Sonority hierarchies vary somewhat in which sounds are grouped together. The one below is fairly typical:
It should be noted that more finely nuanced hierarchies often exist within classes whose members cannot be said to be distinguished by relative sonority. In North American English, for example, of the set /p t k/, /t/ is by far the most subject to weakening when before a vowel not stressed (v. the usual American pronunciation of /t/ as a flap in later, but normally no weakening of /p/ in caper or of /k/ in faker). Similarly, Romance languages often show geminate /mm/ to be weaker than /nn/. In such cases, many phonologists refer not to sonority, but to a more abstract notion of relative strength, which, while once posited as universal in its arrangement, is now known to be language specific.