Kai von Fintel

On its surface, a sentence like _If Laura becomes a zombie, she wants you to shoot her_ looks like a plain conditional with the attitude _want_ in its consequent. However, the most salient reading of this sentence is not about the desires of a hypothetical zombie-Laura. Rather, it asserts that the actual, non-zombie Laura has a certain _restricted attitude_: her present desires, when considering only possible states of affairs in which she becomes a zombie, are such that you shoot her. This can be contrasted with the _shifted reading_ about zombie-desires that arises with conditional morphosyntax, e.g., _If Laura became a zombie, she_ would want _you to shoot her_. Furthermore, as Blumberg and Holguín (J Semant 36(3):377–406, 2019) note, restricted attitude readings can also arise in disjunctive environments, as in _Either a lot of people are on the deck outside, or I regret that I didn’t bring more friends_. We provide a novel analysis of restricted and shifted readings in conditional and disjunctive environments, with a few crucial features. First, both restricted and shifted attitude conditionals are in fact “regular” conditionals with attitudes in their consequents, which accords with their surface-level appearance and contrasts with Pasternak’s (_The mereology of attitudes_, Ph.D. thesis, Stony Brook University, Stony Brook, NY, 2018) Kratzerian approach, in which the _if_-clause restricts the attitude directly. Second, whether the attitude is or is not shifted—i.e., zombie versus actual desires—is dependent on the presence or absence of conditional morphosyntax. And third, the restriction of the attitude is effected by means of _aboutness_, a concept for which we provide two potential implementations. We conclude by discussing our analysis’s prospective repercussions for the theory of conditionals more generally.

The simplest story about modals is also the canon: modals are context-dependent quantifiers over a domain of possibilities. Epistemic modals quantify over possibilities compatible with what is known. Just whose knowledge, or whose information state, is relevant is something decided by the context. The canon is simple and elegant but not quite right. The problem with the canon is that it is weighed down by the fiction that there is a determinate context in which an utterance of a bare epistemic modal is issued. On the story of this chapter, it is indeterminate just which group, or just which aggregated information state, is quantified over by such bare epistemic modals. This chapter explores how a story that exploits this contextual indeterminacy can explain what needs explaining without departing from what's right about the canon and so without positing anything semantically peculiar about bare occurrences of epistemic modals.

This article surveys the state of the art in the field of semantic universals. We examine potential semantic universals in three areas: (i) the lexicon, (ii) semantic “glue” (functional morphemes and composition principles), and (iii) pragmatics. At the level of the lexicon, we find remarkably few convincing semantic universals. At the level of functional morphemes and composition principles, we discuss a number of promising constraints, most of which require further empirical testing and/or refinement. In the realm of pragmatics, we predict that Gricean mechanisms are universal, but suggest that the precise nature of presuppositions may be subject to cross-linguistic variation. Finally, we follow E.L. Keenan in concluding that the overarching universal of effability or translatability between languages cannot be upheld in its strongest form. A recurring theme throughout this survey is how much work still remains to be done in the relatively young field of cross-linguistic formal semantics.

In "*Must* ...stay ...strong!" (von Fintel & Gillies 2010) we set out to slay a dragon, or rather what we called The Mantra: that epistemic *must* has a modal force weaker than expected from standard modal logic, that it doesn't entail its prejacent, and that the best explanation for the evidential feel of *must* is a pragmatic explanation. We argued that all three sub-mantras are wrong and offered an explanation according to which *must* is strong, entailing, and the felt indirectness is the product of an evidential presupposition carried by epistemic modals. Mantras being what they are, it is no surprise that each of the sub-mantras have been given new defenses. Here we offer them new problems and update our picture, concluding that *must* is (still) strong.

1 This paper has been presented at the workshop “Time and Modality: A Round Table on Tense, Mood, and Modality”, Paris, December 2005, at a CUNY linguistics colloquium in May 2006, and at the 6th Workshop on Formal Linguistics in Florian´opolis, Brazil, August 2006. We thank the audiences at those presentations, in particular Orin Percus, Tim Stowell, Marcel den Dikken, Anna Szabolcsi, Chris Warnasch, Roberta Pires de Oliveira, Renato Miguel Basso, and Ana M¨uller. We thank Noam Chomsky, Cleo Condoravdi, and Irene Heim for very helpful conversations about this material. We thank Bridget Copley for sharing with us her recent manuscript “What Should Should..

Why are some conditionals subjunctive? It is often assumed that at least one crucial difference is that subjunctive conditionals presuppose that their antecedent is false, that they are counterfactual (Lakoff 1970). The traditional theory has apparently been refuted. Perhaps the clearest counter-example is one given by Alan Anderson (1951: 37): If Jones had taken arsenic, he would have shown just exactly those symptoms which he does in fact show. A typical place to use such a subjunctive conditional would be in the course of an argument that tries to bolster the hypothesis that Jones did in fact take arsenic. But then it would of course be self-defeating to presuppose that the hypothesis is false. Thus, something else must be going on.

The compositional semantics of sentences like Only mammals give live birth and The flag flies only if the Queen is home is a tough problem. Evidence is presented to show that only here is modifying an underlying proposition (its ‘prejacent’). After discussing the semantics of only, the question of the proper interpretation of the prejacent is explored. It would be nice if the prejacent could be analyzed as having existential quantificational force. But that is difficult to maintain, since the prejacent structures when encountered on their own are naturally read as having a lawlike flavor, which in many analyses is attributed to the semantics of implicit operators alleged to be present in them. In the end, an analysis is presented which attributes some very particular properties to these operators and thereby succeeds in providing the target sentences with intuitively adequate interpretations. These complex constructions can therefore be used as a probe into the nature of implicit quantification in natural language

Epistemic modals are standardly taken to be context-dependent quantifiers over possibilities. Thus sentences containing them get truth-values with respect to both a context and an index. But some insist that this relativization is not relative enough: `might'-claims, they say, only get truth-values with respect to contexts, indices, and—the new wrinkle—points of assessment (hence, CIA). Here we argue against such "relativist" semantics. We begin with a sketch of the motivation for such theories and a generic formulation of them. Then we catalogue central problems that any such theory faces. We end by outlining an alternative story.

A primary goal of research in the semantics/pragmatics interface is to investigate the division of labor between the truth-conditional component of the meaning of an expression and other factors of a more pragmatic nature. One favorite strategy, associated foremost with Grice (1967, 1989), is to keep to a rather austere semantics and to derive the overall meaning of an utterance by predictable additional inferences, called ``implicatures,'' which are seen as based on certain principles of rational and purposeful interaction. In this chapter, I will explore a di¨erent way in which the truth-conditional component is complemented in context. Imagine that we have persuasive evidence that an expression a in context c expresses a proposition p. The straightforward way of capturing this in a semantic system is to attribute to a a context-dependent meaning that maps c to p in a systematic and adequate way.

* This work has been evolving for a while now. Some parts trace back to the few pages on the context-dependency of quantifiers in my dissertation. Reading Recanati’s paper on domains of discourse made me rethink some of my earlier conclusions without in the end actually changing them much. Other parts formed the material for several discussions in my seminar on context-dependency at MIT in the fall of 1995, which included several sessions exploring the issues raised in an early version of Kratzer. Connecting the choice function approach to indefinites with a general analysis of contextual domain restriction was an idea that I considered then without however working out any details. Conversations with Lisa Matthewson and studying her paper inspired me to tackle the problem again. Connections to questions about the context-dependency of conditional sentences also tickled my fancy, since I have been working on conditionals for a while now. Having the opportunity to present this material at the Vilem Mathesius Series provided the impetus to flesh out some of my vague thoughts. Major influences come from the work done in recent years at MIT by Lisa Matthewson, Philippe Schlenker, Julie Legate, and Uli Sauerland. I thank Irene Heim, Lisa Matthewson, Orin Percus, Philippe Schlenker, and Jason Stanley for illuminating discussions about this material. At this point, I see this talk as a report on the state of the art, exploring an intriguing possible connection, but I do not claim substantial intellectual property rights. Mistakes are certainly mine, achievements are probably not