Introduction

We often express our knowledge about the world in sentences such as the following:

(1)

1. a. Ravens are black.

2. b. Tigers have stripes.

3. c. Mary jogs in the park.

We refer to such sentences as generics. They appear to express some sort of generalization: about ravens, about tigers, and about Mary, respectively. Yet it is far from clear exactly what they mean. What does it mean to say that some generalization holds?

It turns out that there are a great many theories trying to answer this question. This, in itself, is a fact that is in need of explanation. Why are generics so puzzling? What is it about them that forces researchers to come up with one theory after another, with no clear agreement on what the correct theory is? And, if they are so strange, why are generics so prevalent?

In this article we will consider some of the puzzles concerning generics, why they are so hard, and the various solutions proposed. Let me say at the outset that readers who expect to find definitive answers to these puzzles will, regrettably, be disappointed. But if not the answers, I hope this article will at least convey the depth and significance of the problems.

Two types of generic

Before we begin our investigation, we need to note that generics do not constitute a uniformphenomenon. A distinction that is often drawn is between direct kind predication and characterizing generics.

To see this distinction, consider the following typical example:

(2) Dinosaurs are extinct.

There is no individual dinosaur that is extinct; individual dinosaurs are just not the sort of thing that can be extinct – only the kind dinosaur can have this property. A natural account of (2) is that it predicates the property of being extinct directly of the kind dinosaur. Krifka et al. (1995) refer to such sentences, which predicate a property directly of a kind, as cases of direct kind predication. They distinguish between them and sentences such as (1a), which predicate a property of instances of a kind, and not of the kind as a whole; these are named characterizing generics.