POLYNESIAN REUCS IN MELANESIA. 133 



Distinctively Melanesian mutations are : M to mw, to n, to ng, to v, to 

 r, to t. The mw-change is scarcely to be classed as a mutation; it is a 

 fashion in pronunciation which exists but in a restricted area of the Banks 

 Group in the northern New Hebrides. 



The purely Melanesian mutations are : M to v, to r, to /, to n, to ng. 



Common to the two families are: V to h, to w. The latter alone has 

 width of extent in either; x-h in Polynesia is found in but one word each 

 in Niue, Nukuoro, and^Rotuma; and in Melanesia in but one word common 

 to Saa and Wango. f j 



Distinctively and wholly Melanesian are these mutations: V to/, to b, 

 to mb, to p, to pw, to u, to t, to ku, to extinction. 



Common to the two families are : F to v, to h, to p, to b, to w, to extinc- 

 tion. In Polynesia the widest extent marks the mutations to //, to v, and 

 extinction; in Melanesia the order of frequency is v, w, p, h, extinction. 



Distinctively Melanesian mutations are: F to mb, to mbw, to k, to kr, 

 to n, to ng, to s, to th, to t, to u. Of these the mb and mbw are reducible 

 to p (b), and the u to the common w. 



The purely Melanesian mutations are: F to k, to kr, to n, to ng, to s, 

 to th, to t. 



Common to the two families are : P to b {mb), to v, to h, the last occurring 

 in a single word common to Rotuma and Wango. 



Distinctively and wholly Melanesian are: P to mp, to m, to kpw, to/, to 

 bu, to /, to iv, to extinction. 



In a former paper in which I subjected the truly Polynesian lan- 

 guages to a similar detailed examination* it was pointed out that 

 with certain exceptions noted the wdiole play of consonant mutation 

 was a vertical or series matter : 



One more preliminary statement: we have already said that for con- 

 venience we should enter upon our alphabetical conspectus the aspirate in 

 the neighborhood of each of the three series. The convenience is this, that 

 the aspirate is not palatal, not lingual, not labial, yet it lies as close to the 

 one as to the other. We shall find it involved in all these changes, but it 

 does not affect the rule which we are about to enunciate. 



With the three exceptions noted (s-v, ng-n, t-k) the whole play of con- 

 sonant mutation in Polynesian is a matter of vertical change. When a 

 palatal changes it changes to another palatal, lingual modified remains 

 lingual still, and labial remains labial even though its play of mutation 

 carries it bodily into the vowel tract. But there is no horizontal move- 

 ment, the labial under stress of change does not become palatal or lingual. 



♦"Samoan Phonetics in the Broader Relation," 17 Journal of the Polynesian 

 Society, 217. 



