X-Cc: Multiple recipients of list WELSH-L To: Multiple recipients of list WELSH-L In-Reply-To: (null) The best general book on the mutations (these included) is T. J. Morgan _Y Treigladau a'u Cystrawen_. Historically, it's words that originally ended in vowels (before loss of original Celtic < Indo-European final syllables in early post-Roman times) that trigger lenition. That's the historical principle. Since apopocope however there has been a lot of analogical mix and match. In the examples you cite, there is a pronoun element immediately preceding the lenited word. Nouns or mere pronominal inflection (e.g. mae gennyt gi gwyn.) would do the same. Reconstructing earlier forms some of the forms that now trigger these mutations would originally have ended in a vowel and some would not. In Middle Welsh the mutations are more irregular in such cases and not always written, but you could see how the analogy spread. Morgan would cover it. Synchronically, the current native-speaker feeling might be, in something like Mae gennyf gi that it has do with separating the verb from its subject (ci). In the impersonal (e.g. gwelir ci) you don't lenite the passive object\subject, which might be because it is not felt to contain an incorporated pronoun in the inflected ending (again synchronically). Once the mutations ceased to be phonetically conditioned they have had to respond to a different set of forces as morpho-phonemes. In a nucleus of cases, the lenition clearly reflects old intervocal consonant weakening, e.g. gwraig wen `fair married woman' < Celt. *wraki+ winda+, vs. gw^r gwyn < *wiros windos, &c.