255 



we should of course expect the ending -it (*uu'it instead of 

 iiiX'ut). We do not meet with this difficulty in those words 

 which end in -ik (malik etc.) or -eq (taleq) in the singular. But 

 in those words of this kind which end in -oq or -ag (na'laqaq) 

 in the singular (in the plural respectively -nt or -at) there are 

 accordingly two irregularities to be explained: the absence of 

 uvularization in the last syllable but one and the presence of 

 a {ä) or и in the last syllable. Many more examples of non- 

 uvularized plural forms might be added to those already given 

 (cf. pp.243 and 248): 



niq'aq (snare), т^\ш. n'i^'ät; qatiq'aq (aback), phir. qat'i^'ät', 

 nan^oq (bear), plur. nkin-ut] tas*eq (lake), plur. t^dt'Ht etc. 



Most of these plural formations may belong to a latest 

 stage in the development of the plural inflexion of nouns; 

 perhaps there is no longer any question here of an assimila- 

 tion of q (or k) with the inner consonant of the stem, but 

 they may simply be analogical formations after the earlier 

 models. The shifting of stress has been felt as the main 

 principle in the formation of the plural form and along with 

 it has followed quite mechanically, so to speak, that change 

 of the stem consonant which consists in its lengthening and, 

 in so far as it is a non-nasal, its unvoicing. As for the suffix, 

 however, the original manner of constructing the plural is 

 retained, namely by the addition of -t to the stem vowel. 



§ 35. Historical perspective. Although next to 

 nothing is known about the language spoken in large portions 

 of the Eskimo world — I include here the American polar 

 archipelago and in part Baffin Land, the north coast of the 

 American continent south of the archipelago, the coasts of 

 Hudson Bay, the northern coasts of Alaska and the Asiatic 

 side of Bering Strait — yet the three or four dialects that 



