I thank both Pat Keating and Janet Pierrehumbert for their thoughtful and constructive oral comments on my paper during the conference. I especially thank Janet Pierrehumbert for offering the above written commentary.

Pierrehumbert raises the issue of the connection between generality and reductionism (“A researcher … has a choice whether to aim for reduction or for generality”). The connection or lack of it is, I think, a very simple matter. In explanations the link is obligatory. To explain the unfamiliar by reducing it to the familiar means to bring the unknown into the fold of the known and therefore to enlarge the domain to which the explicanda apply, thus achieving greater generality. An example is Watson and Crick's explanation of the genetic code and the mechanism of inheritance by reducing it to previously known chemical facts, e.g. how adenine bonds only with thymine and cytocine only with guanine in such a way as to guarantee the construction of exact copies of molecular chains (DNA) consisting of those substances. Needless to say, in cases like this it may take genius and inspiration to figure out which facts to bring together into an explanation.

One can achieve generality without reduction but then this is a form of description. Explanation is deductive generalization; a systematic description of a sufficiently wide range of phenomena is inductive generalization. One of the points in my paper was that nonlinear phonology is a good description of certain sound patterns. Pierrehumbert seems to agree with me, then.