Evidentiality is often classified under the general category of epistemic modality. At the same time, canonical modals like English must are sometimes argued to encode a kind of inferential evidentiality. By examining the semantics and pragmatics of Japanese inferential evidential yooda Davis & Hara (2014) and Hara (2017) show that evidentials constitute a category distinct from epistemic modality like daroo. That is, while p-daroo denotes quantification over possible worlds (Must(p) at f, g, w), p-yooda presupposes that the event p usually causes a state q (Must_p(q) at f_c, g, w) and asserts that the speaker perceives q. The analysis correctly predicts that the assertion of p-yooda has a contextual requirement that there is an effect state q usually caused by p, but does not commit the speaker to p. On the other hand, the assertion of p-daroo entails that the speaker is (weakly) committed to p, but its causal dependency (i.e., q causes p) only arises as a pragmatic inference.
In this talk, we aim to show that this heterogeneity between evidentiality and modality is psychologically real based on the results of the EEG experiment conducted by Hara et al. (2020). In particular, only the anomalous yooda sentences with mismatching causal context elicit N400/P600 effect, while no effect is observed from the anomalous daroo sentences. In the literature of psycholinguistics, two kinds of semantic/pragmatic violations are often conflated. Pylkkänen et al (2009) point out that anomalous sentences such as Dutch trains are sour, which are often considered semantically/pragmatically ill-formed and elicit N400 effect in ERP studies, are actually violations of world knowledge. Thus, we argue that the anomalous yooda sentences are instances of world knowledge violations that force the parser to infer an unrealistic causal dependency, while the anomalous daroo sentences are instance of purely linguistic/pragmatic violations.

In this paper we provide a (non-classical) account of the probability of conditionals, and two logics of conditional reasoning: (i) a logic C of inference from certain premises that generalizes deductive reasoning; and (ii) a logic U of inference from uncertain premises that generalizes defeasible reasoning. Both logics rest on the Cooper conditional (Cooper 1968), a variant of the De Finetti conditional also considered by Belnap, Olkhovikov, and Cantwell. The two logics preserve Import-Export and are connexive, handling the interaction of conditional and negation in a way that departs from classical logic. But whereas C is monotonic for the conditional, U is not, and whereas C obeys Modus Ponens, U does not without restrictions. We use the distinction between the two systems to cast light, in particular, on McGee's puzzle about Modus Ponens.

In Japanese linguistics, conditional clauses have been mostly studied from the perspectives of semantics and pragmatics (Akatsuka 1985; Arita 2007; Masuoka 1993b; Suzuki 1992; Takanashi 2003 among others), and little attention have been paid to syntactic aspects (Kuno 1973; Mikami 1960; Takanashi 2003). The aim of this study is to reveal the internal syntax of Japanese conditional clauses through a detailed investigation of the so-called Conditional Topic construction. I will try show that Conditional Topic constructions show close parallelism with Sluicing constructions. Furthermore, I argue that both Conditional Topic and Sluicing are derived from underlying Cleft, which is derived from the so-called no-da in-situ focus construction. Based on these observations, I argue that conditional clauses and Conditional Topics are derivationally related, i.e., Conditional Topics are derived from the underlying Conditional clause structure that involves the derivation of no-da in-situ focus constructions and Cleft constructions.

The variably strict semantics for conditionals is faced with a challenge: How do we ensure that A&B remains a live possibility after updating with a Sobel sequence of the form (i) if A, C; (ii) if A and B, not C? Willer (2017) points out that the posterior context will always entail that A&B is false, under the assumption that Modus Ponens is valid for bare indicative conditionals. Thus, it seems that we either have to reject Modus Ponens, or adopt a different solution.
One way forward is to drop the assumption (implicit in prior work) that all felicitous Sobel sequences are classically consistent. Križ (2016) argues for a referential analysis of if-clauses in which conditionals have a homogeneity gap, parallel to plural definites. In Križ's framework, conditionals with a truth value gap can still be true enough, and Sobel sequence conditionals might be felicitous without being strictly true. I show that this provides a solution to Willer's problem without dropping Modus Ponens, and without adopting Willer's dynamic analysis. The resulting view has similar virtues to Moss’s (2012) pragmatic approach to reverse Sobel sequences, but differs in its basic semantic assumptions. I lay out the formal details of a Sufficient Truth account, apply it to forward and reverse Sobel sequences, and discuss whether the full power of the variably strict semantics is necessary once we adopt the Sufficient Truth framework.

Counterfactual conditionals and other modal constructions display X-marking (von Fintel and Iatridou 2022), i.e. an array of morphological features that include past tense. Why is past morphology associated with a counterfactual meaning? This question has proven very hard to answer. Existing accounts are divided in two camps. "Past-as-Past" accounts interpret the relevant past morphology as real, and try to use it to fix the modal base of counterfactuals. "Past-as-Modal" accounts claim that, in counterfactuals, past morphology has a nonstandard modal meaning. Both families of accounts run into major problems. I suggest that both are right in some ways, and wrong in others. Past morphology does receive its standard meaning in counterfactuals, but this meaning includes a modal element: tenses impose constraints on the domain of quantification of modals. This account has independent motivation, and manages to predict compositionally the right meaning for counterfactuals without stipulations.

There has been increasing confirmation, supporting Bayesian approaches in the psychology of reasoning, of the conditional probability hypothesis that the probability of the natural language conditional, P(if p then q), is the conditional probability of q given p, P(q|p). Some studies have found possible exceptions to the hypothesis when p and q are independent. Other studies have not supported this conclusion. But the former studies have encouraged the development of truth condition inferentialism, which implies that there must be a compelling argument from p to q, with p pivotal in it, for a conditional if p then q to be true. There are clear examples of uses of if p then q, sometimes expressible with "even if", that are true although p and q are independent. I will call such cases true independence conditionals. Inferentialists have tried to dismiss these uses as "non-standard", but usually, when even if not-p, q holds, there is an implicit if p then q that is also an independence conditional. Uses of independence conditionals cannot be called "non-standard", and it is circular to claim that a theory only applies to "standard" cases, with the "standard" cases taken as the ones the theory applies to. A theory defended in this way is untestable. Moreover, research by inferentialists themselves has disconfirmed truth condition inferentialism. They try to argue that their disconfirming findings are the result of belief bias, but there are serious problems with this argument. Other psychological results also disconfirm truth condition inferentialism. Independence conditionals play an important role in human reasoning, and I will draw attention to questions that should be asked about them.

In this talk I explore Spanish bare if-clauses and argue that they support theories in which the interpretation of if-constructions relies on speakers’ assumptions regarding the relation between the protasis and the apodosis (Franke 2009, Sano & Hara 2014, Francez 2015, Goebel 2017, Biezma & Goebel 2022 a.o.). Spanish bare if-clauses convey an array of meanings not available in regular if-constructions and have been traditionally taken to be the result of reanalysis. In this talk I argue that their interpretation is not different from that of adjoined if-clauses. The analysis I offer shifts the burden of the interpretation to the dynamic update. In the broader agenda, I argue that by transcending the sentential domain, and relating the utterance to the overall context in which it is embedded, we can explain complex meanings without appealing to reanalysis and/or ad-hoc semantics.

Logical consistency, and its generalization to uncertain reasoning in the form of probabilistic coherence, are basic foundational principles of reasoning that hold independently of the meaning we ascribe to conditionals. Consistency, and its generalization to coherence, also forms the basis of a Bayesian probabilistic logic, going back to the work of Ramsey and de Finetti. Their analyses imply that a conditional satisfies the Equation P(if p then q) = P(q|p), a relation that has received strong empirical support as a model for natural language conditionals. We can also draw on coherence-based probabilistic logic as a methodological tool to compare theories of how people understand and reason with conditionals. Specifically, we can discover whether people’s reasoning is coherent above chance levels more often under the assumption of one interpretation of natural language conditionals than under another. And we can make quantitative comparisons of coherence under the assumption of different theories of conditionals. To make such theory comparisons as informative and as possible, it is important to look at cases in which the probability of being coherent just by chance is comparable between theories; and to examine not just the range of outcomes compatible with a theory, but also the range of in principle plausible outcomes that the theory rules out – and whether the latter is substantive enough for the theory to be empirically testable. A limitation of theory comparisons that draw on probabilistic coherence is that they apply only to a fixed point in time. Yet in practice reasoning and language use unfold dynamically over time, adapting to ever changing informational contexts with belief updating. We discuss an important non-logical variable in dynamic reasoning, the notion of discourse coherence as developed recently by Daniel Lassiter, and its implications for further psychological research on conditionals.

I advocate a theory of conditional meaning that foregrounds the effects of antecedents on the discourse context in which they occur: if A sets up a temporary context in which the truth of A is taken for granted, and this assumption can continue arbitrarily far in the discourse before being jettisoned. On this approach, the effects of conditional antecedents on sentence meaning are secondary and derivative from their discourse effect. The account builds on suppositional theories of conditionals from philosophy and psychology, and lesser-known formalisms for handling suppositional reasoning discussed by Zeinstra, S. Kaufmann, and Isaacs & Rawlins. In addition to simplified account of modal subordination that the latter authors highlight, seeing the world with this inverted spectrum gives us a new perspective on the probabilities of conditionals, Khoo's Bounding puzzle, and the fact that all constructions with suppositional meaning have similar domain-restricting effects on modals and other operators. The latter point, in particular, is difficult to explain if for a sentence-level theory of conditionals, because it is forced to treat domain restriction as a lexically encoded property of particular operators rather than an inevitable side effect of making a supposition.