En réponse à Ian Spackman : >Surely it's not a theoretical impossibility for a phonemic cluster to be >realised as an acoustic affricate? Of course not, especially diachronically speaking (affricates often originate with consonant clusters). But such a situation is according to me very temporary. When it arises, the phonemic cluster gets quickly reanalysed as a single phoneme, in the sense that it's how people will hear it. Christophe Grandsire. http://rainbow.conlang.free.fr You need a straight mind to invent a twisted conlang.