________________
AUGUST, 1899.]
A THEORY OF UNIVERSAL GRAMMAR.
199
as there are functions which it can fulfil. The functions a word fulfils in any particular sentence can be indicated by its position therein without or with variation of form, and, because of this, the form which a word can be made to assume is capable of indicating the class to which it belongs for the nonce. It is further obvious that words transferable from class to class belong primarily to a certain class and secondarily to the others, that a transfer involves the fulfilment of a new function, and that a word in its transferred condition becomes a new word connected with the form fulfilling the primary function, the relation between the forms, i. e., the words, so connected being that of parent and offshoot. Form, therefore, can indicate the class to which a parent word and its offshoots respectively belong.
This is the induction that leads me to argue that form grows out of function, or, to put it in the familiar way, accidence grows out of syntax, because when connected words differ in form they must consist of a principal part or stem, and an additional part or functional affix. The function of the stem is to indicate the meaning of the word, and the function of the affix to modify that meaning with reference to the function of the word. This modification can be expressed by indicating the class to which the word belongs, or by indicating its relation or correlation to the other words in the sentence.
But the stem itself may consist of an original meaning and thus be a simple stem, or it may contain a modification of an original meaning and so be a compound stem. A compound .stem must consist of a principal part or root and additional parts or radical affixes, the function of the root being to indicate the original meaning of the stem, and of the radical affixes to indicate the modifications by which the meaning of the root has been changed into the meaning of the stem.
Further, since words fulfil fanctions and belong to classes, they must possess inherent qualities, which can be indicated by qualitative affixes.
Thus it is that the affixes determine the forms of words, bringing into existence what is usually called etymology or derivation. They are attachable, separably or inseparably, to roots and stems and words by the well-recognized methods of prefixing, infixing, and suffixing, either in their full or in a varied form. It is the method of attaching them by variation of form that brings about inflexion in all its variety of kind.
Such is the line which I have long thought inductive argument should take, in order to work out the grammar of any given language or group of languages logically, starting from the base argument that speech is a mode of communication between man and man, expressed through the ear by talking, through the eye by signs, or through the skin by touch, and taking a language to be a variety or special mode of speech.
The grammar, i. e., the exposition of the laws, of any single language seems to me to stop at this point, and to carry the argument further, as one of course must, is to enter the region of Comparative Grammar. In doing so one must start at the same point as before, viz., the sentence, but progress on a different line, because hitherto the effort has been to resolve the unit of language into its components, and now it has to be considered as being itself a component of something greater, i. e., of a language.
To continue the argument. Since a sentence is composed of words placed in a particular order without or with variation of form, its meaning is clearly rendered complete by the combination of the meaning of its components with their position or forms or both. Also, since sentences are the units of languages, words are the components of sentences, and languages are varieties of speech, languages can vary in the forms of their words, or in the position in which their words are placed in the sentence, or in both. And thus are created classes of languages. Again, since the meaning of a sentence may be rendered complete either by the position of its words or by their forms, languages are primarily divisible into syntactical languages, or those that express complete meaning by the position of their words; and into formative languages, or those that express complete meaning by the forms of their words.
