In this article we note that diacritics, both in terms of their definition by the IPA, and in studies of transcriber reliability, are treated as a single group. Further, they are usually treated as being used purely to refine the meaning of a sound and, as such, as having less status phonetically than full symbols. It is argued here that diacritics should be classified into at least two major categories, and it is shown how one of these categories is the equivalent of a ‘full’ symbol. Apart from the implications this has for reliability measures, it is argued in conclusion that a more neutral definition of diacritic by the IPA is required.