Context and Content Lectures 2009 (Slides)École Normale Supérieure, École des Hautes Études en Sciences Sociales, Institut Jean Nicod (2009)
The lectures will use modal constructions, including various kinds of conditionals, as probes into the nature of context-dependency in natural languages. There will be three major recurrent themes: Variable Economy, Syntactic Contextualism, and Indeterminacy.
1. Variable Economy
An influential line of research in contemporary contextualism, represented by Jason Stanley, Zsóltan Szabó, and Kai von Fintel, says that the impact of contexts on contents is via syntactically represented variables. The discussion so far has mainly focused on implicit restrictions for quantifiers, where unpronounced domain variables have been posited. To do the job they are designed to do, domain variables have to range at least over properties, however, if not over entities of even higher types. This is implausible. The properties needed can’t seem to be picked up in realistic contexts. An alternative was proposed by Roger Schwarzschild in his 2009 EALING lectures: domain restrictions for quantifiers can be projected via domain fixing functions from ‘anchors’, which are always individuals or situations, hence entities of the most basic types e and s. If there are only a few obvious choices for domain fixing functions, we might hypothesize that it is just anchors that are syntactically represented via pronouns whose values have to be contextually supplied. Such a hypothesis does not only contribute to a constrained theory of context dependency, it also contributes to research programs that aim at constraining the types for natural language variables more generally, as exemplified in work by Gennaro Chierchia from the eighties, and more recently in Meredith Landman’s dissertation.
The lectures will explore whether Schwarzschild’s ‘deconstruction’ of implicit quantifier domain restrictions can be carried over to implicit restrictions for modals like must, might, or can. Modals require an ordered domain of possibilities for their interpretation, and this domain of possibilities is often not explicitly expressed, but contextually supplied. Could we assume that modal restrictions, too, can be systematically projected from anchors that are situations or events via a small set of obvious domain fixing functions?
2. Syntactic Contextualism
Syntactic contextualism assumes that many semantic properties of lexical items are not truly properties of those lexical items themselves, but are contributed by inflectional items surrounding them. Syntactic contextualism has been successful with nominal and verbal inflection (Hagit Borer), adverbial modification (Guglielmo Cinque, Marcin Morzycki), sentential complementation (Keir Moulton), and modals (Valentine Hacquard). In the area of modality, Hacquard has proposed a unified semantics for modals of all flavors, arguing that the different flavors of modals are, in fact, determined by different pieces of inflection surrounding modals in different positions in the hierarchy of verbal inflectional heads. More specifically, according to Hacquard, different kinds of event arguments are made available in different syntactic positions, and those different types of event arguments determine domains of possibilities for modals in systematic ways.
The lectures will attempt to unify Hacquard’s and Schwarzschild’s visions. Using Schwarzschild’s terminology, what Hacquard proposes amounts to taking event arguments as representations of suitable anchors from which domains of possibilities can be projected via domain fixing functions. The question that needs to be explored further is where exactly those arguments come from. Are they the familiar Davidsonian event arguments, or rather representations of topic or resource situations that we know are responsible for other types of domain restrictions, as shown in the work of François Récanati, and most recently in Florian Schwarz’s dissertation.
Even if the impact of context on content is via variables of the lowest possible types, there will always be cases where contexts leave the values of context variables genuinely underspecified. In the case of modals, contextual indeterminacy has the effect that the domains of possibilities for modals remain underdetermined. In those cases, addressees can attempt to shift modal boundaries in ways unintended by speakers, as observed in David Lewis’ Scorekeeping in a Language Game. Are there constraints for modal boundary shifts, and if so, where do they come from? The lectures will discuss this question for conditionals. If conditionals are modal constructions, and if domains of possibilities for modals are systematically projected from anchors via domain fixing functions, the constraints on boundary shifts for conditionals should be derivable from default settings for modal anchors and domain fixing functions. This might explain, for example, why the default interpretation for indicative conditionals is the material implication interpretation, hence the weakest possible interpretation for a modalized conditional. On the other hand, the default interpretation for a counterfactual is logical implication, hence the strongest possible interpretation for a modalized conditional.
Citation InformationKratzer, Angelika (2009). Context and Content Lectures. Paris, École des Hautes Études en Sciences Sociales.