This paper proposes an analysis for the long-standing puzzle observed by Chemla (2007) regarding the anti-duality of the French universal quantifier tous, which arises even though French has no word for 'both' to feed a Maximize Presupposition competition. This phenomenon has been cited as an example in language where a dual 'conceptual alternative' is at play (Buccola etal., 2018), but no formal account of it has been put forth. Furthermore, a naive implementation of the idea overgenerates anti-duality inferences in other expressions, such as each, which, and one in English and French, which might be expected to be observed due to anti-dual counterparts in some languages like Icelandic and Japanese. We propose an account where French tous has an unpronounceable dual universal alternative built from a dual core concept, competition with which is licensed by the existence of a pronounceable expression equivalent in meaning, which we call an 'Indirect Alternative'. This proposal accounts for tous's anti-duality and lack of anti-n-ality for n > 2, as well as the lack of anti-duality in other quantifiers.
In this paper, I present novel data that challenge the dominant approach to modelling linguistic relevance, according to which a sentence is relevant if and only if it is informative to the Question Under Discussion (QUD). In particular, I show that a sentence can fail to be informative to the QUD and, despite this, still be perceived as relevant. In the final section, I outline the theoretical implications of this finding.
This paper discusses tense markers that convey an indirectness implication by comparatively analyzing sentences with the past tense suffixes -sqa in Cuzco Quechua and -tayna in Southern Aymara. The main contribution is to provide support for a (non-modal) evidence acquisition time analysis of these markers, which, I claim, is to be preferred over an analysis that relies on spatio-temporal distance and speaker perception. I provide novel data that highlight that acquiring evidence for a proposition is the key factor in the cases under examination (instead of relying on perceptual distance). I further show that 'late realization' scenarios are not possible in these languages, and discuss how the evidence acquisition time analysis captures this. The discussion provides the first systematic comparison between Cuzco Quechua and Southern Aymara from the perspective of formal semantics, which supports the well-known descriptive statement in the Andean literature that the grammars of these languages are parallel.
Conditional statements often imply meanings that extend beyond their literal content, influenced by contextual factors such as the Question Under Discussion (QUD) and a speaker’s epistemic state. One such inference is Conditional Perfection (CP), wherein a statement like, “If p then q,” is strengthened to imply that the antecedent p is the only contextually salient condition under which the consequent q holds (e.g., “p if and only if q”). While previous research has debated the role of QUDs in CP, the role of speaker knowledge has yet to be systematically tested. In two experiments, we investigated how these contextual factors influence conditional reasoning. In Experiment 1, participants were more likely to derive CP when the conditional statement was an answer to a QUD that focused on the antecedent compared to those that were answers to either a neutral QUD or one that focused on the consequent. In Experiment 2, CP was more likely when the speaker was knowledgeable about all alternative antecedents than when they had partial knowledge. These findings support the view that CP arises through a process of exhaustification, wherein the conditional statement is strengthened by negating relevant alternatives. More broadly, they suggest that conditional inferences depend on reasoning about alternative utterances, paralleling other forms of pragmatic enrichment such as scalar implicatures.
Disjunctive sentences can in certain cases convey strengthened, conjunctive inferences, a phenomenon also observed with indefinites. This behavior has been captured on the grammatical theory of exhaustification and alternatives, on which strengthening is modulated by what alternatives are generated by grammar. We examine a series of predictions of the theory arising from a grammatical property shared by disjunction and indefinites: their ability to undergo scope shift in certain configurations in which their alternatives-conjunction and universal quantifiers-cannot. We show that in such configurations, the absence of corresponding conjunctive and universal alternatives may allow exhaustification to generate conjunctive inferences. We present evidence that this indeed occurs, building on and extending earlier observations of Santorio (2018, 2020) and Bar-Lev and Fox (2020).
This squib presents evidence against a proposed agenthood constraint on non-culminating construals of accomplishment predicates (Demirdache and Martin 2015). This constraint on the agenthood properties of the causer subject has been argued to result in two types of languages with non-culmination construals: (i) languages in which any form of non-culmination requires agent control, and (ii) languages that require agent control only if non-culmination manifests itself in a lack of change of state entirely. New data from the Interior Salish language challenge the necessity of an agenthood constraint and show that there are languages in which agent control is not required for non-culmination, even when there is no change of state.
Mandarin Chinese lexically distinguishes the disjunctors in alternative questions (h & aacute;ishi) and in disjunctive assertions (hu & ograve;zhe), reflecting a distinction that Haspelmath (2007) and others have called interrogative versus standard disjunction. I argue that the two disjunctors share their basic syntax and semantics as junction heads (J) that project their disjuncts as Roothian alternatives, which are then interpreted by a corresponding question-forming operator or existential operator. I motivate this view from island insensitivity and focus intervention effects, which I show to apply in parallel to both alternative question formation with h & aacute;ishi and the scope-taking of hu & ograve;zhe. H & aacute;ishi also allows for a number of non-interrogative uses, subject to significant speaker variation. I argue that these patterns reflect broadly two types of grammars: those where h & aacute;ishi syntactically enforces that its alternatives be interpreted by certain operators, and those that do not. For the latter, more liberal speakers, h & aacute;ishi can be used non-interrogatively in the same environments that wh-phrases can be. The study and analysis of this pattern of variation leads to the conclusion that a so-called "interrogative disjunction" could be so specified via its syntactic specification or through its semantics alone, with both strategies being attested by different sets of Mandarin Chinese speakers.
The goal of this paper is to reconcile two observations regarding homogeneity and non-maximality. First, homogeneity is sensitive to constraints on presupposition accommodation, both in the positive case and under negation (Wehbe 2022). Second, there are asymmetries between positive and negative sentences with definite plurals (Bar-Lev 2021). We argue that taken together, these two observations support an account of homogeneity in terms of presuppositional exhaustification.
There is a long line of previous research arguing for a tight connection between the meaning of definites and demonstratives, see e.g. Ahn (2019); Dayal and Jiang (2022); Ebert et al. (2020); Elbourne (2008); King (2001); Roberts (2003); Wolter (2006). New data from Ga (Kwa), an under-researched language spoken in Ghana, directly supports the view that demonstratives are kind of marked definites: the meaning of demonstratives is transparently composed from the meaning of the definite determiner and the meaning of the particle restricting the interpretation of the DP. This transparent structure raises, however, a general compositional problem, i.e. how the anti-uniqueness inference of the demonstrative is derived, if the definite determiner conveys uniqueness. The paper solves this issue, addressing therefore one of the most pertinent issues in the semantics of demonstratives and definites across languages.
I argue for a semantic distinction between two classes of attitude complements. One class is best modeled in terms of possible worlds compatible with what the attitude holder believes/says, in the tradition of Hintikka. The other is best modeled in terms of centered worlds representing the de se perspective of the attitude holder, in the tradition of Chierchia (in turn inspired by Lewis). Much work has assumed that all attitude complements are to be treated semantically in this second manner. I refer to this hypothesis as Uniformity. Uniformity predicts that all attitude complements should be equally semantically able to host elements that must refer de se, such as shifted first person indexicals or relative tenses. Drawing on new evidence from Nez Perce, I demonstrate that this prediction is false, and argue that the best explanation for the distribution of dedicated de se elements comes from variation in whether attitude complements denote sets of centered tuples or rather sets of possible worlds.
Under what conditions can a pronoun refer to a given antecedent? It has long been noted that there is a connection between pronouns’ accessibility conditions and patterns of presupposition satisfaction. In this paper, this connection is explicitly spelled out in the form of the existence generalization: a pronoun may refer to an (indefinite) antecedent if and only if a witness for that indefinite can be presupposed to exist at the point where the pronoun is used. We show that, while dynamic approaches and E-type approaches do expect some parallels to hold between presupposition satisfaction and accessibility conditions for pronouns, they do not validate the existence generalization as a matter of principle. As a result, they face various under-generation problems, which they must solve through a set of unrelated assumptions. This work proposes a system that revives the choice-functional approach of Egli and Von Heusinger (1995) ; van der Does (1993) and can account for the generalization in full. In this system, a pronoun’s accessibility conditions is an existence presupposition simpliciter and the existence generalization is validated as a matter of principle. With relatively minimal tooling, this approach derives an interesting range of cases (bathroom sentences, cataphoric possibilities, donkey sentences, subordination). Some limitations and possible extensions are discussed: existential/universal readings and non-indefinite antecedents.
A prominent view of homogeneity takes it to be the result of a trivalent distributivity operator. Križ (2015) challenged this view by pointing out the existence of non-distributive plural predication, which is nonetheless homogeneous. In this paper I argue that this challenge disappears once a trivalent cumulativity operator (based on Link’s star operator), which derives meanings that are underspecified for distributivity and collectivity, is taken to be responsible for homogeneity instead. I point out that when this view is combined with a solution offered by Schwarzschild (1994) and Heim (1994) for the challenge of deriving both underspecified and ambiguous meanings (using ‘covers’), the challenge of deriving both distributive and non-distributive homogeneity is immediately resolved. I further argue that the view that emerges provides an explanation for the variation between predicates with respect to homogeneity and underspecification, and correctly predicts it to be context-dependent.
Morpho-syntactic markedness for number does not always seem to align with a singular or plural semantics. I show that in Alasha Mongolian (Mongolic) unmarked inanimate nouns are number neutral whereas their animate counterparts are strictly singular (cp. Bylinina and Podobryaev 2020). Unmarked inanimates, however, can be strictly singular if modified by a subclass of APs (e.g. big) and numerals. Plural-marked nouns can be exclusively or inclusively plural depending on upward/downward entailingness (like English). Adopting Harbour’s (2007, 2011, 2014) theory of number, I propose that the generalizations are best explained if unmarked number neutral nouns lack NumP, which would otherwise mark the NP for [$\pm $atomic], whereas their animate counterparts always project it. I argue that inanimates may project NumP if there is morpho-syntactic evidence to do so (some APs, overt plural-marking, and numerals). In addition, I propose a new solution for the inclusive/exclusive ambiguity of the plural: it is the result of syntactically conditioned allosemy at LF. Last but not least, the results of the analysis paired with cross-linguistic observations about number give rise to a novel generalization that correlates morphological markedness and semantic interpretation.
A key breakthrough in research on the semantics/pragmatics interface was the discovery of ‘locally computed’ scalar implicatures. While initial accounts modified the interpretive procedure to account for them (Chierchia 2004), the dominant view within the grammatical tradition to scalar implicatures is now that local implicatures are computed by an invisible but syntactically real version of only, the exhaustivity operator Exh (Chierchia et al. 2012). We argue that all current accounts (including the operator-based view) are faced with a challenge because some expressions display a Janus-faced behavior: they simultaneously seem to include and to exclude Exh. Take the discourse Jane flew to Spain or Portugal during the break. However Bill doesn’t believe it. The first sentence is naturally understood to mean that Jane flew to Spain or Portugal but not both, but the second sentence means that Bill doubts that Jane flew to Spain or Portugal (or both). We uncover several types of Janus sentences of this sort using pronouns and (in an appendix) using ellipsis. Janus sentences are found with unaccented scalar terms, but usually not with controls involving overt only or accented scalar terms. Without solving the puzzle, we discuss one broad solution strategies. This solution consists in complicating the semantics of the exhaustivity operator so that it can always be present, but with semantic effects that are systematically trivialized in some environments. This view becomes nearly indistinguishable from one where exhaustification is made part of the interpretive procedure itself, as in Chierchia (2004).
This paper offers a unified account of the English particle just that covers its exclusive, emphatic/intensifying, precisifying, unexplanatory, unelaborative, and counter-expectational uses. Drawing on an insight from another semantic domain, we claim that the chameleon-like behavior of just can be made sense of if we treat it as having a domain-widening function. The key proposal is as follows: the use of just indicates that the speaker is considering the widest set of alternative answers relevant at the context. The analysis relies on the notion of the optimal construal of an underspecified question, which makes use of a comparison between the inquisitivity of questions, modeled as the width of a question. The optimal construal of a question further depends on its answerability - i.e. whether the speaker considers a true answer to be accessible at the context (satisfying Quality) and whether the speaker considers addressing it to be relevant to discourse goals (satisfying Relevance). The diverse contextual effects of just that are observed arise from the interaction between the way in which the set of alternative answers to the underspecified question is construed and what is taken to be the speaker's motivation for signaling that the widest answerable construal of the question is being addressed via the prejacent.
Studies show that adults interpret simplex disjunction ("The mouse carried an apple or anorange") inclusively ('The mouse carried one, possibly both') or exclusively ('The mouse carried one butnot both'), while they generally interpret complex disjunction (e.g., "either. . . or") exclusively. Children,however, tend to be inclusive or conjunctive ('The mouse carried both') with simplex and complexdisjunctions alike. While previous studies focused on one simplex and one complex disjunction, thepresent study investigates multiple simplex disjunctions (neutral and prosodically marked "sau") andcomplex disjunctions ("sau. . . sau", "fie. . . fie") in child Romanian. We ask whether children’s conjunctiveinterpretation of disjunction is an experimental artifact that arises in contexts where the disjunctivestatement mentions all visible objects in the display, in which case it should disappear in contextswhere the disjunctive statement is made more informative. We also ask what role prosodic andmorphological markedness play in the interpretation of disjunction. We conducted two Truth ValueJudgment Tasks in prediction mode. In Experiment 1, the visual displays contained two objects,and the disjunctive test sentences mentioned both; while adults were exclusive, children respondedin a conjunctive manner to "fie. . . fie", but were inclusive elsewhere. In Experiment 2, the visualdisplays contained two additional unmentioned objects; children were inclusive with "sau"-baseddisjunctions, showing no sensitivity to prosodic/morphological markedness, but a considerablenumber of children persisted in their conjunctive behaviour with "fie. . . fie". Our findings supportthe view that, at least for some disjunctions, the conjunctive interpretation is a genuine semanticinterpretation in development.
This paper proposes a formal account of indefinites' exceptional quantificational and binding scope properties-static and dynamic exceptional scope. I show how minimal dynamic extensions of tools independently motivated and widely used for questions and indefiniteness offer a unified explanation of indefinites' multifaceted exceptional scope behavior. The account improves on existing static and dynamic theories of indefiniteness, and predicts a wide range of attested exceptional scope behavior for a range of expressions, both indefinite and not.
Recent experimental literature has investigated across-scale variation in scalar implicature calculation, probing why lexical scales differ from each other in their likelihood of being strengthened (e.g. old $\rightarrow $ not ancient v. smart $\rightarrow $ not brilliant). But in existing studies of this scalar diversity, less attention has been paid to potential variation introduced by the carrier sentences that scales occur in. In this paper, we carry out a systematic investigation of the role of sentential context on scalar diversity, focusing on scales formed by two gradable adjectives. We find within-scale variation: different subject nouns (e.g. The employee is smart v. The scientist is smart) have a significant effect on how robustly a scalar implicature arises. We then explore the relationship between a noun’s prior likelihood of exhibiting the stronger adjectival property (e.g. brilliance) and the rate of implicature calculation, and find that they are negatively correlated. We also test whether a previously identified factor in scalar diversity, adjectival threshold distance between the weaker (smart) and stronger (brilliant) adjective, is sensitive to the subject noun manipulation, but do not find evidence for this. In addition to their theoretical import, our findings also highlight the methodological importance of controlling carrier sentences.
Predicativists hold that proper names are count nouns with a predicative meaning, and treat bare singular names as predicative DPs headed by an unpronounced definite. However, bare singular names exhibit differences in grammatical behavior from ordinary definite singulars. One difference, it has been argued, is that while ordinary definite singulars can be interpreted generically, bare singular names cannot. This is not right: bare singular names can have generic uses. I present the evidence and offer an argument that generics with bare singular names are good news for predicativists.
