catalogue of unknown or unfamiliar names, belonging to races and tongues that stand far off from our interests; but if its result shall be to give us a comprehensive view of the grand outlines, geographical and structural, of human speech, our hour will not have been spent unprofitably.
It must be borne in mind from the outset that the best classification of human languages now attainable is neither exhaustive, nor equally certain and reliable in all its parts. While nearly the whole field has been explored, it has not been explored everywhere with equal minuteness and care, nor by equally trustworthy investigators. In language, as in geography, there are few extensive regions which need any longer be marked "unknown;" yet there are many of which only the most general features have been determined: and that, perhaps, in part by inference, in part upon information which may turn out incorrect. It may be said in general that, where travellers' reports, or mere vocabularies, have alone been accessible as the ground of classification, the results reached are of superficial character and provisional value. No family of languages can have either its internal or its external relations well established, until its material has been submitted to analysis, the genesis and mode of construction of its forms traced out, and its laws of phonetic change deduced from an examination and comparison of all the accessible phenomena—until, in short, its vital processes are comprehended, in their past history and their present workings. To accomplish this for all existing and recorded human speech will be a slow and laborious task; and, for a long time to come, we must expect that the limits of families will be more or less altered, that languages now separated will come to be classed together, and even that some of those now connected will be sundered. It is not alone true that penetrating study often brings to light resemblances between two languages which escape a superficial examination; it also sometimes shows the illusiveness of others which at first sight appeared to be valid evidences of relationship. In a preliminary comparison, chance coincidences are liable to be overvalued. Moreover, the first tentative groupings are wont to be made by the more sanguine and enterprising class
21
