496
AN GAODHAL.
where every vowel is broad. In the Turkish if we
take the root sev (to love), we have for example,
sevildirememek ("not to be able to causes oneself
to be loved"); and bashlaya namak ("not to be
able to begin"). In Yakootic the vowel harmony
is very strictly observed and more developed than
elsewhere as the broad and slender vowels are
there, again subdivided into heavy and light,
which makes the harmonization of the syllables
very complicated, but most rigorously fixed and
determined in every case. In all these languages,
it is invariably the stem which dictates the nature
of the vowels that are going to stand in the suffixes.
The principle of vowel-harmony constitutes one of
the chief distinguishing features, one of the most
striking peculiarities of this far-spread family of
languages; and where this law is disregarded, it
must be considered simply as the result of pho¬
netic decay; while we see it most strongly showing
its power where artificial influences, such as writ¬
ing and literature, have least interfered. In the
same way, we find that in old Irish there was a
time when this rule, (narrow with narruw, broad
with broad), was not called into action. So we have
in Mongolion a first step towards a loosening of
this principle in the fact of i having become neu¬
tral, either broad or slender. Other languages of
that class have a hard i (represented in transcrip¬
tion by y) and a soft or weak i; the hard or broad
is wanting in Mongolian, hence no further con¬
trast exists in relation to that sound, and
it has become neutral and apt to follow
any syllable whether broador slender. Also
in the Moksha-Mordwin tongue we find the
vowel-harmony imperfectly and inconsistently
applied, probably the consequence of its
having been hindered in its full developmant, or
its being counteracted by long continued foreign
influences: although also there the rule is, gen¬
erally speaking, that the stem vowel should take
the lead and determine the class of all the follow¬
ing vowels in the same word, e. g., sivel (meat),
genitive, sivelin, sedi (heart), sedida (from a
heart), etc.
In Tcheremisian we find that there are two dia¬
lects among that tribe which live in the govern¬
mental distrists of Viatka and Kasan. These two
dialects are divided by the Volga river. On one
side the language has the law of vowel-harmony,
while on the other side of the river this law does
no longer exist. Also in the Telugu language,
traces of vowel-harmony are found. Thus, the
copulative particle is ni, after a a preceding i, i
or ei. but it is nu when u or hard vowels precede.
The Dative participle ki in the former case, and
ku in the latter. So in the decension: e. g., katti
(knife). Plural with the ending lu becomes kattulu
instead of kattilu: Dative kattiki, but in the
Plural kattuluku. So in the verse e. g, kalugu (t
be able), Aorist Kalugudunu (all broad vowels),
but Preterite Kaligitini (with the slender vowels)
As to the consonants there are in the Ural-Altaic
languages only the gutturals that are double in
nature and receive a double form according to
their being broad or slender; the former requires
hard vowels, the latter slender vowels.. Such is
the case in the Tartar-Turkish languages, in Mon¬
golian and Tungusic, also in the Ostyak something
similar is observed. In the Arabic where we have
no vowel harmony, there are, nevertheless, to a
limited extent, it is true broad and slender vowels;
in the pronunciation of a word, it depends on
whether the leading consonants be hard (broad or
soft (slender, narrow), to have the accompanying
vowels pronounced with a hard or soft sound,
that is broad or slender, Thus s, d, t, z, h, k, are
pronounced when markad with fatha, as e, while
the emphatic consonants s, a. y, th, bh, g, with
fatha are pronounded as a. The application of
this law of vowel-harmony takes, moreover some¬
what different forms in the several branches of
these languages : as we see, likewise in Irish that
the addition of a slender syllable produces a cor¬
responding change or attenuation in the one that
precedes; this is just the reverse of what takes
place in the Ural-Altaic languages. There occurs
in fact something similar to what we have seen in
regard to the initial consonant, changes in Celtic
when compared with the terminal changes in the
Sanskrit and other Aryan languages; the phonetic
influence in Sanscrit going always back to the pre¬
ceding part of the word or to the preceding word,
thereby moving as it were, in the opposite direct¬
ion of what it does in Celtic. Thus, also in the
vowel-harmony in the Ural-Altaic languages moves
onward from the root or stem to the termination;
the vowel-harmony in Celtic moves backward from
the ending. Another difference is — in the former
language it runs through the whole word, forming
a homogeneous chain or series of syllables to
which the key-note, so to say is given by the root-
vowel; in Celtic it only affects the preceding, con¬
tiguous part of the word. It is also worthy of
notice that in the suffixes which we add to words
or stems in those Ural-Altaic tongues, only vowels
of one and the same class, as that of the root, or
as that of the last syllable of the root (if there are
more than one) are also allowed to occur. Hence,
every such suffix presents a double appearance, or
has two forms, in which the consonants remain
the same, but the vowels are of different class; so
that one form is with strong or broad vowels, and
the other form with weak or slender, — either of
which is used as circumstances (resulting from the
nature of the root vowel or the radical syllable),
may require. Thus, we have in Hungarian the
endings or Genitive and Dative ang, nak and nek;
accus at and et, the endings of the comparative
are abb and ebb, the pronominal suffixes am and
en (my), ad and el (thy), a and e (his), atok and
etek (your), ok and ek (their), etc.: in Turkish, we
