Language is a social product, at once the creation and the creator of society. It is independent of the caprice of the single individual, and the Emperor Tiberius could no more change a Latin word[77] than the slavish obedience of a Benedictine monastery could turn sumpsimus into mumpsimus. Unless the community as a body agree to accept the new word or form, Cæsar himself is powerless to introduce it. The changes undergone by language are brought about by the action of circumstances over which the individual has no control. They are circumstances which affect the whole community, not the individual member of it. The primary condition of speech that it should be significant requires that it should be stamped and recognized by the common consciousness. Now, the circumstances that affect a whole community will always act in the same way should the conditions remain the same. Individual caprice is rendered impossible, and the forms assumed by language will be found referable to general laws. We have to deal, not with the infinite complexity of individual motives and caprice, but with the consentient action of many minds swayed by the same feelings, surrounded by the same atmosphere. The joint action of a multitude eliminates the accidental differences of individual character; all that is left is just that in which all agree, the result of the influences of which all alike are sensible. The circumstances that determine the common nature of a society determine also its common utterance, and this common utterance we call its language. It embodies all the past life and history of the community that speaks it; each phase in the development of its speakers is reflected in it as in a mirror, and its worn-out words and forms are so many crystallized embodiments of dead and bygone thought, so many fossil relics, as it were, of the past strata of social growth. The facts of language—its sentences and its words—are the result of the action of general laws and conditions; by comparing and classifying them we can discover what these general laws are, and how they act. A knowledge of these laws and their action constitutes glottology or the science of language; the use of the comparative method by which they are discovered constitutes comparative philology.

Comparative philology, therefore, furnishes the materials whereby the science of language investigates such questions as the origin of speech, the nature of roots, or the meaning of flection. It may be said to comprise both comparative and historical grammar, comparative grammar being primarily occupied in comparing the grammatical forms and syntax of different languages of the same group; historical grammar in tracing the history of the forms and syntax of a single language. The two studies, however, necessarily overlap, comparative grammar requiring a knowledge of the individual languages compared at the successive periods of their history, or restoring the older forms of the individual languages by means of comparison, and historical grammar calling in the aid of the allied dialects to supply the deficiencies of the literary or monumental record. Quite apart from either is philology proper in the old sense of the word, which busied itself solely with literary languages and the literature they enshrine. The business of philology is to compare author with author, style with style, to determine the employment of words and phrases in the writers it investigates and pronounce upon their correctness, to emend the readings of MSS. and imitate the idiosyncrasies of particular writers. From the old-fashioned classical philology to the so-called philosophy of speech there is a wide leap, but both have been equally transformed by the new comparative method. The philosophy of speech in the hands of men like Harris or Stoddart[78] endeavoured to attack the problems of language by “the high priori road,” and by unverified and unverifiable reasoning from the phænomena of modern dialects to discover the origin of speech and the relation between grammar and logic. The philosophy of speech under the guidance of comparative philology has become the science of language, which may be said to comprehend both. The questions which the à priori method failed to resolve are now yielding their answers to à posteriori research, and the results already obtained have overthrown the unsubstantial speculations of the last century. The science of language has been variously termed “La Linguistique,”[79] “Linguistic Science,” Glottic,[80] and Glottology,[81] and it stands in the same relation to comparative philology that physiology stands to comparative anatomy.

Now, the ultimate facts with which comparative philology has to deal are sentences and the words that have been evolved out of them. These words and sentences must be real and not imaginary—that is, they must either belong to some living speech, or be preserved in a written record, or else be restored by a sound comparison of existing words which presuppose some common ancestor. Where such real and well-attested words are not to be had, no conclusions can be drawn. Unless inscribed monuments are hereafter brought to light or comparison with the Malayan dialects results in the recovery of a common parent-speech, the condition of the Polynesian languages 1,000 years ago must remain unknown. Much no doubt may be effected by comparing the scattered relics of these languages together, by showing that a sibilant, for instance, has been preserved in Samoan which has become a simple aspirate elsewhere, or that a guttural is retained between two vowels in Maori which has been dropped in most of the other Polynesian settlements; but to assert that some thousand years back they resembled another language to which they bear little similarity at present, would be to argue without data, and to violate the fundamental principles of comparative philology.

The object of the science of language is threefold:—

(1). It compares and classifies sentences, grammatical relations and words.

(2). It compares and classifies languages and dialects.

(3). By means of this comparison and classification it discovers the laws which govern language in general and certain languages and dialects in particular.

Thus by comparing the languages of the Aryan family we discover the phonetic law that an English th must always represent t in Sanskrit, Greek, and Latin, unless the action of other determinate laws interfere, and by comparing different groups of languages together, we find that the dual number everywhere preceded the plural. There are still many tongues in which the plural is formed by reduplication, tongues, that is, where duality, the repetition of the idea, is or has been the only conception of plurality yet reached; and others in which the number “three” is denoted by words like prica, “many” (in the dialect of the Puris of South America,) expressive of vague indefiniteness, and an inability to form a clear idea of anything beyond “two.” Indeed, in our own Aryan family of speech there was a time when one, and two, or that which was “divided” (δύω, δίς, διὰ, &c.) from one, were the only numerals known, and it required a fresh effort of thought to attain and conceive of a new numeral, which was accordingly named tri, tres, three, or that which is “beyond” (trans, through, Sansk., tar-â-mi, “I pass beyond”).

The laws of speech may be either primary or empirical. Empirical laws are those generalizations made from the survey of a limited number of phænomena, the reason of which we do not know. All we know is that given one particular fact, another particular fact follows, or that wherever we meet with a particular class of phænomena the same generalization is sure to hold good. Thus in astronomy, Kepler’s discovery that the planets move in an ellipse may be termed an empirical law, and the same may be said of the phonetic law mentioned above which obliges us to compare an English th with the Greek and Latin t. Primary laws are those higher and more comprehensive laws or generalizations which embrace the empirical laws and give the reason of them. Such a primary law is gravitation, such, too, probably is the law of natural selection. In the science of language examples of these primary laws would be the law that all language is based on roots, or the law of economy in the use of speech. The determination of the primary laws of language leads us very nearly into the charmed land of metaphysic; as the physicist with his doctrine of force is transported out of the region of pure experiment and observation, and brought face to face with metaphysical problems, so is the scientific student of language with his doctrine of roots. Hence that part of the science of language which stands in the most direct relation with the old philosophy of speech, which would investigate such subjects as the origin of gender and case, or determine the priority of thought or language, has sometimes been called linguistic metaphysics.

When once the laws of language have been laid down we are able to apply them to our facts (that is, words and sentences), to whatever period these belong. The science of language, like all other sciences, rests upon the postulate of uniformity. So long as the conditions remain the same, the laws of the science will act with undeviating regularity. It does not matter whether the words we are dealing with are still living and spoken, or have been dead and obsolete for thousands of years; if we can show that they fall under the action of a particular law, we can apply that law to them in either case with equal certainty. When once we have ascertained that an English d represents a Sanskrit t, only those Sanskrit words which contain a t must be compared with English words of Teutonic origin which have a d in the corresponding place, whatever their antiquity may be. A knowledge that an English d answers to a Sanskrit and Latin t, and an English h to a Sanskrit and Latin c (k or ś) shows that the English hundred has the same origin as the Latin centum, and the Sanskrit śatam, and that, consequently, our linguistic ancestors were able to count as far as one hundred before they separated from each other, the one to conquer India, the other to occupy Europe. Words, in fact, are like the fossils of the rocks; they embody the thought and knowledge of the society that first coined and used them, and if we can find out their primitive meaning by the aid of the comparative method, we shall know the character of the society that produced them, and the degree of civilization it had attained. The palæontologist can reconstruct the animal life of the past ages of the globe with no greater ease than the comparative philologist can reconstruct the life of bygone and forgotten communities. If the fragment of a fossil bone can tell us the history of an extinct world, so, too, can the fragment of a word reveal to us the struggles of ancient societies, and ideas and beliefs that have long since perished.

But the laws of a science must be verified before they can be accepted as such. However brilliant or ingenious a hypothesis may be, it remains a hypothesis, more or less probable, until it has been verified by experiment and observation. It is to history, to psychology, and to physiology that the science of language has to look for the verification of its laws. In the phonautograph of König, or the phonograph of Edison, we can discover the very forms assumed by the waves of air set in motion by each sound we utter; and the first lessons of psychology confirm the conclusion of glottology, that the concrete precedes the abstract. Sometimes it is not so much the law, the generalization itself, that can best be verified; but the application of it to the phænomena of speech. Thus, a sound application of the laws of language makes it clear that the words possessed in common by Spanish and Arabic are not due to a common ancestry, but to contact between the two tongues, and the history of the Moorish conquest of Spain confirms the conclusion.

But we may ask, What is meant precisely by that comparison of words and sentences on which the laws of language are said to rest? A word, a sentence, a grammatical form, consists of two elements, one, the articulate utterance, the other, the signification or thought which the utterance symbolizes. Sound and sense are the two factors which make up speech, and it is, therefore, in respect of both sound and sense that our comparisons have to be made. Comparative philology divides itself into phonology and sematology, to which, perhaps, we may also add morphology. Phonology is the science of sounds, sematology the science of meanings, and morphology the science of grammatical forms. But inasmuch as grammatical forms are but a combination of the relations of the sentence (or rather of the meaning those relations convey to the mind) and of the phonetic sounds by which they are expressed, morphology may be strictly included partly under phonology, partly under sematology. We must never forget that the study of sounds is intended to be the vestibule through which we approach the thought within. The phonological investigations we carry on, the phonological laws we formulate, are the outworks by which we may storm the fortress of the inward signification. They enable us to trace to a common source words that have flowed through diverse regions, or to discover the origin of some strangely-changed form of grammar, but the value they possess is the value that belonged to the magic ring of the Nibelungs: it gives access to the treasure, but is not the treasure itself. Phonology is not commensurate with comparative philology, as seems sometimes to be thought. It forms but one side of the science, the instrument by which we discover the true force and meaning of sentences and words.

As the instrument of linguistic science, however, phonology is of the highest importance. In fact the modern science of language is wholly based upon it, and that which distinguishes comparative philology from the abortive attempts of former centuries is its scientific investigation into the laws of articulate utterance and of phonetic change. Here, and almost here only, we can as yet trace the nature and working of the laws of speech. It is only because we know that an English h and d must answer to a Sanskrit k (ś) and t that we are able to assert that the primitive Aryan community had attained the conception of “one hundred.” Sematology is still in a far more backward state; its laws are still a subject of investigation, and the differences of opinion that exist as to some of the great questions of linguistic science show only too plainly how much in this department of it still remains to be done. But the relative position of phonology and sematology is, after all, but natural. Phonology deals with the outward and physical, that which, can be weighed and measured, and imitated by mechanical contrivances; sematology belongs to the inward and the spiritual—to that realm of thought, in short, which can only be examined in so far as it makes itself accessible to the inspection of the senses, and submits itself to the action of physical laws. Thought seems infinite, manifold, and free, determining and determined by itself. Like the wind, it “bloweth where it listeth;” we hear “the sound thereof, but cannot tell whence it cometh and whither it goeth.” All the capriciousness and complex mobility of the individual appears to belong to it; we may formulate the laws of thinking, but not of the forms which that thinking takes. The vocal organs, on the other hand, through which thought becomes realized in speech, are subject to all the conditions of the material world. The utterance of each articulate sound and its relations to another are conditioned and defined by the physical constitution of man, by the circumstances in which he finds himself, and by measurable laws of sound. The outward form of language, the flesh-garment, as it were, in which thought clothes itself, falls entirely into the domain of physiology and acoustics. Here we can observe and experimentalize, can weigh and measure, can even reproduce artificially for ourselves. Every consonant and vowel can be accurately determined, the machinery and effort needed to produce them precisely known, the variations they are capable of exactly ascertained. But when we turn to the informing thought, to that inner essence which gives life and reality to each modulation of articulate sound, all appears different. What wonder that the science of significations should be so far behind the science of sounds?

Let us not forget, however, that thought, in so far as it finds its expression in language, is not so infinitely free and capricious as we might at first sight suppose. The very fact of its finding expression in language, that is, of being embodied in articulate sounds, implies restraint and submission to conditions. Thought is thus, as it were, arrested and crystallized; it is only gradually and in consequence of ascertainable causes that the signification attached to a particular sound or group of sounds comes to be changed. That these sounds should symbolize certain ideas is, after all, a matter of convention; it follows from the tacit agreement, not indeed of isolated individuals, but of individuals as forming a society. Changes, therefore, in the signification of words and sentences can only result from causes which affect the whole society, and as such causes necessarily work slowly and by degrees, significant change can accordingly be brought under the action of general laws. But these laws can only be established by the help of phonology: until we know what words and forms the laws of phonology will allow us to compare together and refer to a common origin, we cannot begin to discuss the genesis and history of the significations they bear. No doubt structure, that is, the conception of the sentence formed by a language, and the order in which the several parts of a sentence are arranged, is a very important element in the classification of languages; still it is only one element, and unless phonology prove that the roots and derivatives of two idioms are related, no amount of structural similarity will justify us in deriving them from the same stock.

Phonology, then, is the key and mainstay of modern linguistic science; it guarantees the correctness of the results already obtained, and is the indispensable preliminary to future researches. As will be shown in a later chapter, our knowledge of sounds and their laws is now tolerably complete. So, too, is the application of this knowledge to certain groups of language. The phonological laws of the Aryan family, for instance, are pretty well ascertained; we know what sounds in one member of the family answer to other sounds in another member, and what particular changes of sound are permissible within each of the several members themselves. It follows from the physical formation of the organs of speech that the various sounds capable of being articulated are limited in number. Prince Lucien Bonaparte has enumerated as many as 385, though some of these are not to be met with in any known language or dialect.[82] The number of different sounds occurring in any single language is not large among European languages; for instance, Modern Greek, Spanish, and Illyrian have but five vowel-sounds, while Gaelic, which has the largest number, possesses twenty-one, Portuguese and English following next with nineteen a-piece. So far as consonantal sounds are concerned the number tends to diminish with the culture and age of a language, and the evidence of facts is against identifying the hypothetical alphabet to which the sounds of the various Aryan dialects can be reduced with the actual alphabet of the parent Aryan speech. The physical formation of our vocal organs, due to climate, food, habit, and inherited aptitudes, obliges us to pronounce in a particular way. There are sounds, for instance, which birds and animals can make, but we cannot; while nothing is harder than to catch and reproduce the exact pronunciation of a foreign tongue. The Polynesian turns David into Raviri, Samuel into Hemara, London into Ranana, and Frederick into Waratariki, and the word steel has been adopted in the Sandwich Islands in the shape of tila. It has been said that a foreigner can never speak another language so perfectly as to conceal all traces of his origin, and though this is going too far, it is quite certain that there are languages the pronunciation of which can never be thoroughly acquired after the age when growth has ended and the organs of speech have ceased to be plastic. There are numerous sounds which particular races or individuals are unable to imitate successfully; and those who have watched the attempt of children to learn their mother-tongue know how slowly some special sound is often acquired, and how in some cases it is never acquired at all. The sound which one person will pronounce as r will be pronounced l by another. Thus, the Chinese change every l into r, and the nearest approach they can make to the pronunciation of Christ is Ki-li-sse-t(ŭ). The Japanese, on the other hand, cannot manage l, and in their mouths accordingly idolatry becomes idoratry. The native children of Bengal, quick as they are in other respects, seldom pronounce rightly those English words which begin with a sibilant and a mute when a consonant precedes them, ten stamps, for instance, being made into ten-y-stamps, and this string into this-y-string. The same sound which is pronounced without difficulty in certain combinations may be a hopeless puzzle in others, and the English tourist who mispronounces Boulogne and Cologne, will yet ask for an onion and talk of a barrier. No individual, it would seem, pronounces all his sounds exactly like his neighbours, and even the same individual will vary his pronunciation of the same word in the course of a few seconds. Variations of pronunciation, in fact, are like the variations we observe in plants and animals, and if any variation becomes marked and is rendered popular and general from some cause or other, it brings about an alteration in the form of words. Such alterations resemble new species in natural history, and we may compare the different species of pigeons or dogs with the differences of pronunciations given by different dialects to what was originally the same sound. Changes in the pronunciation of words are constantly going on, causing a language to alter its form and appearance or to branch out into dialects. As these changes are determined by circumstances and physical necessities, and not by the arbitrary will of the individual, the laws they follow can be discovered and laid down. The laws once known, we can tell what words and sounds in different dialects, or in the different periods of the same dialect, may be compared together and referred to a common source, supposing, that is, that the significations they bear allow us to ascribe the identity of their phonetic elements to anything more than coincidence. The laws of phonology enable us to assert that the Greek καλός, and the English hale or (w)hole, may be traced back to a common origin so far as their outward crust and garment—the phonetic sounds of which they are composed—is concerned; it then remains for sematology to decide whether the ideas of “beauty” and “soundness” can be connected together. Distinctions between sounds must be studied in spoken languages, and we must not forget that it is always very difficult to discover what was the exact sound attached to a word no longer spoken, but preserved only in the custody of writing.

Different tribes and races vary much as to the sounds which they find it easy or hard to pronounce and imitate. A sound which has been changed into a certain other sound in one language, may have been preserved or changed into quite a different sound in another language. In our Aryan group the palatals were originally gutturals; in Malayan, on the contrary, dentals. Because our Teutonic forefathers turned k into h, we must not conclude that such a change was possible all over the world, and that wherever we come across an h we are at liberty to assume an earlier k. Indeed, there is clear evidence that in some languages h may become k. The phonetic laws which hold good of one group of languages, or of one member of a group, do not necessarily hold good of another.

In comparing languages we have first to compare their grammars, not their vocabularies. The reason of this is obvious. It is in the sentence, not in the isolated word, that languages agree or differ, and grammar deals with the relations that the several parts of the sentence bear to one another. Single words may accidentally resemble each other in both sound and sense, and yet belong to languages which have nothing in common. In the Quichua, or dialect of the Incas, three words—inti, “sun;” munay, “love;” and veypul, “great”—resemble the Sanskrit indra, manyu, and vipula,[83] but this is the only likeness that can be detected between the two tongues. So, too, the Mandshu shun, “the sun,” coincides in sound and meaning with the English word, like the Mandshu sengi and Latin sanguis, “blood,” or the North American Indian potómac and the Greek πόταμος, “river.” Such accidental coincidences turn up all the world over. The number of articulate sounds used in actual speech is, after all, not so very large, nor also the number of different ideas needed by primitive man; and when we bear in mind the probable onomatopœic origin of the greater part of our vocabulary, it is not wonderful that these coincidences should occur. Indeed, the wonder would be if they did not. But a coincidence of this sort is one of the surest evidences we can have that the words which seem to resemble one another have no connection whatsoever. As Professor Max Müller has said, “sound etymology has nothing to do with sound.” Language is continually changing; and as the phonetic and significant changes in it are occasioned by outward conditions and circumstances which vary from age to age and from country to country, they must necessarily take a different direction in the mouths of different speakers. The very fact that the English call and the Greek καλέω have almost every letter in common, ought to have raised a presumption against their identity, even before the law was known that an English c answers to a Greek γ, and a Greek κ to an English h, and that, consequently, the true Greek representative of call is γήρυω, and the true English representative of καλέω is hail.

But if we are not to compare words of the same sound and sense together, how, it may be asked, are we to ascertain the relationship of two or more languages, and discover what sounds correspond to each other in them? Our only guide is grammar. If we find that two languages express the relations of grammar in the same way, and by the help of the same machinery, we may conclude that the two languages come from a common source, and, therefore, possess a common stock of words. Under grammar will also be included structure—that is to say, the order and position of the parts of the sentence, as well as the conception of the sentence itself. Grammar and structure, therefore, are the clue by which comparative philology must be guided in its researches. It was the neglect of such a clue that caused Latin and Greek to be compared with Hebrew, and made the etymological dictionaries of the last century a rubbish heap of wasted labour. Those languages only which agree in their way of viewing the relations of thought can be grouped together. When once agreement in grammar and structure has determined the probable connection of two tongues, the aid of phonology may be called in to complete and verify the inquiry. Where the grammars are really connected, we may feel quite certain that there will be a community of roots. Where, on the contrary, there is no connection between the grammars, a community of roots must be due to accident. What proved the existence of an Aryan family of speech, and thereby founded comparative philology, was not the resemblances between individual words, striking as these were, but the exact correspondence between the grammatical forms of the several members of the family. The lists of words drawn up by Sir W. Jones, by Adelung or by Vater, remain mere literary curiosities. The comparative philology of Aryan speech was really created by the comparative grammar of Bopp. When once the grammatical relationship of the Indo-European languages had been established, there was a solid basis for phonology to work upon, and it was not long before Grimm discovered the laws which regulate their interchanges of sound.

But in comparing grammar and structure, we must be careful to exclude the accidental, or rather the phænomena due to the peculiar circumstances in which an individual tongue has been placed. We ought to be able to trace the history and development of each special language as far back as possible, ascertaining its oldest forms and noting the successive changes they have undergone. For this purpose it is necessary that the language should be a literary one, and that the various phases of its growth should have been preserved on monuments or in books. Where this is not the case, we have to fall back upon a simple comparison of existing dialects, and endeavour to restore from these the common forms to which their variant derivatives seem to point. The greater the number of dialects the more satisfactory will be the results of our comparison; accidental resemblances will be better eliminated, and intermediate forms are more likely to be preserved. Where the dialects to be compared are few, we have to contend against one of two difficulties—either the differences between them are so slight—as in the case of the Semitic languages—that the parent-speech from which they branched off must be too recent to throw any light on its earlier history and relationships; or else the differences are so great, the time during which they have been separated so considerable, that the links have been lost by which we may connect them together and reduce them to a single origin.

Phonology requires a knowledge of the past history and development of the languages it deals with even more than the study of grammar. In the comparison of words we may lay down the general rule that roots and not derivatives should be compared together. We should trace the history of the words we examine as far back as may be, should reduce them to their simplest forms, and strip off the accretions that have grown round them like the lichen round the stone. Words derived from the same radical will often assume different forms in different languages, or even in the same dialect; while words derived from different radicals will, on the other hand, assume the same form in different languages, or even in the same one. Captive and caitiff have the same origin in the Latin captivus; sound may be either the Latin sonus or subundare, or the Anglo-Saxon sund, “hale,” or sund from swimman. The American potómac, quoted above, is a compound, while the Greek πόταμος comes from the root πο-, which we find in πίνω and πότος, in the Sanskrit pânam, “a drink,” and our own potion. The lexicographers who have declared monkey to be a corruption of mannikin were little aware that the word is really the Italian monichio, the derivative of monna, and that monna, again, is a contraction of madonna, mea domina. Before we know the history of a word, we must not venture to compare it with another, though it may happen that the history will be learnt through the process of comparison itself. Thus we know that the Gothic fimf, “five,” has lost two gutturals, as well as a final labial, from the analogy of the Latin quinque (for quinquem), the Sanskrit panchan and the Lithuanian penki, and we can thus trace it back to the period when the Aryans of Europe and of Asia were still undivided. But at this point our materials fail us. We may feel pretty sure that quemquem, the original Aryan word for “five,” is a simple root, and that its numerical meaning is a derived one; we may even hazard the guess that it has been formed by reduplication, but beyond this a sound method of etymology cannot go. To connect it with the Semitic khâmésh, as Ewald has done, is to violate the rules of comparative philology. We know the history neither of khâmésh nor of quemquem.

In comparing words together, it is safest to begin with two classes of words, those which, like the numerals, have acquired a fixed and arbitrary meaning, and terms of relationship and every day use. In the case of the former, the signification, once fixed, remains unaltered, however much the phonetic crust of the word may change, while new names are less likely to come into vogue; in the case of the latter, the very frequency of their use tends to keep them in existence. If a few families here and there adopt new modes of expression, still it may be expected that the larger part of the community will be more conservative. Hence, when we find two languages agreeing in their numerals and words expressive of common objects and ideas, we may infer that they are related to one another. The pronouns are not so sure a criterion, as they have generally been worn down by constant use to monosyllabic forms, while their antiquity prevents us from discovering their true history and origin. Like the names of “father” and “mother,” moreover, the first and second personal pronouns show a tendency to be represented in most languages by the simplest and earliest sounds uttered by the child.

The laws of phonology must be established by as large a number of instances as possible. In no other way can the chances of accident or mistake be avoided. A law, in fact, must hold good of all the phænomena that are summed up under it, and the more numerous the phænomena, the wider and more firmly established will the law be. Grimm’s laws of the interchanges of sound in the Aryan family of speech depend on the observation and comparison of a very large number of words. As soon as it was found that English words which contained a th answered in signification and general form to Latin and Greek words which had a t in the same place, it was possible to formulate the law: English th = Latin and Greek t; all that remained was to verify the law by fresh instances, and in this way to strengthen the proof of the connection of the two languages. If it could be shown that real exceptions to the law occur which are not due to the interference of other laws, the law would have to be given up, however numerous might be the apparent instances on which it rested. The progress of comparative philology is continually strengthening its phonological laws and adding to their number.

The intimate connection of sound and sense must never be lost sight of in etymological research. They are as it were the outer and inner sides of the same object. Where the significations are unrelated, we cannot connect two words which agree in phonetic sound any more than we can connect two words of the same signification but different sound. In our own group of tongues the two separate roots dhā “to suck,” and dhā “to place,” for example, are identical in sound; and if we turn to languages like Chinese or Ancient Egyptian, we shall find numberless cases in which the same word, so far as pronunciation is concerned, has a variety of unallied meanings like our English box or scale. Of course, it is not necessary that the signification of the words we compare should be exactly the same; the signification of words changes as much as their outward phonetic form; but we must be able to show that one meaning is derived from the other, or from a common parentage, just as we show that one sound is derived from another or from a common source.

For the purposes of phonology more especially, the study of living spoken dialects is indispensable. No doubt the historical character of glottology requires us to investigate the records of extinct languages with as much care as the facts of living ones, and it is only by learning what a language once was that we can properly know what it is now. Nevertheless, it is only in the modern languages that we can discover the nature and laws of pronunciation; it is only here, moreover, that we are brought face to face with the problems and realities of speech. The biologist, it is true, cannot dispense with the aid of comparative anatomy, but his primary object is the study of the living organism. What has been termed “antiquarian philology” has sometimes stood in the way of scientific progress; sounds have been confounded with letters, and words instead of sentences have been made the units of speech. Antiquarian philology, furthermore, still has the shadow of classical scholasticism hanging over it; it will need a long education before the world is disabused of the idea that superiority in literature means superiority in language, and that a scientific study of language is identical with the old-fashioned “philology” of the classical scholar. Before the forms of an extinct speech can be made available for scientific investigation, they must be revivified by the translation of their written symbols into phonetic sounds, and how hard such a task is need not be pointed out. If we wish to work back to the former pronunciation of a language we must start from its modern and actual pronunciation, and in spite of all that we can do, in spite of slow and patient induction and a careful weighing of the facts, our conclusions will be at the best imperfect and approximative. The older and more scanty the remains of a language, the more defective and uncertain will be our restoration of its pronunciation. In the larger number of cases we have to be content with merely approximative results. What Mr. Ellis and Mr. Sweet have done for the pronunciation of early English, is due to the abundance of the data and the unbroken tradition which they embody; to restore the pronunciation of Latin is a work of greater difficulty, to restore that of ancient Greek of greater difficulty still. In short, the records of dead speech must be interpreted by the facts of living language, just as the conditions which brought about the deposition of the rocks can only be explained by the forces still at work upon the surface of the globe. Here as elsewhere in science, we must proceed from the known to the unknown. The laws of consonantal change laid down for Latin and Greek, for Sanskrit and Zend, for Keltic and Old High German, receive their verification and explanation from the Romance dialects of modern Europe; while it is in the study of savage idioms, in the languages of Bushmen and of Kafirs, of North American Indians and of Papuans, that some of the most precious facts of linguistic science have been obtained. An extinct literary language, indeed, is by its very nature less serviceable to the comparative philologist than the artless jargons of barbarous tribes. It is artificial rather than natural, and the product of individual idiosyncrasies rather than of the whole community. The further removed it is from the fresh current of living speech, the less capable it becomes of strictly scientific treatment. The individual element, with all its arbitrary capriciousness, has entered too largely into it. The grammatical forms invented and enforced by ignorant grammarians, the words coined after false analogy by the Homeric rhapsodists and their successors, or the stilted phrases and inverted expressions employed by a particular writer and his imitators, all belong to the domain of the “philologist” rather than to that of the scientific student of language. He has nothing to do with textual criticism or the study of style, much less with the successful reproduction of the idiosyncrasies of classical authors.

Philology in the narrower sense of the term has to prepare materials for comparative philology in so far as the latter is concerned with literary languages or dialects. In its turn it is guided in its researches and kept within the limits of scientific accuracy by comparative philology which tests and rectifies its conclusions, and prevents for the future attempts like that of Buttmann to derive ἄφνος from ἄφθονος or that of K. O. Müller to extract πελασγός from πελαργός. The particular can only be understood in the light of the universal, and as long as we are dealing with one language only our comparisons must be limited to that language alone at different stages of its growth, and will consequently sometimes lead us astray. Error can only be avoided by making our field of comparison as wide as possible, and so bringing our theory to the test of the greatest possible number of facts. It is evident from this, however, that the comparative philologist will have a special and minute acquaintance with but a few out of the many facts which come before his view. The memory even of a Mezzofanti is limited, and the ordinary student of language must be content to derive from others a large proportion of the materials on which he works. Caution in the choice and use of his authorities is here absolutely requisite, and it ought to be the business of the specialists in each language to see that the facts presented to him are thoroughly accurate and exact. Their work is the foundation upon which the structure of comparative philology has to be built.

But the comparative philologist cannot dispense with a specialist’s knowledge of at least two languages. In no other way can he have that intimate acquaintance with the inner life of speech requisite for his studies, or possess the necessary instinct for selecting the right authorities to whom to trust when dealing with tongues with which he is unacquainted. The more languages he knows thus thoroughly the better, especially if these belong to different classes of speech. Unless the Aryan scholar is acquainted with a Semitic language, his theory of flection is likely to be one-sided and faulty, and unless he have a further knowledge of some agglutinative dialect, his views on the relation between flection and agglutination must be received with a certain amount of distrust. Grammars and dictionaries will not give us that grasp upon the inner structure and spirit of a dialect which is all-important in determining some of the chief problems of speech. They present us only with the external facts of a language: before we can think in it, before we can place ourselves in the mental attitude of its framers and speakers, we must be saturated with it, as it were, and have that knowledge of it which can only come from daily and constant use.

At the same time, it must not be forgotten that the comparative philologist should not introduce the frame of mind of the specialist into his comparative inquiries. The specialist who takes up comparative philology as a subsidiary pursuit is likely to spoil it in the taking. The minor details of his special subject, whether it be Greek or Sanskrit or Hebrew, will assume an unreal importance in his eyes, and the main phænomena to which his attention ought to be directed will be correspondingly dwarfed. Bopp was the father of comparative philology simply because he was not a specialist in any one of the Aryan languages; had he been a Sanskritist, and nothing else, he would doubtless have produced an excellent Sanskrit grammar, but not the famous text-book of scientific philology. The errors into which he fell have since been corrected by the special students of the various languages he handled so freely: the knowledge he acquired of them was sufficient for the great purpose he had in view, and an exhaustive study of any one of them would merely have consumed the time and energy which were needed for his other work.

We can now see clearly what is the object and scope of the science of language. It has to do with language in all its forms as the significant utterance of society. Where utterance ceases to be significant, the science of language also ceases to investigate it. Beyond the barrier of roots it is unable to pass; other sciences—ethnology, psychology, physiology—must be called in if we wish to know what lies beyond that barrier, what, in short, were the inarticulate utterances and gestures which gave rise to articulate speech. Glottology has to investigate the origin of language so far as it is really language, but no further. By the use of the comparative method, words, forms, sentences, dialects, and languages are classified and traced back to their most primitive form, and the laws which govern their development and relationships determined and explained. In this work of comparison, phonology and sematology ought to go hand in hand, since language consists in the intimate union of sound and thought; but inasmuch as the facts and laws of phonology can be more readily discovered and tested than those of sematology, it is necessary that our linguistic researches should have their starting-point on the phonological side. Inasmuch as language is the reflection of the thought of a community, the history of words and forms, as determined by the application of the laws of glottology, will be also the mental and spiritual history of the community that used them. Like the geologist, therefore, who can reconstruct the material history of the earth and restore the various forms of life that have successively peopled it, the scientific student of language can read the past history of human society in the fossil-records of speech. By tracing the Greek δῆμος to the root δα, “to divide,” he can show that private property in Attica originated in that allotment of land by the commune which still prevails among the Slavs, while not only the existence but even the mode of life and intellectual horizon of the primitive Aryans has been revealed by comparative philology with more certainty and minuteness than could have been done by any chronicle, however perfect. But perhaps the most important of the results obtained by the application of the comparative method to language, has been the light thrown upon the origin and nature of mythology and the history of religion. Two new sciences, those of comparative mythology and comparative religion, have grown up under the shelter of glottology, and form subordinate sciences dependent upon it. In the more immediately practical sphere of education, again, the science of language has lightened the labours of the learner by explaining the reason of the rule while it insists upon the reversal of the old unscientific mode of teaching languages by beginning with the dead ones, and points out that the method of science and of nature alike is to proceed from the known to the unknown. By breaking down the prejudices that have so long maintained our present cumbrous and inaccurate spelling, it is preparing the way for a reform in that direction, with its consequent saving of time and labour, while the construction of an universal language is the aim towards which its students ultimately look.

But meanwhile, though much has been accomplished, much more still remains to be done. Comparative philology and the science of language are not yet a century old, and the problems of speech that still await solution are many and important. The previous chapter will have shown how various are the opinions still held as to the nature of language and its science, while the belief that the exceptional—we might almost say abnormal—Aryan family of speech is the type and rule of all others still unconsciously influences a large amount of philological reasoning. Is the science of language a physical or a historical one? Did roots constitute a spoken language or are they phonetic types which never entered into actual speech? Have isolating languages become agglutinative and agglutinative languages inflectional? Do dialects precede the common language or does the common language precede dialects? Have the languages of the world been all derived from one or two primitive centres or do they point to an infinite diversity of origin? Such are some of the questions which still await an answer, and the answer requires more investigation, more patient observation and induction, and, above all, more labourers in the field of research.