Speaker: Marcin Morzycki
Title: Integrating nonlocal readings of adjectives: more than just occasional
Abstract: It has been a longstanding puzzle that in sentences like The occasional sailor strolled by, the adjective can, on one reading, systematically be paraphrased with an adverb: ‘Occasionally, a sailor strolled by’ (Bolinger 1967, Stump 1981, Larson 1999, Zimmermann 2003, Schäfer 2007, DeVries 2010, Gehrke & McNally 2010, 2015). These are usually regarded as an idiosyncratic grammatical curiosity. I will argue that such nonlocal readings of adjectives are both far more common and more systematic than usually recognized. Similar effects occur with average (The average American has 2.3 children, paraphrasable with ‘on average’; Kennedy & Stanley 2009), whole (The whole ferret is submerged, paraphrasable with ‘wholly’; Morzycki 2002, Moltmann 2005), wrong (You opened the wrong bottle, paraphrasable with ‘it was wrong for you to open that bottle’; Larson 2000, Schwarz 2006), undisclosed (Solange is staying at an undisclosed hotel, paraphrasable with ‘it was not disclosed which hotel Solange is staying at’; Abusch & Rooth 1997), and others. Empirically, my aim will be to demonstrate that despite real idiosyncrasies, the regularities are sufficient to require a unified account of such readings. They fall into three classes distinguished by restrictions on the quantificational force of their determiner. I'll argue that what may lie at the heart of this phenomenon is the coupling of a mostly adjective-like syntax with a determiner-like semantics. Restrictions on the determiners that can occur in these constructions, as well as some notoriously idiosyncratic readings they receive there, may follow largely from the compositional mechanism required to intrepret adjectives whose semantics is determiner-like. More provocatively, I'll suggest that certain complexities that arise on some readings of average and perhaps other nonlocal adjectives can be understood as an independent and, in a principled sense, secondary layer of complexity superimposed on a simpler basic semantics (a semantic virus, analogous to the syntactic viruses proposed by Sobin 1994, 1997 and Lasnik & Sobin 2000).