by Dr Ken George
When I learned Unified Cornish in the 1970s, people asked me: "How do you know that you are pronouncing it correctly?" This bothered me considerably, so that I decided to study the problem in depth. I first learned Breton, the language closest to Cornish, and spent a year in Brittany, during which time I studied Celtic linguistics, and was awarded the Diplôme Supérieur d’Études Celtiques. On return to Cornwall, I spent three years conducting research on the phonological history of Cornish, and submitted the results in a thesis, for which I was awarded the degree of Doctorat du Troisième Cycle by the University of Western Brittany. I also investigated how close Cornish was to Breton throughout the centuries .
My study of the phonology of Cornish was a broad-brush one. Many of the details have been filled in since, but certain features need further mention.
The feature which most distinguishes Cornish from Breton and Welsh is the change of Old Cornish d to Middle Cornish s, e.g. ‘father’ is tad in Breton and Welsh, but tas in Middle Cornish. This sound-change is called assibilation.
The same phenomenon is evident in the middle of words, but here there is an added complication; the original -d- sometimes appears as -s- and sometimes as -g-; e.g. Breton pediñ, Welsh pedi ‘to pray’, but Middle Cornish pysy ~ pygy. The change to -g- is called palatalization. Nance was well aware of this variation; he gave both forms in his dictionaries, as s (the favoured form) and j.
In my thesis, I misunderstood the nature of the change to -g-. Nicholas Williams (1990) argued that, particularly before high front vowels, Old Cornish /d/ evolved in two different ways: assibilation to [z] and palatalization to [dZ]. He also suggested that the difference was dialectal; assibilation in the east and palatalization in the west.I revisited assibilation and palatalization and the question of dialects. There probably were dialects in traditional Cornish, but until recently it has been possible to explain all observed features without recourse to invoking them. In the case of s and j, however, a study of place-names showed that s dominated in Powder hundred and j in Pydar hundred.
One non-English feature of Cornish is the existence of double consonants; here is a handout in Cornish which may help the learner. The double consonants /mm/ and /nn/ are noteworthy, because they suffered pre-occlusion, and became [bm] and [dn] in Late Cornish; the case of alemma > alebma was examined by Nicholas Williams in his book The Cornish consonantal system, but his reasoning appears false .
In my thesis, I showed that Middle Cornish had two long o-type vowels: [):], which came mainly from Old Cornish /)/, and [o:], which arose when Old Cornish /ui/ lost its diphthongal character. The two long-o vowels are phonemically distinct, as is shown by the minimal pair bos ['b):z] ‘to be’ v. boes ['bo:z] ‘food’. This discovery was contested by Nicholas Williams in his book Cornish Today (1995), and so the evidence was presented again in Kernewek Kemmyn - Cornish for the 21st century, and again in Agan Yeth no. 2. Williams could still not accept this, and continued to argue in his book Towards authentic Cornish (2006). So I presented the evidence in a different way , but I doubt if Nicholas Williams will ever acknowledge my discovery.
A residual problem is how to spell the closer vowel /o/ .
Apparently free variation between y and e is found in such words as kynsa ~ kensa ‘first’. A more consistent vocalic alternation is observed (but not in all texts) between pairs of words such as gwyth ‘trees’ and gwethen ‘tree’ (monosyllables v. polysyllables). Dunbar and George (1997) suggested that this second variation reflects a phonetic difference only of quantity (long v. half-long versions of the same vowel), but Nicholas Williams claims that it represents primarily a difference in quality.
In an attempt to throw more light on this problem, I began to look again at the long front vowels . The results of an examination of w-diphthongs also have a bearing on this topic. More work is required, however.