On the Semantic Markedness of F-Features

When linguists talks about features, they usually talk about markedness as well. One reason is that feature systems are more efficient if there is an unmarked default value contrasted with a marked value. Nevertheless it is often difficult to determine which feature value should be regarded as the unmarked one. Φ-features are a particularly interesting case since they are important in many different domains of linguistic inquiry, and therefore markedness considerations arise in several different ways. To my knowledge, Greenberg (1966) was the first to investigate markedness in the domain of φ-features. He presents several tests from different domains for markedness. Later works (Noyer 1992, Harley & Ritter 2002), focus more narrowly on morphological markedness. My focus, however, is semantic markedness. One of Greenberg’s test for markedness, which I discuss below as Dominance, is semantic. I develop three other tests for semantic markedness. Using these tests, I then investigate the semantic markedness of person, number, and gender features. In the person domain, I conclude that there is clear evidence that third person