Expressives and Identity Conditions

We present diverse evidence for Pullum and Rawlins's (2007) claim that expressives behave differently from descriptives in constructions that enforce a particular kind of semantic identity between elements. Our data are drawn from a wide variety of languages and construction types, and they point uniformly to a basic linguistic distinction between descriptive content and expressive content (Kaplan 1999, Potts 2007).

1 Expressives

We use the label expressive to pick out the class of emotive morphemes, words, and constructions studied by Cruse (1986:271–273), Kaplan (1999), and Potts (2005:chap. 5). Typical examples are epithets like the jerk, expressive attributive adjectives like damn, honorifics, some discourse particles, and some uses of diminutive suffixes. We do not venture a definition of expressive here (see Potts 2007). For our purposes, an intuitive characterization is preferable, since we aim to provide additional descriptive tests for expressivity.

Expressives can be (and often are) fully integrated into the phonology and morphosyntax of the phrases that contain them. For the most part, they can be characterized in familiar terms. For example, expressive adjectives are strictly prenominal, which places them in a syntactic class with former and main.

(1)

1. a. the main road

2. b. *This road is main.

(2)

1. a. the damn dog

2. b. *The dog is damn.

(3)

1. a. the bloody television

   b. #The television is bloody. (literal reading only)

In languages with case marking on prenominal adjectives, such as German, expressives receive the usual case morphology.

(4) Du  hast  kein    verdammtes Wort gesagt.

you have no.ACC damn.ACC    word said

'You didn't say a damn word.'

Similarly, epithets have roughly the distribution of typical argument nominals (Lasnik (1989) and Büring (2005:123) provide more fine-grained characterizations). Honorifics in languages like Japanese and Korean trigger predictable morphosyntactic changes that are in keeping with other facets of those languages (Harada 1976, Boeckx and Niinuma 2004). And expressive particles in German are governed by diverse conditions on their position in the clausal hierarchy. Thus, in general, we find considerable evidence that expressives are part of the morphosyntactic system.

A growing body of evidence suggests that expressives form a linguistically coherent class. For instance, Aoun and Choueiri (2000) and Aoun, Choueiri, and Hornstein (2001) report that Lebanese Arabic epithets trigger special morphological effects, and Schwarz (2008) observes that German epithets resist determiner-preposition incorporation (e.g., vom Mann 'by the man', but *vom Idioten 'by the idiot'). And Potts and Roeper (2006) find that phrases like you fool! are morphologically impoverished in systematic ways across languages.

The next four sections present additional evidence for the linguistic coherence of expressives and suggest furthermore that this coherence is rooted in a basic semantic distinction between expressive content and descriptive content.

2 Balanced Constructions

We begin with three English constructions:

(5) Water or no water—I'm not hiking in      (NP or no NP)

this heat.

(6) Sue is as crazy as crazy can be.    (as AP as AP can be)

(7) I'll talk with the president, and the

president alone.                    (X and X alone)

The NP or no NP construction is the focus of Pullum and Rawlins (P&R) 2007. Examples like (8) seem to indicate that the two NPs in this construction must match.

(8)

1. a. *Water or no H₂O—I'm not hiking in this heat.

2. b. ??likelihood or no probability

(P&R 2007:284)

However, P&R provide corpus evidence that the matching condition cannot be stated in terms of string identity. They observe first that one of the NPs can be an elided version of the other.¹

(9)

1. a. War with Iraq or No War, Innocent People Are Likely to Die.

   (P&R 2007:(6a))

2. b. . . . its willingness to print this story, anonymous source or no, would seem to suggest there's some legitimacy to it . . .

   (P&R 2007:(7))

More importantly for our purposes, expressives are free to violate string identity for the NPs, as in (10).²

(10)

1. a. day trip or...