Day 1 (16 December 2016)
|09:00 - 09:30||Registration|
|09:30 - 09:45||Opening|
|09:45 - 10:45||Friederike Moltmann: Modal Objects and their Truthmakers|
|10:45 - 11:00||Coffee Break|
|11:00 - 11:45||David Boylan: Miners and Modals|
|11:45 - 12:30||Ethan Jerzak: Two Ways to Want|
|12:30 - 14:00||Lunch Break|
|14:00 - 15:00||Floris Roelofsen: Information, Issues, and Live Possibilities|
|15:00 - 15:45||Dirk Kindermann: Fragmented Contexts|
|15:45 - 16:00||Coffee Break|
|16:00 - 16:45||Mark Bowker: Situated Content and the Problem of Underdetermination|
|16:45 - 17:45||Ede Zimmermann: Multiple Indices vs. Indirect Senses|
Day 2 (17 Decmber 2016)
Day 3 (18 Decmber 2016)
|10:00 - 11:00||Nikola Kompa: Language and Embodiment|
|11:00 - 11:15||Coffee Break|
|11:15 - 12:00||Markus Kneer: Truth-Assessment and Retraction of Epistemic Modals:
|12:00 - 12:45||Dietmar Zaefferer: Bridging the Gap Between Language and Action with an Agent-Based Situation Theory|
|12:45 - 14:00||Lunch Break|
|14:00 - 15:00||Sebastian Löbner: Frames as Informational Holograms - Towards an Integrating Theoretical Model of Syntax, Semantics, Utterance Meaning, and Context in Frame Theory|
- Mihnea Capraru: Sundial Semantics and Embedded Communication
- Simone Carrus: Literal and Non-Literal Uses of Slurs: The Case of the Denial
- Milana Kostic: Towards a Probabilistic Account of Epistemic Modality
- Insa Lawler, Florian Hahn & Hannes Rieser: Multi-Modal Context-Dependency
- Milo Phillips-Brown: I want to, but ...
- Hannes Rieser: A Process Algebra Account of Speech-Gesture Interaction
The background of this talk is the development of an informational conception of logic that is based on the methodology of the philosophy of information, and in particular on the thesis that information is always assessed at a given level of abstraction. Here, I wish to specifically explore the similarities and dissimilarities between an informational perspective on logic based on situation semantics and my own account by systematically contrasting the relational and situated accounts of information.top
I consider Sobel-like sequences of attention and belief change in which a reasoner shifts between belief and belief in the negation as they sequentially attend to new issues. I argue that these sequences support a granularity model of attention (cf. Yalcin , , de Jager , Franke and de Jager , Fritz and Lederman , [author]) and also point to ways of extending a possible worlds picture of cognition with a storage-retrieval mechanism so as to better handle non-monotonic belief change.top
I argue that the problem of the essential indexical is an instance of a problem of the essential index: the truth-conditions of judgements (beliefs or assertions) are not fully determined by their contents. While their contents have relative truth-conditions, empirical judgements have absolute truth-conditions, involving a particular index of evaluation. I argue that the problem of the essential indexical, understood as a threat to the doctrine of propositions, actually arises from the contrast between de re and de dicto beliefs, and invites us to regard de re beliefs as empirical beliefs to be evaluated at an object of acquaintance.top
Speakers do not intend their utterances to express semantic contents. Given even a modest informational situation, there are often a number of different yet equally viable candidates for the content of the sentence uttered, each of which would communicate the same propositions. It would therefore be unnecessary, or even unreasonable, for speakers to intend any one of these contents in particular. The view is presented as a response to Stanley and Szabó’s (2000) problem of underdetermination. An interesting consequence is the potential compatibility of prima facie conflicting linguistic theories.top
Kolodny and MacFarlane (2010) argue that the deontic “ought” and “should” are information sensitive. This paper aims to show that the epistemic “ought” and “should” are also information sensitive, that this has important consequences for implementing information sensitivity and that it helps resolve certain methodological questions raised in this literature.top
Asudeh and Giorgolo (2016) provide a category-theoretic account of co-denoting names in glue semantics, developing work in denotational semantics and functional programming (Wadler 1992) to model how partial perspectives on situations influence interpretation of discourse situations and other agents’ informational situations. They suggest that their semantics is compatible with minimalism, given (Burke 2015), which has recently paired glue semantics with the sorts of syntactic representations assumed in Tree-Adjoining Grammar (Joshi et al. 1975) and the Minimalist Program (Chomsky 1995). We show that their theory can be recast as proposed and that the recasting has a number of advantages.top
This paper argues that language can embed itself in the environment in such a way as to off-load the semantic encoding of information, and thus to partially relieve the speakers of the need to represent the information mentally. Take, for instance, the (imaginary) Sundial people, who travel eastward and westward, tell the time only by sundials, and thus never become aware that the same hours come at different times in different places. Despite this absence of mental representation they can still communicate successfully, by uttering such sentences as, ‘Everywhere we go we dine at 7.’top
A lively debate on the nature and mechanisms of slurring is being daily enriched by philosophers, linguists and psychologists. The fundamental and undisputed property of slurs is their ability to derogate and offend individuals and groups. This power is attributed to the slurs content by the so-called content-based theories (Macià, 2002; Schlenker, 2007; Potts, 2007; Hom, 2008; Williamson, 2009; McCready, 2010; Croom, 2011; Camp, 2013; Hom & May, 2013; etc.), in opposition to the non-content based ones (Anderson & Lepore, 2013; Numberg, 2013).
Among the hypotheses proposed on account of the Derogatory Force (Hom, 2008), some claim that the derogatory content possesses an “at-issue” component, others claim that its nature is univocally pragmatic. In this article, we will face a specific phenomenon: the behavior of the derogatory content when it is object of a denial (It's not true that P / P is false).
Starting from the intuitions proposed in Camp (2013), we will then look to this phenomenon assuming a dual account (Hom & May, 2013), thus adopting the idea according to which the Derogatory Force of slurs would be expression of two components of meaning: an at-issue component (only in assertions) and an additional pragmatic component (in any kind of structure).
My aim is to show that the analysis of non-literal vs literal uses of slurs leads to different conclusions. Indeed, in so-called literal uses (e.g. 'faggot' used to target a male homosexual), both the dual account and the pure pragmatic account potentially result in the same reliable analysis. That is, the denial appears to be unable to block the derogation, inasmuch it is conveyed pragmatically. By contrast, in the case of so-called non-literal uses of slurs (e.g. 'faggot' used to target a male heterosexual), the two accounts turn out to be similarly inadequate. In light of this fact, especially if we were not very confident in the classical dichotomy of literal vs non-literal uses, we could take seriously the analysis proposed in Croom (2015) according to which, at least to analyze the content of slurs, a family resemblance account of categories has to be preferred to the classical theory of concepts. Furthermore, I will show that the information about the target, available to the speaker, influences the use of the denial in a potentially controversial way.top
I will present a simple-minded view of perception as the classification of objects and events in terms of types viewed as cognitive resources. The theory of types that I am using is TTR (Type Theory with Records) which borrows a great deal from work in logic and computer science in a tradition initiated by Per Martin-Löf. It provides a rich type theory, that is, it includes types, not just for basic ontological categories such as entities and functions, but also types of objects such as Tree and Boy, and types of events (or situations) such as Hugging-of-a-dog-by-a-boy. Types may be complex objects constructed from other types in a type theoretic universe. I will suggest that natural language semantics is at bottom based on our cognitive ability to perceive objects and situations in terms of types. To this we have added the ability to reason in terms of the types themselves. Thus, for example, we can consider types of situations without actually perceiving a situation of the type and even types of situations which are impossible. And we can also consider the types which other agents believe or desire to be instantiated or stand in some other attitudinal relation to.
I will argue that types can be used to give a model of propositions (following the dictum known as “Propositions as Types” which Martin-Löf took over from intuitionistic logic). The sentence 'A boy hugged a dog' is true if there is a situation of the type Hugging-of-a-dog-by-a-boy and false if there is nothing of that type.
I will also argue that types can be used to model mental states. For example, a type modelling a belief state is a type of the way the world would be if our beliefs were true and a type modelling our desires is a type of the way the world would be if our desires were fulfilled.
The semantics of the attitudes such as believe or want is modelled by matching types corresponding to the attitudinal complements against the type representing the appropriate mental state. Thus 'a believes p' is true just in case p matches the type, Tbel(a), modelling a’s beliefs. Basically p matches Tbel(a) just in case p is a subtype of Tbel(a). That is, any situation of type p would also be a situation of type Tbel(a). We will develop a refinement of this basic intuition and show how it treats examples of “hyperintensionality”. We will also develop a notion of point of view which enables us to superimpose a type on another type (using a notion of asymmetric merge, similar to priority unification in feature based systems). This facilitates a semantic treatment of attitude reports in which the speaker’s point of view is superimposed on the type corresponding to the mental state of the person whose attitude is being reported. We will discuss how this relates to classical problems involving proper names, as well as some cases of intentional identity and a puzzle originally introduced by Janet Fodor concerning sentences like Charley wants to buy a coat like Bill’s.top
Erica Cosentino and Markus Werning: Does Pragmatic Modulation Occur Before or After Sentence Meaning Composition is Completed? N400 Effects of the Interaction Between Context-Induced Affordances and Lexical Meaning.
In the current ERP study, we investigates the time course of the interaction between the lexically specified telic role of a noun (the function of the object to which the noun refer to: e.g., a funnel is generally used to pour water into a container) and an ad-hoc affordance contextually induced by the linguistic discourse (e.g., the funnel is glued to the wall and the agent wants to hang the coat). Exploring the pattern of the N400 modulations, we argue that pragmatic modulation occurs before sentence meaning composition is completed.top
I present unexplored uses of 'wants', on which information inaccessible to the desirer herself helps determine the truth of the desire report. I show that theories by Stalnaker, Heim, and Levinson fail to predict both these readings and true indicative conditionals with 'wants' in the consequent. These features are related by modus ponens reasoning.
To solve both puzzles I develop a relativist semantics, on which desire attributions express information-neutral propositions. I compare 'wants' with 'ought', which exhibits similar unembedded and compositional behavior. Finally I sketch a pragmatic account of the purpose of desire attributions that explains the evolution of this relativistic behavior.top
Stalnaker’s influential conception of the conversational context – like any account of pooled information – suffers from a version of the problem of logical omniscience. For Stalnaker’s possible-worlds account, this is the problem of idealizing interlocutors’ rational abilities: the total set of interlocutors’ joint presuppositions is consistent (if it is to carry any information) and it is deductively closed – that is, any information entailed by the presuppositions is itself presupposed. In this talk, I argue that the conversational common ground is fragmented. That is, it does not form a single consistent, deductively closed set of presuppositions, but is rather organized into a number of fragments, each of which contains presuppositions that together are consistent and deductively closed, but which need not be consistent with presuppositions in other fragments. In the first part of the paper, I provide a number of arguments to motivate the thesis that the common ground is fragmented: from the need to model (a) inconsistent discourse, (b) joint reasoning in conversation and its failure, (c) cases of presupposition accommodation in disagreement, and (d) the felicity of repeating information already in the common ground. In the second part, I develop the outlines of a model of the common ground that makes room for fragmentation. The model draws on Roberts’ (1996, 2004) question-under-discussion model to individuate fragments. Using questions, I hope to also shed some light on the vexed question of the individuation of fragments for individual mental states of agents.top
Within the paradigm of Situated Cognition, which spans a rather wide and varied research area, the claim that cognition is embodied has gained some prominence. My concern will be with approaches to language comprehension that are united by the claim that language comprehension is embodied. One way of spelling this out is to say that in order to understand a linguistic expression one has to simulate the corresponding experience (denoted by the expression). This in turn requires activation in sensory and motor regions of the brain because in simulating a particular experience we exhibit roughly the same patterns of neural activity that accompanied the initial experience. In the paper, I will first examine some of the evidence that has been brought forth in favor of the claim that language comprehension is based on simulation. In a second step, I will ask how simulation (sensorimotor activation) is supposed to yield understanding. The remainder of the paper will be devoted to the topic of abstract terms. More specifically, I will address the questions of whether we understand abstract concepts by means of metaphors and whether metaphor comprehension itself is embodied. I will argue that both questions have to be answered in the negative. Finally, I will claim that the conception of understanding employed by embodied approaches to language comprehension is seriously limited.top
In this paper a recent proposal for the semantics of epistemic modal 'must' will be analyzed. Instead of the traditional account (Kratzer, 1991) which esentially treats 'must' as a universal quantifier over the subset of possible worlds, Lassiter proposes a probabilistic account of epistemic modality. He argues that such an account can capture the perceived gradability of 'must' better, as well as fulfill the following desiderata: a) it can capture the fact that 'must' marks propositions believed indirectly, based on inductive inference, b) the predicted dual of the interpretation of 'must' is a plausible interpretation of 'might' and c) 'must p' is asserted when p is believed with some high, but non-maximal probability. Lassiter modifies an existing account (Swanson, 2006) of epistemic modality, which gives an interpretation of 'must' as built around scales (similarly to the interpretation of gradable adjectives) by employing probabilistic graphical models. However, his proposal leaves a couple of issues open: 1) on such interpretation 'must' is not closed under conjunction (thus undermining the possibility of incorporating it in a compositional framework) 2) such interpretation leaves the question of threshold setting for the acceptance of p open. I will propose some amendments of Lassiters account and show that such modified account avoids these issues. On this account 'must p' is interpreted as true if p is entailed by a consistent stably-probable proposition (i.e. a proposition which is more probable than its negation conditioned on any proposition with which it is compatible). It can be shown that such interpretation of 'must p' fulfills requirements of closure under conjunction (Leitgeb, 2014) and is rendered true if p is accepted with sufficiently high, but non-maximal probability. Moreover, in such settings, the value of the threshold is determined in a principled way, i.e. it is dependent on the underlying partition of the information state of the agent.top
In our talk, we tackle a yet untreated issue in debates about context-dependence, namely some multi-modal dimensions of information-dependency. We first argue that the meaning of speech-accompanying iconic gestures is dependent on its co-occurrent speech to some degree. Since there is no solution prototype for modeling such kinds of information-dependency in either formal semantics or formal gesture research, we provide a first approach to model multi-modal meaning based on a gesture semantics computed as a function of its speech context.top
Kristina Liefke: Rich Situated Propositions: The ‘Right’ Objects for the Content of Propositional Attitudes
This talk presents a novel account of natural language meaning on which the meaning of sentential utterances is semantically rich and informationally situated. In virtue of its situatedness, the meaning of a sentential utterance varies with the informational situation of the cognitive agent interpreting the utterance. In virtue of its richness, the meaning of an utterance contains information beyond the sentence’s lexically encoded information. The agent-sensitivity of meaning solves a number of problems for propositions which have recently resurfaced in the philosophy of language (esp. the cognitive accessibility problem, the substitution problem, and the rigid granularity problem). It is inspired by the idea from cognitive science and artificial intelligence that cognition is always strictly tied to the situational context of the cognition. In reference to the name of the latter approach, i.e. situated cognition, we call our account Rich Situated Semantics.top
Sebastian Löbner: Frames as Informational Holograms - Towards an integrating theoretical model of syntax, semantics, utterance meaning, and context in Frame Theory
A network of object nodes interlinked by attribute links, a given frame is usually considered from one perspective: as information about the object represented by the distinguished “central node”. Correspondingly, the frame is “read” as providing type information about this distinguished object. However, a coherent frame provides equally complex information about every single node it contains; it has as many “readings” as it has nodes. The total information in a frame can be attributed to each node in it – like a hologram viewed from different angles provides perspectivized images of the same object.
If frames are used as a framework of linguistic theory, the hologram property of frames has far-reaching consequences and explanatory potential. It bears directly on the operations of unification, frame embedding, or integration of two (or more) frames into an overarching frame. By these operations, two or more frames get connected to form a greater frame; all information contained in one of the frames integrated now bears on all objects figuring in the others. Pustejovsky (1991) introduced the notion of “cocomposition” as a mechanism responsible for the fact that the same expression can take on different senses in the course of, and due to, compositional interaction with another expression: “the product of function application [e.g. of a verb to its object, S.L.] would be sensitive to both the function and its active argument” (p. 421). Cocomposition analyzes composition beyond general semantic rules of meaning combination, like function application, by taking a closer look at the interaction of meanings on the basis of partial decomposition or of a rich type system (cf. Asher 2011). If semantic composition is modeled as fame unification, due to the hologram property of frames all composition is cocomposition.
In a frame approach to language, semantic and syntactic composition can be integrated into one frame, as sketched in Löbner (2014: 45–48). In addition to discussing the aforementioned theoretical consequences for semantic composition, the talk will sketch the added descriptive and explanatory potential of the frame-theoretic approach that results from embedding the frame of syntactic and semantic composition into a frame of an utterance, and the utterance frame, in turn, into frames of the world.top
Computational semantics has been among the major diculties in development of computerised processing of human language. A major reason for that is the most prominent phenomenon of human language — paramount ambiguity. Many of the ambiguities can be resolved in communications by context information, in specific situations. Other ambiguities persist even in specific contexts. Thus, we are presented with demanding challenges for development of computational theories of language, meaning, and information. Adequate theories need to represent partiality and underspecification of information, in its natural residing in situations.
The ideas of Situation Theory, as a mathematical modeling of information, was introduced by Jon Barwise, in the 1980s. The initial motivation was to provide computational semantics of human languages with mathematics of situated information. Since then, simplified versions of Situation Theory have been used for applications and implementations in various areas, including in computational grammar of natural language. A prominent application has been in HPSG, and, more broadly, in other information systems. In the same time, mathematics and formalization of Situation Theory have been largely open, difficult topics.
In the first part of the talk, I introduce a new development of Situation Theory. The development employs a specialized formal language for a higher-order type-theory of situated information. The terms of the formal language represent situation-theoretic objects. The language has specialized terms for constrained computations by mutual recursion. Specialized terms represent networks of parameters that are simultaneously constrained to satisfy restrictions. Language terms include components for integration of locations in 3-dimensional space and time. The second part of the talk represents application of the Typed Sit- uation Theory for representation of semantic content of English expressions, via syntax-semantics in Generalized Constraint Based Lexicalized Grammar (GCBLG). While GCBLG is closest to HPSG, it is a generalization of the approaches within the class of CBLG. By this, we target applications to varieties of grammars.top
This talk will outline a novel semantics of modals based on modal objects and their satisfaction conditions. Modal objects are, for example, obligations, permissions, needs, and abilities. They have truthmakers or satisfiers, which are situations or actions, and, if they are modal objects of necessity, falsifiers or violators. The talk will make use of a development of Fine’s recent truthmaker semantics with its central notion of exact truthmaking and its notion (exact) false-making, allowing exact truthmaking (satisfaction) and false-making (violation) to also hold between an action or situation and a modal object. On the semantics of modal sentences that will be outlined, modal verbs take modal objects as their implicit (Davidsonian) argument and the subject or complement clause or prejacent of the modal acts as a predicate characterizing the modal object in terms of its satisfiers (truthmakers) and possibly violators (falsifiers). The semantics will assign the very same logical form to sentences with modals of necessity and sentences with modals of possibility.
I want pizza, but I don’t want heartburn. When we constrain the metasemantics in standard ways, all prominent semantics for 'want' badly mishandle sentences like these. Suppose I believe I’ll get heartburn if I have pizza. Then these views wrongly predict that 'I want pizza' and 'I don’t want heartburn' can’t both be true. We need to thread a needle between getting these core cases right without overgenerating in others. My goal here is largely negative: I present novel cases, showing that neither of two promising proposals threads this needle. I then briefly sketch a solution.top
The talk is based on extensive corpus work dealing with the interaction of gesture and speech in natural settings. The problem handled is how gesture and speech processes taken as semiotic systems can be modelled in a formal language. The main argument in the talk is that this cannot be achieved in paradigms currently under use such as DRT, SDRT or HPSG. The proposal is to turn to process algebras in the tradition of Milner’s ?-calculus. The special process algebra discussed in the talk is a newly developed ?-? calculus which can transport typed ?-expressions over communicating channels.top
Both in inquisitive semantics and in update semantics, sentences are not evaluated with respect to single possible worlds, but rather with respect to sets of possible worlds, which are usually regarded in this setting as information states (but may also be regarded as partial worlds or situations). The two frameworks differ, however, in how they construe the central semantic relation between sentences and information states. As a consequence, the task of integrating them raises some non-trivial issues. We will offer a preliminary proposal for how to address these issues, and will illustrate some of the specific features of the resulting framework.top
Markus Werning: Evidence for Single-Type Semantics – An Alternative to e/t-based Dual-Type Semantics
Partee (2006) conjectures a formal semantics for natural language (hereafter, single-type semantics) which interprets CPs and referential DPs in the same semantic type: properties of situations. Partee’s semantics contrasts with Montague semantics and its recent contenders (dubbed dual- or multi-type semantics) which assume distinct basic types for the semantic values of referential DPs (i.e. individuals) and CPs (i.e. propositions, truth-values, or sets of assignment functions). Partee’s conjecture is motivated by results in event semantics and discourse representation theory, which support the indirect uni-directional shiftability between propositions and individuals. However, none of these results supports the identity of the types for individuals and propositions (e.g. by displaying a modeling advantage of single-type semantics over dual-type semantics). Our paper compensates for this shortcoming. In particular, it identifies a number of arguments for the adoption of single-type semantics which display this semantics’ greater unificatory and explanatory power. The latter are based on the ability of single-type semantics to provide a uniform account of the distributional similarities between DP and CP, to explain the truth-evaluability of DP fragments, and to explain the semantic relations between CPs and referential DPs. To further support single-type semantics, we defend it against a number of standard objections.top