We offer a new account of the semantics of predicates of personal taste (PPTs) like tasty and fun which, unlike recent proposals (Lasersohn 2005; Stephenson 2007a, 2007b), does not appeal to a judge parameter as a component of the evaluation index. We identify empirical shortcomings of previous proposals, arguing that PPTs have a first-person-oriented meaning component even in cases that seem to involve an exocentric interpretation. We propose that the interpretation of PPTs involves firstperson-oriented genericity in the sense of Moltmann (2006, 2010a). When I say This cake is tasty, I say roughly that for all worlds w and all individuals x such that x is relevant in w and I identify with x, the cake is tasty to x in w. We explain the shifting of the first person orientation from speaker to attitude holder in attitude reports by taking both matrix and embedded sentences to express properties rather than propositions (Stojanovic 2011). In both cases, an abstraction operator in the left periphery of the clause binds the variable responsible for the first-person-oriented interpretation of the sentence. The paper closes with a comparison with a similar proposal by Moltmann (2010b, forthcoming) and a discussion of the implications of our semantics for the analysis of attitudes de se.

This paper presents novel data regarding the logophoric pronoun in Ewe. We show that, contrary to what had been assumed in the absence of the necessary fieldwork, Ewe logophors are not obligatorily interpreted de se. We discuss the prima facie rather surprising nature of this discovery given the assumptions that de se construals arise via binding of the pronoun by an abstraction operator in the left periphery of the clausal complement of an attitude predicate, and that logophors are elements that are obligatorily bound by such abstractors. We show that this approach can be reconciled with these facts given the additional assumption that elements that are 'de se' bound can interact with the concept generator variables posited by Percus and Sauerland (Proceedings of Sinn und Bedeutung 7, 2003a; Proceedings of NELS 33, 2003b) to derive de re interpretations of embedded nominals. The proposed set-up has consequences for our understanding of puzzles raised by Heim and Sharvit concerning binding-theoretic effects with de re elements, and for the derivation of the obligatorily de se interpretation of controlled PRO.

In a partial control configuration the denotation of the controller is properly included in the understood subject of the infinitive. This paper proposes a compositional semantics for partial control-the first such proposal that we are aware of. We show that an account of what determines whether a given predicate licenses partial control follows naturally from the analysis without additional syntactic assumptions. At the heart of the proposal lies the idea that partial control predicates are attitude verbs and as such, quantify over a particularly finegrained type of modal base-so-called centred worlds. Unlike in traditional semantics for attitude reports, however, the lexical entry of these predicates requires that the property expressed by the control complement is applied not to the coordinates of this modal base, but rather to world, time and individual arguments that stand in a systematic relationship to those coordinates. This makes sense of the observation, going back to (Landau, 2000), that the ability of a control predicate to license partial control is intimately connected to its temporal properties. The Semantics of Partial Control 1 (Jackendoff and Culicover, 2003) advocate a semantic approach to partial control. Problems with their approach are discussed in (Landau, 2013a). See also (Hornstein, 2003), who suggests that partial control may not involve a distinct syntax from exhaustive control, but instead arises via a meaning postulate associated with the semantics of the PC class. With neither of these approaches was a complete compositional semantics for PC provided. 2 These lists are based on the useful characterization of the PC and EC classes in (Landau, 2000). The most important differences between his lists and ours are (i) our re-classification of claim and pretend as members of the EC class, based on data that will be introduced in section 3.4; (ii) the addition to the PC class of the verbs expect, vote, advise, recommend and a variant of remember that takes a gerundive complement, which we call rememberPC; (iii) the addition to the EC class of the 2 4.

No abstract

Hazard is the potential of something to cause harm; risk is the likelihood of harm occurring. Chemicals regulation is largely focused on minimising risks associated with chemicals — and rightly so. However, in the EU the hazard classification of individual chemicals can impact significantly the regulation of products containing those chemicals, regardless of the actual risks that the products may pose to human health or the environment. This can lead to undesirable consequences, such as restrictions on the use of safe products, substitution towards less safe products, and disincentives to innovate. Such hazard-based regulation tends to be at odds with World Trade Organization rules and has raised significant concern with EU trading partners. This journal is an ideal forum for discussing how the EU can move towards a better and more coordinated legal framework for the regulation of chemicals.