dc.description.abstract | This thesis aims to provide a degree-based account of the scalar PAN-construction (Van Miltenburg & Zwarts 2013), illustrated in (1 a,b). Such constructions contain a Preposition, an Adjective, and a Noun (usually denoting some abstract property, like importance). The adjective says something about the degree to which this property holds. So in (1 a), the meeting has a high degree of importance, and this is indicated by
the size-adjective great (and vice-versa for the size-adjective small in (1 b)). When the adjective is left out, as in (1 c), the property is usually judged to hold to a significant degree. For example, (1 c) seems to imply that the meeting has a greater-than-average level of importance. That is: it seems to imply that the meeting is important. But where does this ‘greater-than-average level of importance’ come from?
(1) a. This meeting is [ PP of great importance. ]
b. This meeting is [ PP of little importance. ]
c. This meeting is [ PP of importance. ]
I argue that the scalar PAN-construction is interpreted through degree semantics (see e.g. Kennedy 2007 and the references therein), and that we can have a unified theory covering both the PAN-construction and expressions such as those in (2 a,b) that are discussed in Morzycki 2009.
(2) a. John is a big idiot.
‘John’s idiocy is big’
b. Mary’s a big goat-cheese enthusiast
‘Mary is very enthusiastic about goat-cheese’
Morzycki’s account of (2 a,b) assumes a covert MEAS operator that is fairly task-specific so as to account for what he calls the bigness generalization. I.e. the fact that John is an ADJ idiot does only receives a degree reading when idiot is modified by a bigness-denoting adjective such as big, and fails to get this reading with adjectives like small, little, etc. I show that Morzycki’s (2009) theory cannot readily be extended to cover PAN-expressions, unless we assume a more general version of MEAS (behaving more like a grammatical operator), and posit that nouns like idiot are evaluative (cf.
Constantinescu’s (2011) idea that such nouns just have an inherent bigness). This evaluativity in conjunction with what I will call the Modifier Domain Requirement can be used to account for the bigness generalization. With these elements in place, there can be a general semantics for gradable nouns. | |