About 'about'

Abstract

I provide a compositional account of about-PPs in combination with attitude predicates and content nouns, and as a predicate. The account requires that attitude predicates are properties of content-bearing eventualities, rather than relations that take propositions (or other clause denotations) as arguments. I argue further that the relevant notion of 'content' must be extremely general, allowing for question-like, proposition-like, and hybrid meanings.

Semantic Information and the Syntax of Propositional Attitude Verbs

“…Two NP structures were selected: single (38a) and double objects (38b). NPs were chosen so as not to have an interpretation in which they could be interpreted to have propositional content (Anand & Hacquard, 2014;Moulton, 2009a, b;Rawlins, 2013;Uegaki, 2012). A third feature relevant to NP complements-passivization-was also included (39).…”

A computational model of S-selection

“…7 In this methodology, one (i) selects a set of verbs and a set of subcategorization frames, (ii) computes the cartesian product of these sets, (iii) constructs sentences for each element of this cartesian product, and (iv) gathers acceptability judgments about those sentences on an ordinal (1-7) scale. We selected 1,000 clause-embedding verbs to include in our large-scale experiment using a search of VerbNet (Kipper-Schuler 2005) seeded by the clauseembedding verbs discussed in Hacquard & Wellwood 2012;Anand & Hacquard 2013Rawlins 2013;White et al 2014. This seed contained ∼500 verbs.…”

“…There are roughly three approaches in the literature to explaining the syntactic their distribution: (i) responsives are ambiguous between a declarative-and an interrogative-taking variant-e.g., they S-select both propositions and questions (Karttunen 1977a;Heim 1994;George 2011);13 (ii) responsives only S-select propositions, and interrogatives either (iia) always denote propositions (Groenendijk & Stokhof 1984), or (iib) they are somehow cast into propositions by an operator or coercion rule (Egré 2008;Spector & Egré 2015); or (iii) responsives only S-select questions, and declaratives (iiia) either always denote questions (Rawlins 2013), or (iiib) they are somehow cast into questions by an operator or rule (Uegaki 2012(Uegaki , to appear, 2015. We cannot do justice to this rich literature here, but even at this coarse level of analysis we can fruitfully use the abstract characterizations above to make predictions about the results our model should produce if any one option is correct.…”

“…This in turn may be thought of as requiring a hybrid type that can represent multiple kinds of content-e.g., informative and inquisitive content (cf. Groenendijk 2009;Ciardelli, Groenendijk & Roelofsen 2013;Rawlins 2013).…”

“…This question has a long history in the semantics literature (Karttunen 1977a,b;Groenendijk & Stokhof 1984;Heim 1994;Ginzburg 1995;Ginzburg & Sag 2001;Lahiri 2002;George 2011;Uegaki 2012Uegaki , to appear, 2015Rawlins 2013), with many potential answers. Our aim is to relate our modeling results to these proposals and to argue that current proposals likely need to be relativized to particular verb classes, split in particular between cognitive and communicative verbs.…”