34 BUREAU OF AMERICAN ETHNOLOGY [bull. 40 



cography; parts expressing the modifying relations, as the subject- 

 matter of grammar. In modern Indo-European languages the num- 

 ber of ideas which are expressed by subordinate elements is, on the 

 whole, limited, and for this reason the dividing-line between grammar 

 and dictionary appears perfectly clear and well drawn. In a wider 

 sense, however, all etymological processes and word compositions 

 must be considered as parts of the grammar; and, if we include those, 

 we find that, even in Indo-European languages, the number of classi- 

 fying ideas is quite large. 



In American languages the distinction between grammar and 

 lexicography often becomes finite obscure, owing to the fact that the 

 number of elements which enter into formal compositions becomes 

 ver}^ large. It seems necessary to explain this somewhat more fully 

 by examples. In the Tsimshian language we find a very great number 

 of adverbial elements which can not be considered as entirely inde- 

 pendent, and which, without doubt, must be considered as elements 

 modifying verbal ideas. On account of the ver}^ large number of these 

 elements, the total number of verbs of motion seems to be somewhat 

 restricted, although the total number of verbs that may be com- 

 bined with these adverbial ideas is much larger than the total number 

 of the adverbial ideas themselves. Thus, the number of adverbs 

 appears to be fixed, while the number of verbs appears unlimited; 

 and consequently we have the impression that the former are modi- 

 fying elements, and that their discussion belongs to the grammar 

 of the language, while the latter are words, and their discussion 

 belongs to the lexicography of the language. The number of such 

 modifying elements in Eskimo is even larger; and here the impres- 

 sion that the discussion of these elements belongs to the grammar of 

 the language is increased by the fact that they can never take an 

 initial position, and that they are not placed following a complete 

 word, but are added to an element which, if pronounced by itself, 

 would not give any sense. 



Now, it is important to note that, in a number of languages, the 

 number of the modifying elements may increase so much that it 

 may become doubtful which element represents a series of ideas 

 limited in number, and which represents an almost unlimited series 

 of words belonging to the vocabulary. This is true, for instance, in 

 Algonquian, where in almost all verbs several elements appear in 

 conjunction, each in a definite position, but each group so numerous 



