Seminaire SynSem

Janek Guerrini (Frankfurt)

Italian pre-nominal modification as a window into semantic composition

In this project, I argue for a new classification of adjectives, based on mode of composition and disregarding their emergent entailment pattern. It divides adjectives into set-denoting adjectives like ‘Greek’ or ‘hexagonal,’ context-sensitive functional adjectives like ‘good’ and ‘fake,’ and non-context-sensitive functional adjectives like ‘former’ and ‘potential.’ The classification is based on whether adjectives can occupy given syntactic positions, which are specialized for specific semantic modes of composition, and on the readings they can give rise to in these positions (see Cinque, 2010; 2014). A prominent role in this classification is assigned to the pre-nominal position in Italian, a marked position which has often been proposed to be specialized for ‘functional’ adjectives (Cinque, 2010; 2014) since set-denoting adjectives (i.e. ⟨e, t⟩ type adjectives) are banned in it (*un greco avvocato, ‘a Greek lawyer’). What emerges is a subdivision similar to one proposed by Higginbotham (1985) on different grounds, which also echoes a number of distinctions drawn in the literature, such as for instance by Siegel (1976) and Partee (2010).

Having established this classification, I present two case studies where the relevant tests are used as a window into the compositional behavior and core meaning of two groups of adjectives: ‘privative' adjectives like ‘fake’ and color adjectives like ‘red.’ ‘Fake’ has traditionally been considered a challenge to conventional semantic accounts of adjectival modification due to the complexity of its meaning (Del Pinal, 2015; 2018, a.o.). Based on data from the Italian pre-nominal position, I argue that ‘fake’ has in fact a very simple, conjunctive meaning: a fake N refers to something that (i) intends to resemble a typical N and (ii) is not, in fact, an N. This captures a number of puzzling data points and makes fine-grained and correct predictions for the range of readings of complex cases such as iterated ‘fake,’ as in ‘fake fake N.’

Concerning color adjectives, I show that Italian highlights aspects of their meaning that are only emergent in other languages previously considered in their analysis such as English. When post-nominal, Italian color adjectives have the range of readings that is familiar from English: both ‘red pen’ and ‘penna rossa’ may denote a red-inked pen or a pen with a red outer surface. When pre-nominal, instead, Italian color adjectives can only target the visible part of objects denoted by the noun they modify – ‘rossa penna’ (‘red pen,’ pre-nominal) can only denote a pen with a red outer surface. This systematic behavior is difficult to explain on most existing accounts of color adjective composition, which rely too heavily on context (Travis, 1985; Rothschild and Segal, 2009; Del Pinal, 2018) or bona fide ambiguity (Kennedy & McNally, 2010). My approach argues, in the spirit of Szabó (2001), that color adjectives have an open property slot. This is generally filled contextually, but must be filled by a property identical to the one provided by the input noun in Italian pre-nominal position – an assumption independently justified by other work on other Italian pre-nominal adjectives, cf. Cinque (2014). This also captures their behavior with the tests from the overall classification I argue for, which is strikingly similar to other context-sensitive functional adjectives like ‘fake’ and ‘good.’