Blair Collection > Celtic studies

138 EheVs Celtic Studies.
f immediately preceded the frequent loss of tlie p, thus for ex-
ample patar may have first changed into fatar^ and then into
athir; loss of a vowel in da ("of which", also "which" and "if")
for dian (from do-an, cf Z. 892), the auslaut of Avhich may still
be recognized in the eclipse foUoAving ; consonantal metathesis in
hearla, heurla for helre (hngua, sermo), for which herli is once
found in Z. 9, in haistim for haitsimm (baptize), eistini for eitsimm
(ausculto), eashog (jNIid. Irish easpog) for epscoj), Cornish, escojy
(episcopus).
However necessary in such cases we may find the Old Irish
in the elucidation of the ]\Iodern Irish forms, and however
clearly we may thereby discern the error into which the dii-ect
comparison of the latter with those of the other languages might
lead us, the comparison of the newer forms is not less instructive
and important for correctly understanding the older ones, nay, is
often indispensably necessary, and a closer attention to those
forms would have saved Zeuss from many errors. As sufiicient
preliminary investigations have not yet been made to render it
possible to give a systematic representation of Irish phonology,
I shall only touch in the following pages upon a lew points to
which my studies have led me.
§. 2. VocaUsmus.
The most difficvilt part of the Irish phonetic system to bring
to a fixed standard, is the Irish vocalismus, because three kinds of
e and o appear to exist, which do not always admit of being dis-
tinguished with certainty, and further, because even the question
of the priority of a or o, a or e, u or o, i or e in individual cases
is oiten beset with insuperable difiSculties (at least for the pre-
sent). In order to indicate graphically the threefold genesis of
the e and o without the use of new type, I propose, fii'stly, to
leave the e and o, which have arisen du-ectly from a without the
action of another vowel, unmarked, equally whether they
sounded e and o in Gaulish, or came into existence later by the
simple weakening of a (perhaps in the auslaut from e and 6?) ;
secondly, to mark the umlaut caused by i and u with the sign of
shortness, by which we gain at once a sign for original and
secondary i and u^ for ai and az<'"* diphthongal and such as arises
from umlaut; and lastly, to denote the breaking by «, especially
one hand, and dirium, eriU-su, airi (the feminine does not occur), erunn, dirib
(drhulj-si, airriu erriu crru on the other. For the only deviating form O'Dono-
van adduces Middle IvisXxforrnind, with which orrainn accurately agrees.
'"* Perhaps the mo?t convenient way would be just to write this umlaut every-
wlicre ni, au. This mode of marking appears to me to be very convenient for
Zend also, in order to distinguish the Tand u in gairi, taiiriDid, from the original
in yd us.