerland, 2004).4 The alternative sentences (5) and (6) are both more informative than the asserted sentence (1). If the speaker had sufficient evidence that (5) and (6) were true, then she should have uttered these alternatives instead of (1). The fact that the speaker chose not to say (5) or (6) leads the hearer to infer that the speaker does not have sufficient evidence that (5) or (6) are true, generating the ignorance inference in (3).5

Note that there is no lexical replacement involved in generating the ignorance inference, because the alternatives in (5) and (6) correspond to the individual disjuncts of the asserted sentence (1) (Gazdar, 1979, Sauerland, 2004). That is, these alternatives can be generated through a simplification of the asserted sentence, specifically by deleting all but one of the disjuncts.

As we’ve now seen, the exclusivity and ignorance inferences can both be derived through reasoning over alternative sentences that the speaker might have uttered instead of what she actually said. Because of this, some accounts take these inferences to be derived in the same way (e.g., Sauerland, 2004). However, for reasons that we will discuss further, what is most relevant for our purposes is the distinction in the alternatives that are generated for each inference; one involves lexical replacement (i.e. the exclusivity inference), whereas the other involves simplification of the original assertion (i.e. the ignorance inference). Here and in the rest of the paper, we will refer to alternatives of the latter kind as non-replacement alternatives. For discussion regarding the distinction between alternatives generated by deletion versus alternatives generated by lexical replacement, see Katzir, 2007, Fox and Katzir, 2011.

In addition to the substantial theoretical work on inferences like these, there has also been a great deal of experimental work, including developmental studies that assess children’s ability to compute such inferences. The bulk of this work has reported that children consistently derive scalar inferences at lower rates than adults do (Chierchia et al., 2001, Foppolo et al., 2012, Guasti et al., 2005, Noveck, 2001, Papafragou and Musolino, 2003, among many others). However, this work has primarily focused on a rather restricted set of scalar inferences, typically those derived on the basis of alternatives that are generated through lexical replacement.

More recent investigations have expanded the range of inferences investigated to plurality inferences, free choice inferences, ad hoc implicatures, conjunctive inferences of disjunction, and ignorance inferences (Hochstein et al., 2016, Singh et al., 2016, Stiller et al., 2015, Tieu et al., 2014, Tieu et al., 2016, Tieu et al., 2017). Some of these studies report low rates of inference derivation by children (Tieu et al., 2014), as in previous studies; however, others have reported adult-like rates of inference derivation by children (Hochstein et al., 2016, Singh et al., 2016, Tieu et al., 2016). Notably, the inferences that children have been reported to derive readily are ones that involve non-replacement alternatives.

One attempt to explain children’s variable success in scalar inference computation is the Alternatives-based approach (Barner et al., 2011, Reinhart, 2006, Singh et al., 2016, Tieu et al., 2016). On this approach, children’s difficulties in deriving the bulk of previously investigated scalar inferences, and their apparent success on a handful of other inferences, are explained by appealing to children’s limitations in computing the relevant alternatives to the asserted sentence. Furthermore, the Alternatives-based approach makes clear predictions about children’s behaviour in dealing with inferences that have not been investigated previously.

The present paper investigates the predictions of the Alternatives-based approach by investigating children’s and adults’ interpretations of sentences in which a disjunction is embedded under a universal quantifier, as in (7). Such sentences constitute an ideal case study for testing the Alternatives-based approach, because two of the inferences that arise from such sentences are based on non-replacement alternatives. Specifically, the sentence in (7) licenses the distributive inference in (8) (Crnič et al., 2015, Gazdar, 1979). It has also been proposed that children license the conjunctive inference in (9) (Fox, 2007, Singh et al., 2016).

The derivation of the distributive inference in (8) involves the alternatives in (10) and (11), which are non-replacement alternatives as they can be generated by simplifying the asserted sentence in (7). As we will see, the findings of our experimental study reveal that children are adult-like in deriving such inferences. On the Alternatives-based approach, this finding is attributed to the fact that the relevant alternatives (i.e. (10) and (11)) are derived through simplification rather than lexical replacement.6

Moreover, we also found that a few children in our experiment computed the conjunctive inference in (9), as reported in previous developmental studies of children’s interpretation of disjunction (Singh et al., 2016 for disjunction embedded under “every”; Paris, 1973, Braine and Rumain, 1981, Chierchia et al., 2004, Tieu et al., 2017 for unembedded disjunction).

To preview, we will conclude that the results of the present investigation are in line with the predictions of the Alternatives-based approach, providing further support for it as a viable explanation of children’s variable success in computing a large class of scalar inferences.

The rest of the paper is organised as follows. First, we review previous developmental studies of scalar inferences. We then outline a prediction of the Alternatives-based approach, and present an experiment designed to test the prediction. We conclude with a discussion of how our results bear on the Alternatives-based approach.

Children have been found to display variable success on scalar inferences. We will first consider some studies that have reported non-adult-like performance, and then move on to cases in which children were adult-like.

For over a decade, developmental research has found that children derive scalar inferences far less often than adults do (e.g., Chierchia et al., 2001, Foppolo et al., 2012, Guasti et al., 2005, Noveck, 2001, Papafragou and Musolino, 2003, among many others). For example, Noveck (2001) presented underinformative sentences like (12) to a group of 8-year-old children, 10-year-old children, and adults. The participants were asked if they agreed with the sentences. Participants who derived the scalar inference in (13) were expected to disagree with the sentence.

Noveck reported that child participants rejected the underinformative statements far less often than the adult participants did. This pattern of responses was taken as evidence that children derived fewer inferences than adults did. This result is representative of much of the subsequent work investigating children’s performance on scalar inferences (though see Katsos & Bishop, 2011 for evidence that features of the task, such as the number of response options provided to children, can influence their performance). It is worth noting, however, that many of the previous developmental studies have focused on a fairly restricted set of scalar inferences, namely those involving the scalar quantifiers ‘some’/‘all’, the logical connectives ‘or’/‘and’, and the modals ‘may’/‘must’. Importantly, the derivation of such scalar inferences involves alternatives generated through lexical replacement. More recent work has turned to investigating other inferences, including those derived from non-replacement alternatives.

In the recent formal semantics/pragmatics literature, a wide range of inferences have received a scalar inference-based analysis (Chierchia, 2013, Fox, 2007, Klinedinst, 2007, Levinson, 2000, Romoli, 2013, Spector, 2007, Thomas, 2012). Subsequent studies investigating children’s knowledge of these diverse inferences have produced some surprising results (Bill et al., 2016, Hochstein et al., 2016, Singh et al., 2016, Tieu et al., 2014, Tieu et al., 2016, Tieu et al., 2017, Zhou et al., 2013), including the finding that some of these inferences are derived quite readily by young children.

One example of an inference that children have been reported to compute at adult-like rates is the free choice inference, investigated by Tieu et al. (2016). Tieu et al. investigated Mandarin-speaking 4-year-old children’s interpretation of disjunction under deontic modals, as well as English-speaking 5-year-old children’s interpretation of free choice ‘any’ under deontic modals. Both sentence types give rise to free choice inferences, which have received a scalar inference analysis in the literature (Alonso-Ovalle, 2005, Chemla, 2009, Fox, 2007, Klinedinst, 2007, Kratzer and and Shimoyama, 2002).7 For example, a sentence like (14) gives rise to the free choice inference in (15).

Tieu et al. (2016)’s results provided some preliminary evidence in support of the proposal that children are capable of computing inferences that can be generated using non-replacement alternatives, as compared to ones that require lexical replacement of alternatives. Tieu et al. used a Truth Value Judgment Task (Crain & Thornton, 1998), in which the child participants were presented with free choice sentences (e.g., (14)), in contexts that were inconsistent with the free choice inference (e.g., (15)). That is, in the context, the character mentioned in the target sentence (e.g., the elephant) was only permitted to perform one of the mentioned actions (e.g., catch a big butterfly). The study compared children’s performance on free choice inferences and standard scalar inferences involving plain disjunctions (i.e. the exclusivity inference). Tieu et al. found that the Mandarin-speaking children derived the exclusivity inference from test sentences with plain disjunction at a typically low rate (18%), whereas they derived free choice inferences from sentences that contained a deontic modal verb at a significantly higher rate (91%). Notably, both the free choice inference and the exclusivity inference conditions were tested within subjects, using the same Truth Value Judgment Task and similar materials, suggesting the divergent performance on the two inferences cannot be explained by differences in methodology.8

As another example, children have been reported to interpret numeral terms as exact (Barner and Bachrach, 2010, Huang et al., 2013, Papafragou and Musolino, 2003). On certain accounts, such an interpretation involves the derivation of a scalar inference (Horn, 1972, Levinson, 2000); for instance, ‘two’ is interpreted as ‘exactly two’ through the negation of the stronger alternatives ‘three’, ‘four’, etc.

Turning to ad hoc inferences, Stiller et al. (2015) presented 2- to 4-year-old children with sentence descriptions (e.g., My friend has glasses) and three corresponding pictures. The participants’ task was to identify which of three faces the sentence was describing. The three faces created an ad hoc scale (i.e. 〈face with no glasses, face with glasses, face with glasses and hat〉). If participants derived the relevant ad hoc inference for the target sentence description (i.e my friend does not have a hat), they were expected to select the face with glasses and no hat. Stiller et al. reported that 3-year-old children and 4-year-old children derived ad hoc inferences approximately 75% of the time.

In yet another example of children’s relative success, Hochstein et al. (2016) investigated 4- to 5-year-old children’s derivation of ignorance inferences. These authors reported that 5-year-old children derived ignorance inferences at a rate of 70%, while they computed exclusivity inferences only 30–40% of the time.

Finally, children have also been reported to succeed in accessing exhaustive interpretations similar to those in the derivation of scalar implicatures, when interpreting ‘only’-sentences involving context-dependent scales (Barner et al., 2011). Barner et al. (2011) presented 4-year-old children with pictures depicting three objects/animals (e.g., cat, cow, dog), all of which were doing the same activity (e.g., sleeping). Children were asked one of two questions; Are only some of the animals sleeping? or Are only the cat and the cow sleeping? The former question is associated with context-independent alternatives, 〈some…all〉, whereas the latter is associated with context-dependent alternatives, 〈cat, cow, dog〉. A negative answer to these questions was taken as evidence of an exhaustive interpretation, i.e. Only some, not all, of the animals are sleeping. Barner et al. reported that children accessed the exhaustive interpretation in the context-independent condition 33% of the time, whereas they did so in the context-dependent condition 86% of the time.

The majority of previous studies investigating children’s abilities with scalar inferences and exhaustive interpretations have reported low rates of derivation in children. However, some recent studies have revealed a series of inferences that children successfully compute. This disparity in findings calls for an explanation for children’s variable success. The need for an explanation is especially pressing on unified accounts that treat all of the relevant inferences in the same way.

Two main kinds of explanations for the developmental findings have been proposed. On one account, children’s difficulties with scalar inferences are attributed to limitations in their pragmatic knowledge (Katsos and Bishop, 2011, Noveck, 2001, Skordos and Papafragou, 2016). This line is taken, for example, by Katsos and Bishop (2011), who propose that children are more ‘pragmatically tolerant’ than adults. On this view, even when children derive scalar inferences, they are nevertheless more likely than adults to accept a target sentence, despite the fact that the inference makes the sentence false. Similarly, Skordos and Papafragou (2016) propose that children have a limited understanding of which alternative sentences are relevant in a given context (see Noveck (2001) for a similar idea). This limitation is expected to affect children’s derivation of scalar inferences, since alternative sentences need to be perceived as ‘relevant’ before they can be negated.

Such pragmatic explanations are able to account for the typical observation of low rates of scalar inferences in children. However, it is unclear how such accounts can be extended to cases where children perform at adult-like rates in deriving inferences (e.g., free choice inferences). One would expect the proposed limitation ascribed to children (pragmatic tolerance or limited understanding of relevance) to have a uniform effect across different types of inferences, contra recent findings.

A second kind of explanation that attempts to account for children’s selective success on scalar inferences is one we will broadly term the Alternatives-based approach. This approach attributes children’s difficulties to a limitation in children’s ability to generate alternatives. There are different proposals regarding the cause of this difficulty. Some explanations appeal to children’s processing limitations, suggesting that the process of generating alternatives is costly, and therefore challenging for children (Chierchia et al., 2001, Gualmini et al., 2001, Reinhart, 2006, Tieu et al., 2016). Others appeal to limitations in children’s knowledge of lexical scales, suggesting that children often have not learned the scales required to generate alternatives through lexical replacement (Barner and Bachrach, 2010, Barner et al., 2011). This difficulty with generating alternatives means that children will experience difficulties deriving the associated scalar inferences. In this way, the Alternatives-based approach, like the pragmatic-based explanations, can account for why children have been found to struggle with many scalar inferences.

In contrast to the pragmatic-based explanations, however, the Alternatives-based approach can also account for why children have been found to succeed on certain scalar inferences. Crucially, nearly all of the inferences that children have been found to compute successfully do not involve alternatives generated through lexical replacement. Instead, they involve non-replacement alternatives. That is, they involve alternatives generated through the simplification of the asserted sentence (or, in the case of ad hoc implicatures, through alternatives that are made salient and easily retrievable from the context). For example, as already mentioned, a sentence like (14) (repeated in (16)) is often associated with the free choice inference in (15) (repeated in (17)). Deriving free choice inferences as scalar inferences involves the alternative sentences in (18) and (19). However, these are non-replacement alternative sentences as they are generated by deleting parts of the assertion in (16), and not by lexical replacement of scalar terms (Alonso-Ovalle, 2005, Chemla, 2009, Fox, 2007, Klinedinst, 2007, Kratzer and and Shimoyama, 2002). As mentioned, Tieu et al. (2016) found that children readily derived the free choice inferences associated with these kinds of alternatives.

If we assume that, aside from the proposed limitation associated with the lexical replacement of alternatives, children are in all other relevant respects adult-like, then at least some version(s) of the Alternatives-based approach will predict children to readily derive scalar inferences like (17). For example, children’s limited knowledge of lexical scales, as proposed by Barner and Bachrach (2010), will not affect their ability to compute scalar inferences that do not involve knowledge of lexical scales, namely those derived from non-replacement alternatives.

The predictions of proposals attributing children’s difficulties to processing limitations (e.g., Reinhart, 2006), are somewhat less clear (see Singh et al., 2016, footnote 32, for some considerations that may speak against a processing-based account). Such proposals could in fact be taken to predict that children’s behaviour across all scalar implicatures should be uniform, assuming all implicatures are highly demanding in terms of processing resources. On the other hand, if we assume that alternatives derived through lexical replacement are more costly than alternatives generated through some other source (e.g., by simplifying the assertion), then these processing accounts will also predict that children should be more successful at deriving scalar inferences associated with non-replacement alternatives (see Chemla & Bott, 2014). In any case, in the remainder of this paper, when we refer to the ‘Alternatives-based approach’, we are only referring to explanations that predict that children will more readily derive scalar inferences from non-replacement alternatives.

The next section illustrates how the predictions of the Alternatives-based approach can be tested further by focusing on other inferences derived from non-replacement alternatives. Specifically, we will focus on testing the predictions of the Alternatives-based approach for sentences containing disjunction under a universal quantifier.

As noted earlier, a sentence like (20) is associated with the distributive inference in (21) and the conjunctive inference in (22). Crucially, as outlined in Appendix A, the alternatives that are required to compute these inferences are not generated through lexical replacement.9 Rather, the relevant alternatives in (23) and (24) are non-replacement alternatives that can be generated by deleting parts of the asserted sentence in (20).10

The Alternatives-based approach, therefore, predicts that children should readily compute distributive inferences like (21) and conjunctive inferences like (22), in response to sentences like (20), since both kinds of inferences are derived through non-replacement alternatives.11,12

Our predictions for adults, on the other hand, are different. While adults are expected to derive the distributive inference, they are not expected to derive the conjunctive inference. This is because adults, unlike children, can presumably generate the alternative involving lexical replacement (i.e. the alternative in which disjunction is replaced by conjunction), which is thought to ‘block’ the derivation of the conjunctive inference (see Fox, 2007, Singh et al., 2016, Tieu et al., 2017 for relevant discussion).

In sum, we are assuming that the distributive inference and the conjunctive inference are members of a natural class of scalar inferences, namely, inferences derived through non-replacement alternatives (Fox and Katzir, 2011, Katzir, 2007). Notably, for a sentence like (20), the distributive and the conjunctive inferences are the only inferences that can be derived via non-replacement alternatives (see Appendices A and B for details). Importantly for our purposes, the Alternatives-based approach predicts that children should be able to derive both kinds of inferences, since no lexical replacement is required. Since we do not expect adults to compute conjunctive inferences, we cannot expect that the two inferences will be computed at exactly the same rates in children and adults. However, the Alternatives-based approach does predict that children should compute this class of inferences (i.e. either the distributive or the conjunctive inference) at a similar rate to adults.13 This prediction was investigated in the present experimental study by assessing the rates at which adults and children derive distributive and conjunctive inferences from sentences like (20).

Section snippets

Experiment

The present experiment was devised to determine the extent to which children derive distributive and/or conjunctive inferences from sentences in which disjunction is embedded under the universal quantifier. Both of these inferences rely on non-replacement alternatives. Therefore, the experimental hypothesis, based on the Alternatives-based approach, was that the child participants would compute such inferences at adult-like rates. This finding would be in contrast to the findings of much

Discussion

The aim of this experiment was to investigate the predictions of the Alternatives-based approach (Barner and Bachrach, 2010, Barner et al., 2011, Chierchia et al., 2001, Gualmini et al., 2001, Reinhart, 2006, Tieu et al., 2016). Specifically, we aimed to test the prediction that children would be able to compute inferences that were derived from non-replacement alternatives, that is, from alternatives that could be generated from the asserted sentences through deletion.

We did this by

Conclusion

Recent developmental work has led to proposals that children’s variable success in computing scalar inferences can be explained by appealing to the nature of the alternatives involved; we refer to this family of approaches as the Alternatives-based approach. This approach was the source of the specific prediction we investigated – namely that adults and children would derive scalar inferences based on non-replacement alternatives to a similar extent. To test this prediction, we used sentences

Author contributions

C.B, E.P. and S.C. conceived the project; C.B. and E.P. designed the experiment in consultation with J.R., L.T. and S.C; C.B. and E.P. collected the data; E.P. analysed the data; E.P. and C.B. prepared the initial manuscript; J.R., L.T. and S.C. revised the manuscript.

Conflict of interest

The authors declare that no competing interests exist.

Acknowledgements

We would like to thank all the children and families who participated in the study. The research leading to these results was supported by the Australian Research Council Centre of Excellence in Cognition and its Disorders (CE110001021), the European Research Council under the European Union’s Seventh Framework Programme (FP/2007–2013)/ERC Grant Agreement n. 313610, ANR-10-IDEX-0001-02 PSL, ANR-10-LABX-0087 IEC, and the Leverhulme Trust, grant RPG-2016-100. No financial interest or benefit

References (62)

  • Alonso-Ovalle, L. (2005). Distributing the disjuncts over the modal space. In Proceedings-NELS (Vol. 35, p....
  • An, S. (2015). The acquisition of constraints in child mandarin, PhD thesis. Macquarie...
  • Barker, C. (2010). Free choice permission as resource-sensitive reasoning. Semantics and Pragmatics, 3,...
  • Barr, D. J. (2013). Random effects structure for testing interactions in linear mixed-effects models. Frontiers in...
  • C. Bill et al.

    Scalar implicatures versus presuppositions: The view from acquisition

    Topoi

    (2016)
  • Boster, C., & Crain, S. (1993). On children’s understanding of every and or. In Proceedings of early cognition and...
  • Bowler, M. (2014). Conjunction and disjunction in a language without ‘and’. In Semantics and linguistic theory (Vol....
  • Breheny, R., Klinedinst, N., Romoli, J., & Sudo, Y. (2016). The symmetry problem: Current theories and prospects...
  • Chemla, E. (2009). Similarity: Towards a unified account of scalar implicatures, free choice permission and...
  • G. Chierchia

    Broaden your views: Implicatures of domain widening and the “logicality” of language

    Linguistic Inquiry

    (2006)
  • Chierchia, G. (2013). Logic in grammar: Polarity, free choice, and intervention. OUP...
  • Chierchia, G., Crain, S., Guasti, M. T., Gualmini, A., & Meroni, L. (2001). The acquisition of disjunction: Evidence...
  • Chierchia, G., Guasti, M. T., Gualmini, A., Meroni, L., Crain, S., & Foppolo, F. (2004). Semantic and pragmatic...
  • S. Crain et al.

    Investigations in universal grammar

    (1998)
  • L. Crnič et al.

    Scalar implicatures of embedded disjunction

    Natural Language Semantics

    (2015)
  • K. Davidson

    And’or ‘or’: General use coordination in ASL

    Semantics & Pragmatics

    (2013)
  • F. Foppolo et al.

    Scalar implicatures in child language: Give children a chance

    Language Learning and Development

    (2012)
  • Fox, D. (2007). Free choice and the theory of scalar implicatures. In Presupposition and implicature in compositional...
  • D. Fox et al.

    On the characterization of alternatives

    Natural Language Semantics

    (2011)
  • Gamut (1991). Logic, language and meaning. University of Chicago...
  • G. Gazdar

    Pragmatics: Presupposition, implicature, and logical form

    (1979)
  • Cited by (12)

    • Do children derive exact meanings pragmatically? Evidence from a dual morphology language

      2021, Cognition
      Citation Excerpt :

      However, two sets of findings have challenged this line of reasoning. First, many studies have now shown that young children can compute implicatures when stronger scalar alternatives are made contextually available, and that early failures were therefore due to children's inability to access scalar alternatives, not a general pragmatic deficit (e.g., Barner, Brooks, & Bale, 2011; Barner, Hochstein, Rubenson, & Bale, 2018; Chierchia, Crain, Guasti, Gualmini, & Meroni, 2001; Foppolo, Guasti, & Chierchia, 2012; Hochstein, Bale, Fox, & Barner, 2016; Miller, Schmidt, Chang, & Munn, 2005; Long, Shukla, & Rubio-Fernandez, 2020; Pagliarini, Bill, Romoli, Tieu, & Crain, 2018; Papafragou & Tantalou, 2004; Singh, Wexler, Astle-Rahim, Kamawar, & Fox, 2016; Skordos & Papafragou, 2016; Stiller, Goodman, & Frank, 2015). For example, in one study, by Skordos and Papafragou (2016), children failed to compute some vs. “all scalar implicatures under conditions similar to earlier studies, but succeeded when scalar alternatives were made more available.

    • Meaning and Alternatives

      2022, Annual Review of Linguistics
    View all citing articles on Scopus
    View full text