- Language and Society among the Insular Celts AD 400-1000 -
The idea that the genitive might be being used elliptically - '(the stone) of Domnicus'
- can hardly account for the second and third of the examples just given. It is not
much more plausible for the following:
BROHOMAGLI I IC lAC IT I ET VXOR EIUS CAVNE76
Late British Latin must have been in the same situation as the one we find in modern
Italian: -i was the mark of the plural of second-declension nouns, but no longer of
the genitive singular. British, by the fifth century, was in the same position.^77 This is
what explains the numerous relics of the plural in -i (for example, early Welsh meip
from British ~-mapi) as opposed to the very few relics of the genitive singular in -i.
We may contrast this linguistic situation with that of a man who learnt his Latin from
grammars, for example the Anglo-Saxon Bede or the Irishman Tirechan. The former
writes much better Latin than the second, but both have learnt in the same way.
The kind of mistakes made by Tirechan are of two types: one is familiar to all
those who have had to learn a second language after childhood - he makes simple
grammatical errors;78 the second is more interesting - he writes a Latin in which Irish
constructions or idioms prevail over their Latin counterparts.l^9 Tirechan, however,
is perfectly well aware of the distinction between Latin cases, even if he makes
mistakes. Many of those who were responsible for the texts of the inscriptions were
unaware of any case system at all. They had not learnt their Latin from grammars.
Latin was, therefore, in the time of Voteporix, a spoken language, alongside Welsh
and, as I shall now try to show, Irish.
The clearest evidence that Irish was a spoken language in Wales comes from
Kenfig in Glamorgan, many miles to the east of the main Irish settlements in Dyfed.
A bilingual inscription runs as follows:^80
Ogam: POPIA[] II ROL[ .. ]N M[AQ]I LL[E]NA
Latin: PVMPEIVS I CARANTORIVS
The Latin inscription is in square capitals. There is nothing in the form of the inscrip-
tion to suggest that Pumpeius and Carantorius are different persons. The ogam
inscription, however, appears to be in two parts: one, on the left of the Latin inscrip-
76 ECMW, no. 183.
77 J.T. Koch, 'The loss of final syllables and loss of declension in Brittonic', BBCS 30 (1983):
201-33·
78 For example, Tfrechan, Collectanea, 6. I, ed. L. Bieler, The Patrician Texts in the Book of
Armagh (Dublin, 1979), 126, possimus for possumus, acciperunt for acceperant.
79 For example, 24. 1 (p. 140), quae tenuit pallium apud Patricium et Rodanum = gaihes caille
la Patraic 7 Rodin.
80 ECMW, no. 198, now in the Margam Museum; the stone is damaged and the reading there-
fore uncertain, but Nash-Williams's reading is a distinct improvement on that of
Macalister, CIIC no. 409. See also McManus, A Guide to Ogam, § 6.20, who confirms
P[.]P[, and The Royal Commission on Ancient and Historical Monuments in Wales, An
Inventory of the Ancient Monuments in Glamorgan, I.J The Early Christian Period
(Cardiff, 1976), p. 38, no. 849, reads P[O or A]P[ ... ]. On the form of ogam used for /p/
see P. Sims-Williams, 'The additional letters of the ogam alphabet', Cambridge Medieval
Celtic Studies 23 (1992): 39-44, esp. 42.