38 SISSANO. 
it lacks hai and gai, which we might expect; but the ai and kai forms 
are typical of the modulation which we have seen in item 1 and which 
we shall shortly see in the ani stem. The Dobu e’ai (2) seems to be 
ai with a preface; in Ray’s list there are evidences of the use of k, and 
there is nothing corresponding to e’ through which we might arrive 
at a better comprehension of the form. In Mugula qai (5) we en- 
counter the so-called Melanesian q, a phonetic unit which requires 
some explanation. 
From my earlier studies upon this subject I extract the following 
citation: 
We observe a group of forms in which the initial consonant is subjected toa 
wide yet systematic variation. ‘This peculiarity is known as the Melanesian 
q. Itis a composite of k and b and w; in this composite k may become ngg, 
b may become mb or p._ It is not to be interpreted as the effort on the part 
of Melanesians to compass an unwonted Polynesian sound, for it is of far 
greater frequency in Melanesian words for which we can find no affiliation 
with Polynesia. Rather are we to regard it as showing the struggle in sound 
evolution by a primitive people in the genesis of their speech who are coming 
into first possession of a labial mute and whose untrained buccal muscles reveal 
to us the wrestling.* 
Ray records the use within the area of the so-called Melanesian 
languages of Torres Straits as follows (page 419): 
The compound guttural and labial consonant, which is common in the 
Melanesian Islands, is found also in New Guinea, .but nowhere with its full 
sound kpw. Just as in the Island languages, as the guttural is sooner or later 
superseded by the labial the sound of k or p relatively predominates. In 
some languages, or in some words in one language, one or the other element is 
conspicuous, so conspicuous that either the guttural or labial is missed. On 
the mainland of New Guinea the sound is always a combination of a guttural 
with w, as kw or gw. It is remarkable that whilst the combination with a 
labial is not found on the mainland it is very common on the Louisiade Islands, 
where pw and dw are found in all the languages except Misima and Tagula. 
It is clear that q in the Melanesian languages as established most 
largely under the direction of Codrington and q of the New Guinea 
languages refer to different phonetic elements. The confusion arises 
from the fact that the missionaries to whom we owe the first records 
of speech in this region have mistaken q as the representative of any 
obscurely mouthed consonant. In Melanesia it is a labial which is 
thus obscured; in Torres Straits it is a palatal, for Ray writes “‘it is 
always a combination of a guttural with w.”’ 
Accordingly this qai is really kwai or gwai, and that is closely akin 
to kai or gai. 
STEM ani. 
Much of our material assignable to this stem is readily grouped after 
a simple series exhibited in this tabular form: 
ani ana an a 
kani kan 
gani gan ga 
hani hana han 
*The Polynesian Wanderings, 332. 
