in such cases, the language itself affords us no other opportunity of learning its usage, than one single comparison of a word with others in connexion with it. And though the majority of words in a dead language may often occur, yet whenever the number of their senses bears a considerable proportion to the whole number of examples, the authorities for each single sense will be proportionally reduced.

To aid therefore our imperfect means of discovering by observation the usage of words, we must extend our inquiry beyond the mere relation of words to those who use them. We must consider the relation, which words, as signs, bear immediately to the notions, of which they are signs: and we must further inquire into the ground of that relation. For, though the meaning of words is no other than that, in which they have been actually used, we must not conclude, that usage is altogeth er fortuitous. Though the connexion between words and their notions is conventional, that convention may have been regulated by determinate laws. Indeed the connexion between words and their notions may have originated in various causes. But unless the causes are understood, we cannot judge of the effects. Let us inquire therefore into the origin of that connexion, which subsists between words, as signs, and the notions of which they are signs.

A word may be considered at present, either as

something seen, or as something heard: either as a written word, or as a spoken word: either as a visible, or as an audible sign, of its notion. But in the infancy of language, it was only an audible, not a visible sign. A word was then a mere sound, or utterance of the voice, conveying to the hearer some notion entertained by the speaker. And, though the invention of writing was introduced in so early an age, that all remembrance of that invention is lost in the darkness of remote antiquity, a considerable period must have elapsed before spoken words could have acquired a representation in written words. Indeed, before any attempt was made to write by the use of letters, it is probable that in every country some kind of hieroglyphic or picturewriting was employed. But representations of this kind had no connexion whatever with the use of letters: they could not even have led to the invention of letters. They were representations not of the words, but of the objects, to which the words referred. They were easy and obvious representations, when applied to external objects; nor was the transition difficult, when representations were wanted for things abstracted from the observation of the senses. Some resemblance to a visible object suggested a correspondent mark; as, for instance, when a circle, which is a line without end, was used in hieroglyphics, to denote a period without end. But, as soon as men began to write with those characters, which are called letters, they no

longer represented the objects, to which the words had reference. The thing then represented was the sound, or utterance of the voice, which denoted the object. Letters are elements, which are simply expressive of sound; and they were probably suggested by the different forms assumed by the mouth in the utterance of each single sound. In the most ancient languages, each letter was a distinct syllable, or distinct single sound; and hence they were easily combined into forms expressive of combined sounds. In this manner did the spoken word acquire a representation in the written word; and thenceforward they were so identified, that the word became no less a determinate sign to the reader of what was thought by the writer, than it was previously to the hearer of what was thought by the speaker.

In the interpretation therefore of words it is immaterial at present, whether we consider them as visible, or consider them as audible signs. But there is another relation between words and their notions, which has very material influence on the usage of them in every language. And in order to consider in what

understand this relation, we must

manner it is probable, that language itself was originally formed. The first notions, which men must have wanted to convey to others by the means of words, were notions excited by objects of the senses: and, when words had been provided for these notions, the next effort was the invention of words

for notions acquired by reflection. But here a difficulty occurred, which did not occur in the former case. The words, which were first employed in the infancy of language, to denote external objects, were probably, more or less, an echo to the sense. The particular tones, which were uttered by different animals, or were heard in the operations of inanimate nature, suggested probably the sounds or words, by which the first attempts were made to express the correspondent objects. And, though an object, which itself was destitute of sound, was more easily represented to the eye, than to the ear, more easily provided with a picture, than with a word, yet an object, even by its external form, or an action, by the mode of its operation, might have occasioned in the person, who was forming a sound for it, such a formation of the mouth, as produced & correspondent utterance. In short, external objects, as well as external actions, might, in various ways, which it is here unnecessary to detail, have suggested the sounds or words, which were originally used to denote them. But when words were wanted for things, which could be neither heard, nor seen, nor perceived by any other of the senses, there was no clue, which could lead directly to a sound corresponding with the thing to be represented. All notions, acquired by reflection, are excluded by their very origin, from any immediate resemblance with either visible or audible signs. They may operate indeed mediately, if they operate

en the passions: for in that case an effect may be produced, either in the voice, or in the gesture, which may give rise to a sound corresponding with that effect, and therefore indirectly with the cause, which produced that effect. But if the notion was so abstracted from all sensible effect, as to produce no external mark, which might have suggested a correspondent sound, a sound, or word, must have been provided for it in one of these two ways. Either an arbitrary sound must have been invented, without any attempt at similitude between the sound and the thing to be represented by it; or some similitude must have been sought between the abstract notion, for which a word was wanted, and some other notion, already provided with a word. The latter mode was not only more easy and obvious; but also more consonant with an early state of civilization, when the imagination is always more employed in finding resemblances, than the judgment in discovering differences. In such cases therefore, it would frequently, and perhaps commonly happen, that words already provided for one purpose, would, for want of new words, be applied to another purpose, in consequence of some resemblance, whether real or imaginary, between the primary and the secondary purpose.

In this representation of the origin and formation of language, we see the foundation of those distinctions in the senses of words, which are observed in all languages, and which are expressed by