50 J. A. Fontaine, 



indicates an action, and generally our feeling is influenced by 

 this distinction ; but I have tried to show that such was not 

 the case in the Old French period. 



It has already been assumed, in the first part of this trea- 

 tise, that every verb is active, and consequently all the verbs 

 that have come down to the Romance languages from the 

 Latin mother tongue must have been verbs belonging to the 

 active voice. . To explain this phenomenon we must go back 

 to the early period of the Latin, just as to explain the Italian 

 forms strnggere and traggo we have to go back to the period 

 when the classical forms struere and traho were represented 

 by the more primitive forms strngere, trag(Ji)o. The g of 

 these latter forms is still preserved in the sigma perfects, 

 traxi, striixi standing for trag-si, strug-si. By going back to 

 the formative period of the Latin, we shall see that the active 

 voice was the only voice this language then possessed. 



It has been said (cf. Die 'Verbal Flexion der lateinischen 

 SpracJic of Westphal and others) that the Greek constructed 

 its middle and passive voices (the aorist and future excepted) 

 with the same inflexional endings, y^ai, aai, rai, and these 

 endings were originally, as the comparison with other lan- 

 guages proves, endings of the middle voice, which afterwards, 

 through a transfer of meaning, were used as passive endings. 

 The Latin verb must have once possessed endings similar to 

 the fiai, aai, rat of the Greek, and with a parallel meaning ; 

 but they were afterwards lost, a periphrastic expression being 

 introduced in its stead. That periphrastic expression was 

 made up of an active form, e.g. aino, and the reflexive se. 

 But the Latin did not keep this new formation in its primi- 

 tive state, and a fusion of the periphrastic expression into a 

 single word took place, and anio se became avior{e), amas-se, 

 ainar{i)s, etc. Just as the endings y.ai, crai, rac had served 

 as inflexional endings to the midJle and passive voices, so the 

 parallel Latin endings formed its passive and deponent forms. 

 Thus four periods might be distinguished in the verbal growth 

 of the Latin, viz. : The first period, when the Latin had forms 

 similar to that of the Greek in /^ai, aai, rat ; tJie second 



80 



