GRAMMAR OF ISOLATING SPEECH. 19 
. cumstances be brought together, and open-mouthed words all vocalic 
and without a single stouter phonetic element are frequent. But in 
Melanesia there is no such uniformity. Certain of the languages 
prefer the vowel ending; in certain neighbor languages this is highly 
objectionable; terminal abrasion is applied upon the weak ending and 
the vowel is rubbed off for the stronger ending. We find instances in 
great plenty of this feature carried to the second degree; a word in 
passing from a language of open habit to one of the closed habit loses 
its final vowel, but in yet more distant passage through this medium 
to another language of open habit it undergoes still another loss: the 
then final consonant is thrown aside to expose a vowel which in the 
beginning was inner. 
In the matter of the concurrence of consonants there is similar 
disparity. In the Melanesian languages of open habit no two con- 
sonants may come together. There is next a considerable group of 
languages in which certain double consonants not only are permitted, 
but are rigorously required. This is particularly the case with the 
sonant mutes. In this group, somewhat widely dotted over the geo- 
graphical area, the sonant mutes are unspeakable without the sup- 
port of the preface of the nasal, each of its proper series, the palatal 
mute requiring the palatal nasal, the lingual mute the lingual nasal, 
the labial mute the labial nasal. The following illustrate the principle: 
In the mutation from surd palatal mute to spirant the k of the Poly- 
nesian kamu may not in Viti become g simply, but requires the 
preface of the palatal nasal ng, thus becoming uggamu, a double 
consonant which is fairly represented by the ng of our English word 
finger. In like manner ¢ must become nd, and the Samoan fut is 
vundi in Vaturanga, Nggela, and Bugotu. So p must at its simplest 
mutation become mb; the Polynesian pongis appears as mbongt 
in Nggao, Belaga, Nggela, Vaturanga, Bugotu, Omba, and Sesake. 
Because, as already remarked, the labials are but just acquired and 
imperfectly tamed the word may become a wretchedly uncouth 
mouthing, as shown by these other pongis forms—kpwon in Arag, 
Vuras, and Lakon; mpwonin Mota; kwon in Lo; kmbwon in Maewo, 
Gog, Motlav, and Norbarbar; nggmbwon in Volow. This intermedi- 
ate class admits concurrent consonants only in this specific instance, 
not so much double consonants as grace notes. A more considerable 
group of Melanesian languages has no objection to concurrent con- 
sonants, no matter what they may be. 
Such matters of form are far from being the really distinctive 
character of isolating languages. A much more vital quality is the 
evaluation of words as agencies of speech. Not for long ages of 
evolution are we to approach the seemingly simple system of parts 
of speech with which we are familiar until grammatical formalism 
clouds the clarity of our analytic language. The great difficulty 
