64 COMPARATIVE LANGUAGE 



the cases which they investigated. But it would seem important 

 to subject the cases of alleged functional association and analogy- 

 formation to a renewed scrutiny with a view of determining whether 

 the psychological process in these cases has not, perhaps, been mis- 

 interpreted. 



We say, for instance, that in the Oscan dialect the ending of the 

 nominative plural ~os of the masculine -o- stems has affected the 

 corresponding case of the relative pronoun (pos) and displaced its 

 regular and distinctive ending (seen in Latin qui), much as in vulgar 

 English the sigmatic plural of the noun has affected the personal 

 pronoun of the second person, changing you to yous. And, since in 

 these cases neither the meaning of the words nor phonetic resem- 

 blances can have given rise to associative connection, we are inclined 

 to attribute it to functional likeness. It may, however, be submitted 

 that there is another possibility, namely, the transfer of the termin- 

 ation of one word to an adjacent word simply on account of this 

 local contiguity. Words, we must remember, do not in actual speech 

 ordinarily occur isolated, but combined in phrases. Words like the 

 article or pronouns habitually occur in closest proximity to the noun 

 they qualify, and, in general, words with like grammatical function 

 cannot help being placed together in very many instances. Under 

 these conditions it would not be at all surprising if without any 

 realization by the speaker of their functional similarity the ending, 

 or the accent * of one member of the group should encroach upon 

 that of the other member, especially if both form a phonetic unit or 

 speech-bar. Such interference may operate in either forward or 

 backward direction, and its character would not be different from 

 that of the so-called regressive and progressive assimilation of sounds 

 within the same word. And this explanation is actually proffered 

 by Wackernagel (IF, xiv, p. 367), for some transfers of endings: 

 "Transfer of endings," he says (p. 374), "is not only due to propor- 

 tional analogy, but also to the fact that the words affected are 

 construed together. . . . Hence the influence of pronominal words 

 on nouns." And he appears to regard in this light the transfer of 

 the ending -s which, petrified, appears first in numerals like reVo^cs 

 with accusative function, and thence spreads over the adjacent 

 nouns. I am inclined to believe that what is now regarded as " func- 

 tional analogy" owes if not wholly, at least in good part its 

 existence to such spread. When we have, for example, in Greek 6/crw 

 and OTTTW (after cnra) I doubt whether the connecting link between 

 the two words was simply their grammatical category and that the 

 change originated in the isolated words, for the experiments which 

 I made concerning the association of numerals (American Journal 



1 For the latter see Brugmann, Berichte d. sachsischen Gesellschaft d. Wissen- 

 schaften for 1900, p. 371 (with references). 



