RELATIONS OF COMPARATIVE GRAMMAR 47 



present "dialect." For we need a term to designate the speech of 

 certain territories or communities, without regard to the precise 

 degree of variation (or unity) represented. 



But the very existence of dialects as ordinarily understood has 

 been denied, especially by certain eminent Romance scholars, one of 

 whom is honoring this Congress by his presence. Emphasizing what 

 has come to be rightly an accepted belief since J. Schmidt's exposition 

 of his famous wave-theory, namely, that a linguistic change starts 

 at a certain point and gradually spreads over contiguous territory, 

 and that different linguistic phenomena may start from different 

 centres and so cover wholly or partially different territories, they 

 conceive the resulting conditions to be such as would be illustrated 

 graphically by a large series of intersecting circles drawn from differ- 

 ent centres and representing the areas of the different linguistic 

 phenomena. They assert that there will be only an infinite series 

 of gradual variations, that we may if we choose give the name 

 dialect to the area of a particular linguistic phenomenon, but that 

 any broader grouping of dialects is purely arbitrary and unscientific. 

 Such a conception is possible only upon the basis of purely linguistic 

 theorizing, defying every historical probability. If we could imagine 

 a given territory occupied all at once by a people of uniform speech, 

 in settlements equally large and equally distant from one another, 

 like the squares on a checker-board, with no natural boundaries by 

 mountains and rivers, and further imagine that these settlements 

 remained of the same relative strength, no one of them gaining 

 predominance over others, then, indeed, speech-variation might 

 proceed with such a result as has been pictured. But such conditions 

 never exist. Even if the incoming people were wholly homogeneous 

 without even the germs of dialectic variation, which is rarely if ever 

 the case, there would inevitably arise certain social and political 

 groupings which would reflect themselves in speech. Some degree of 

 centralization is as certain in speech as in politics. The evolution 

 of a standard language is only the culmination of what on a smaller 

 scale is always operative. There is no time when the centrifugal 

 force of speech-variation starting from innumerable centres is not 

 being more or less counteracted by a centripetal force combining 

 certain phenomena in groups. The extent and the defmiteness of 

 these groups vary with the historical conditions. How clearly do the 

 linguistic conditions of ancient Greece reflect that particularism 

 which was so characteristic of the Greeks politically! No single 

 standard of speech until a late period, just as there was no political 

 unity, but numerous dialects, as there were numerous states, show- 

 ing centralization within certain limits. And will any one deny the 

 existence there of well-defined dialects so clearly marked by certain 



