Emar Maier

It is rather uncontroversial that there are different ways to report de se attitudes, but there is still disagreement about the number and the nature of the different mechanisms at work. Following Anand (2006), I distinguish three types of de se reporting: one a special case of de re, another expressed by shifted indexicals, and a third expressed by dedicated de se pronouns. For the first two I propose reductions to de re and de dicto reporting, respectively, couched in a dynamic framework where presupposition resolution takes center stage. For the third, I part ways with all current proposals in offering what is essentially a de re analysis of dedicated de se pronouns. I motivate this radical departure with examples of de se pronouns binding de re reflexives, as recently brought into the spotlights by Charlow (2010) and Sharvit (2010).

This paper deals with the semantics of de dicto , de re and de se belief reports. First, I flesh out in some detail the established, classical theories that assume syntactic distinctions between all three types of reports. I then propose a new, unified analysis, based on two ideas discarded by the classical theory. These are: (i) modeling the de re/de dicto distinction as a difference in scope, and (ii) analyzing de se as merely a special case of relational de re attitudes. The resurrection of these ideas takes place in a dynamic setting. My formalization of the first idea involves a modification of the presupposition-as-anaphora resolution algorithm for DRT. The second involves treating acquaintance relations as second-order presuppositions, to be bound in the context by means of higher-order unification, or accommodated if necessary. The resulting framework requires no syntactic distinctions between different modes of attitude, with the exception of a specific subclass of de se reports characterized by special ‘ de se pronouns’ (i.e. PRO and logophors). These special pronouns are handled in syntax; everything alse is passed on to the pragmatic resolution module as it appears on the surface. The more sophisticated contextual resolution process nonetheless ensures adequate output truth conditions for a variety of classical and novel puzzles. In particular, I compare the new pragmasemantic system to the classical, syntactic analysis with respect to iterated and quantified reports, and monstrously shifted indexicals.

Percus & Sauerland (2003) use quantified belief reports of the form 'Only Peter thinks he's...' to argue for dedicated de se LFs. The argument is targeted against any reductionist account that sees de se as merely a particular subtype of de re, viz. a de re belief about oneself from a first person perspective, requiring nothing but an account of de re attitudes. My acquaintance resolution framework is an attempt at just such a reduction and in this paper I extend that theory with a projection mechanism to allow local accommodation of acquaintance relations. With this extension we can account for their data, as well as for some related data involving quantified belief reports familiar from arguments in the de se literature. Note: the embedded video of Peter's mistaken self-identity is urlhttp://ncs.ruhosting.nl/emar/dese.mpgavailable in .mpg here.

The information conveyed by any utterance is a motley ensemble. Utterances carry content about the world as it is according to the speaker, but also about speakers’ attitudes, the way they speak, what has been said before, and so on. There are many kinds of information that are conveyed by way of language, and differences in kind correlate with differences in status. Presupposed information exhibits a distinctive projection behaviour; conversational implicatures are cancellable in a way that asserted information is not; a pronoun’s gender may help to determine a referent, but is otherwise truth-conditionally inert; and so on.

Linguistics and philosophy have provided distinct views on the nature of reference to individuals in language. In philosophy, in particular in the tradition of direct reference, the distinction is between reference and description. In linguistics, in particular in the tradition of generative grammar, the distinction is between pronouns and R-expressions. I argue for a third conception, grounded in dynamic semantics, in which the main watershed is between definites, which trigger presuppositions that want to be bound, and indefinites, which set up new discourse referents. On this view, proper names, indexicals, and definite descriptions are all analyzed as presupposition triggers.

I present a concrete DRT-based syntax and semantics for the representation of mental states in the style of Kamp. This system is closely related to Recanati’s Mental Files framework, but adds a crucial distinction between anchors, the analogues of mental files, and attitudes like belief, desire and imagination. Attitudes are represented as separate compartments that can be referentially dependent on anchors. I show how the added distinctions help defend the useful notion of an acquaintance-based mental file against Ninan’s :368–377 2015) recent challenge involving counterfactual de re attitudes.

English direct discourse is easily recognized by e.g. the lack of a complementizer, the quotation marks (or the intonational contour they induce), and verbatim (`shifted') pronouns. Japanese employs the same complementizer for all reports, does not have a consistent intonational quotation marking, and tends to drop pronouns where possible. Some have argued that this just shows many Japanese reports are ambiguous: despite the lack of explicit marking, the underlying distinction is just as hard. On the basis of a number of `mixed' examples, I claim that the line between direct and indirect is blurred and I propose a unified analysis of speech reporting in which a general mechanism of mixed quotation replaces the classical two-fold distinction.

This thesis deals with the phenomenon of attitude reporting. More specifically, it provides a unified semantics of de re and de se belief reports. After arguing that de se belief is best thought of as a special case of de re belief, I examine whether we can extend this unification to the realm of belief reports. I show how, despite very promising first steps, previous attempts in this direction ultimately fail with respect to some relatively recent linguistic data involving quantified and infinitival reports, logophoric constructions, and monstrously shifted indexicals. Formalizing my idea of a contextual resolution of acquaintance relations in a dynamic framework, I arrive at an alternative analysis that handles all these data.

Karttunen observes that a presupposition triggered inside an attitude ascription, can be filtered out by a seemingly inaccessible antecedent under the scope of a preceding belief ascription. This poses a major challenge for presupposition theory and the semantics of attitude ascriptions. I solve the problem by enriching the semantics of attitude ascriptions with some independently argued assumptions on the structure and interpretation of mental states. In particular, I propose a DRT-based representation of mental states with a global belief-layer and a variety of labeled attitude compartments embedded within it. Hence, desires and other non-doxastic attitudes are asymmetrically dependent on beliefs. I integrate these mental state representations into a general semantic account of attitude ascriptions which relies on the parasitic nature of non-doxastic attitudes to solve Karttunen’s puzzle

I provide a novel semantic analysis of proper names and indexicals, combining insights from the competing traditions of referentialism, championed by Kripke and Kaplan, and descriptivism, introduced by Frege and Russell, and more recently resurrected by Geurts and Elbourne, among others. From the referentialist tradition, I borrow the proof that names and indexicals are not synonymous to any definite description but pick their referent from the context directly. From the descriptivist tradition, I take the observation that names, and to some extent indexicals, have uses that are best understood by analogy with anaphora and definite descriptions, that is, following Geurts, in terms of presupposition projection. The hybrid analysis that I propose is couched in Layered Discourse Representation Theory. Proper names and indexicals trigger presuppositions in a dedicated layer, which is semantically interpreted as providing a contextual anchor for the interpretation of the other layers. For the proper resolution of DRSs with layered presuppositions, I add two constraints to van der Sandt's algorithm. The resulting proposal accounts for both the classic philosophical examples and the new linguistic data, preserving a unified account of the preferred rigid interpretation of both names and indexicals, while leaving room for non-referential readings under contextual pressure.

Chierchia (1989) and others have used the contrast between George hopes that he will win and Georges hopes to win in mistaken-self-identity scenarios, to argue for dedicated de se LFs. The argument, further strengthened by evidence of shiftable indexicals, appears applicable against any reductionist account that sees de se as merely a particular subtype of de re. My Acquaintance Resolution framework is an attempt at such a reduction, and this paper seeks to extend that theory with a logical principle of introspection for belief, to account for the data within a unified treatment of de re and de se.

A number of puzzles about propositional attitudes in semantics and philosophy revolve around apparent referential dependencies between different attitudes within a single agent’s mental state. In a series of papers, Hans Kamp offers a general framework for describing such interconnected attitude complexes, building on DRT and dynamic semantics. I demonstrate that Kamp’s proposal cannot deal with referential dependencies between semantically conflicting attitudes, such as those in Ninan’s puzzle about de re imagination. To solve the problem I propose to replace Kamp’s treatment of attitudes as context change potentials with a two-dimensional analysis.

It appears that in mixed quotations like the following, the quoted expression is used and mentioned at the same time: (1) George says Tony is his ``bestest friend''. Most theories seek to account for this observation by assuming that mixed quotations operate at two levels of content at once. In contradistinction to such two-dimensional theories, we propose that quotation involves just a single level of content. Quotation always produces a change in meaning of the quoted expression, and if the quotation is mixed the shift is, to a first approximation at least, from '...' to ``what x calls '...''', where x is a variable whose value is determined by the context. We argue that quotation is generally context dependent in various ways, and that some of these ways are presuppositional in nature; we present a detailed analysis of the presuppositions in question

In this paper I present a linguistic investigation of the literary style known as free indirect discourse within the framework of formal semantics. I will argue that a semantics for free indirect discourse involves more than a mechanism for the independent context shifting of pronouns and other deictic elements. My argumentation is fueled by literary examples of free indirect discourse involving what I call language shifts: Most of the great flame-throwers were there and naturally, handling Big John de Conquer and his works. How he had done everything big on earth, then went up tuh heben without dying atall. Went up there picking a guitar and got all de angels doing the ring-shout round and round de throne...that brought them back to Tea Cake. How come he couldn’t hit that box a lick or two? (Hurston 1937, Their Eyes Were Watching God) Free indirect discourse combines elements of both, but it seems decidedly closer to direct discourse. I examine the reigning semantic analysis of free indirect discourse, a variant of the demonstrative account of direct discourse featuring “double context dependence” (Banfield 1982,Schlenker 2004). After showing that this type of account has fundamental difficulties with language shift data like (1), I close with a brief sketch of an alternative analysis based on “mixed quotation.”

The quotational theory of free indirect discourse postulates that pronouns and tenses are systematically unquoted. But where does this unquotation come from? Based on cases of apparent unquotation in direct discourse constructions (including data from Kwaza speakers, Catalan signers, and Dutch children), I suggest a general pragmatic answer: unquotation is essentially a way to resolve a conflict that arises between two opposing constraints. On the one hand, the reporter wants to use indexicals that refer directly to the most salient speech act participants and their surroundings (Attraction). On the other hand, the semantics of direct discourse (formalized here in terms of event modification) entails the reproduction of referring expressions from the original utterance being reported (Verbatim). Unquotation (formalized here also in terms of event modification), allows the reporter to avoid potential conflicts between these constraints. Unquotation in free indirect discourse then comes out as a special case, where the salient source of attraction is the story protagonist and her actions, rather than the reporting narrator and his here and now.