Language: Its Nature, Development and Origin

XXI.—§ 1. Introduction.

Much of what is contained in the last chapters is preparatory to the theme which is to occupy us in this chapter, the ultimate origin of human speech. We have already seen the feeling with which this subject has often been regarded by eminent linguists, the feeling which led to an absolute taboo of the question in the French Société de linguistique (p. 96). One may here quote Whitney: “No theme in linguistic science is more often and more voluminously treated than this, and by scholars of every grade and tendency; nor any, it may be added, with less profitable result in proportion to the labour expended; the greater part of what is said and written upon it is mere windy talk, the assertion of subjective views which commend themselves to no mind save the one that produces them, and which are apt to be offered with a confidence, and defended with a tenacity, that are in inverse ratio to their acceptableness. This has given the whole question a bad repute among sober-minded philologists” (OLS 1. 279).

Nevertheless, linguistic science cannot refrain for ever from asking about the whence (and about the whither) of linguistic evolution. And here we must first of all realize that man is not the only animal that has a ‘language,’ though at present we know very little about the real nature and expressiveness of the languages of birds and mammals or of the signalling system of ants, etc. The speech of some animals may be more like our language than most people are willing to admit—it may also in some respects be even more perfect than human language precisely because it is unlike it and has developed along lines about which we can know nothing; but it is of little avail to speculate on these matters. What is certain is that no race of mankind is without a language which in everything essential is identical in character with our own, and that there are a certain number of circumstances which have been of signal importance in assisting mankind in developing language (cf. Gabelentz Spr 294 ff.).

First of all, man has an upright gait; this gives him two limbs more than the dog has, for instance: he can carry things and yet jabber on; he is not reduced to defending himself by biting, but can use his mouth for other purposes. Feeding also takes less time in his case than in that of the cow, who has little time for anything else than chewing and a moo now and then. The sexual life of man is not restricted to one particular time of the year, the two sexes remain together the whole year round, and thus sociability is promoted; the helplessness of babies works in the same direction through necessitating a more continuous family life, in which there is also time enough for all kinds of sports, including play with the vocal organs. Thus conditions have been generally favourable for the development of singing and talking, but the problem is, how could sounds and ideas come to be connected as they are in language?

What method or methods have we for the solution of this question? With very few exceptions those who have written about our subject have conjured up in their imagination a primitive era, and then asked themselves: How would it be possible for men or manlike beings, hitherto unfurnished with speech, to acquire speech as a means of communication of thought? Not only is this method followed, so to speak, instinctively by investigators, but we are even positively told (by Marty) that it is the only method possible. In direct opposition to this assertion, I think that it is chiefly and principally due to this method and to this way of putting the question that so little has yet been done to solve it. If we are to have any hope of success in our investigation we must try new methods and new ways—and fortunately there are ways which lead us to a point from which we may expect to see the world of primitive language revealed to us in a new light. But let us first cast a rapid glance at those theories which have been advanced by followers of the speculative or a priori method.

XXI.—§ 2. Former Theories.

One theory is that primitive words were imitative of sounds: man copied the barking of dogs and thereby obtained a natural word with the meaning of ‘dog’ or ‘bark.’ To this theory, nicknamed the bow-wow theory, Renan objects that it seems rather absurd to set up this chronological sequence: first the lower animals are original enough to cry and roar; and then comes man, making a language for himself by imitating his inferiors. But surely man would imitate not only the cries of inferior animals, but also those of his fellow-men, and the salient point of the theory is this: sounds which in one creature were produced without any meaning, but which were characteristic of that creature, could by man be used to designate the creature itself (or the movement or action productive of the sound). In this way an originally unmeaning sound could in the mouth of an imitator and in the mind of someone hearing that imitation acquire a real meaning. In the chapter on Sound Symbolism I have tried to show how from the rudest and most direct imitations of this kind we may arrive through many gradations at some of the subtlest effects of human speech, and how imitation, in the widest sense we can give to this word—a wider sense than most advocates of the theory seem able to imagine—is so far from belonging exclusively to a primitive age that it is not extinct even yet. There is not much of value in Max Müller’s remark that “the onomatopœic theory goes very smoothly as long as it deals with cackling hens and quacking ducks; but round that poultry-yard there is a high wall, and we soon find that it is behind that wall that language really begins” (Life 2. 97), or in his other remark that “words of this kind (cuckoo) are, like artificial flowers, without a root. They are sterile, and unfit to express anything beyond the one object which they imitate” (ib. 1. 410). But cuckoo may become cuckold (Fr. cocu), and from cock are derived the names Müller himself mentions, Fr. coquet, coquetterie, cocart, cocarde, coquelicot.... Echoic words may be just as fertile as any other part of the vocabulary.

Another theory is the interjectional, nicknamed the pooh-pooh, theory: language is derived from instinctive ejaculations called forth by pain or other intense sensations or feelings. The adherents of this theory generally take these interjections for granted, without asking about the way in which they have come into existence. Darwin, however, in The Expression of the Emotions, gives purely physiological reasons for some interjections, as when the feeling of contempt or disgust is accompanied by a tendency “to blow out of the mouth or nostrils, and this produces sounds like pooh or pish.” Again, “when anyone is startled or suddenly astonished, there is an instantaneous tendency, likewise from an intelligible cause, namely, to be ready for prolonged exertion, to open the mouth widely, so as to draw a deep and rapid inspiration. When the next full expiration follows, the mouth is slightly closed, and the lips, from causes hereafter to be discussed, are somewhat protruded; and this form of the mouth, if the voice be at all exerted, produces ... the sound of the vowel o. Certainly a deep sound of a prolonged Oh! may be heard from a whole crowd of people immediately after witnessing any astonishing spectacle. If, together with surprise, pain be felt, there is a tendency to contract all the muscles of the body, including those of the face, and the lips will then be drawn back; and this will perhaps account for the sound becoming higher and assuming the character of Ah! or Ach!”

To the ordinary interjectional theory it may be objected that the usual interjections are abrupt expressions for sudden sensations and emotions; they are therefore isolated in relation to the speech material used in the rest of the language. “Between interjection and word there is a chasm wide enough to allow us to say that the interjection is the negation of language, for interjections are employed only when one either cannot or will not speak” (Benfey Gesch 295). This ‘chasm’ is also shown phonetically by the fact that the most spontaneous interjections often contain sounds which are not used in language proper, voiceless vowels, inspiratory sounds, clicks, etc., whence the impossibility properly to represent them by means of our ordinary alphabet: the spellings pooh, pish, whew, tut are very poor renderings indeed of the natural sounds. On the other hand, many interjections are now more or less conventionalized and are learnt like any other words, consequently with a different form in different languages: in pain a German and a Seelander will exclaim au, a Jutlander aus, a Frenchman ahi and an Englishman oh, or perhaps ow. Kipling writes in one of his stories: “That man is no Afghan, for they weep ‘Ai! Ai!’ Nor is he of Hindustan, for they weep ‘Oh! Ho!’ He weeps after the fashion of the white men, who say, ‘Ow! Ow!’”

A closely related theory is the nativistic, nicknamed the ding-dong, theory, according to which there is a mystic harmony between sound and sense: “There is a law which runs through nearly the whole of nature, that everything which is struck rings. Each substance has its peculiar ring.” Language is the result of an instinct, a “faculty peculiar to man in his primitive state, by which every impression from without received its vocal expression from within”—a faculty which “became extinct when its object was fulfilled.” This theory, which Max Müller propounded and afterwards wisely abandoned, is mentioned here for the curiosity of the matter only.

Noiré started a fourth theory, nicknamed the yo-he-ho: under any strong muscular effort it is a relief to the system to let breath come out strongly and repeatedly, and by that process to let the vocal chords vibrate in different ways; when primitive acts were performed in common, they would, therefore, naturally be accompanied with some sounds which would come to be associated with the idea of the act performed and stand as a name for it; the first words would accordingly mean something like ‘heave’ or ‘haul.’

Now, these theories, here imperfectly reproduced each in a few lines, are mutually antagonistic: thus Noiré thinks it possible to explain the origin of speech without sound imitation. And yet what should prevent our combining these several theories and using them concurrently? It would seem to matter very little whether the first word uttered by man was bow-wow or pooh-pooh, for the fact remains that he said both one and the other. Each of the three chief theories enables one to explain parts of language, but still only parts, and not even the most important parts—the main body of language seems hardly to be touched by any of them. Again, with the exception of Noiré’s theory, they are too individualistic and take too little account of language as a means of human intercourse. Moreover, they all tacitly assume that up to the creation of language man had remained mute or silent; but this is most improbable from a physiological point of view. As a rule we do not find an organ already perfected on the first occasion of its use; it is only by use that an organ is developed.

XXI.—§ 3. Method.

So much for the results of the first method of approaching the question of the origin of speech, that of trying to picture to oneself a speechless mankind and speculating on the way in which language could then have originated. We shall now, as hinted above (p. 413), indicate the ways in which it is possible to supplement, and even in some measure to supplant, this speculative or deductive method by means of inductive reasonings. These can be based on three fields of investigation, namely:

Of these, the third is the most fruitful source of information.

First, as to the language of children. Some biologists maintain that the development of the individual follows on the whole the same course as that of the race; the embryo, before it arrives at full maturity, will have passed through the same stages of development which in countless generations have led the whole species to its present level. It has, therefore, occurred to many that the acquisition by mankind at large of the faculty of speech may be mirrored to us in the process by which any child learns to communicate its thoughts by means of its vocal organs. Accordingly, children’s language has often been invoked to furnish illustrations and parallels of the process gone through in the formation of primitive language. But many writers have been guilty of an erroneous inference in applying this principle, inasmuch as they have taken all their examples from a child’s acquisition of an already existing language. The fallacy will be evident if we suppose for a moment the case of a man endeavouring to arrive at the evolution of music from the manner in which a child is nowadays taught to play on the piano. Manifestly, the modern learner is in quite a different position to primitive man, and has quite a different task set him: he has an instrument ready to hand, and melodies already composed for him, and finally a teacher who understands how to draw these tunes forth from the instrument. It is the same thing with language: the task of the child is to learn an existing language, that is, to connect certain sounds heard on the lips of others with the same ideas that the speakers associate with them, but not in the least to frame anything new. No; if we are seeking some parallel to the primitive acquisition of language, we must look elsewhere and turn to baby language as it is spoken in the first year of life, before the child has begun to ‘notice’ and to make out what use is made of language by grown-up people. Here, in the child’s first purposeless murmuring, crowing and babbling, we have real nature sounds; here we may expect to find some clue to the infancy of the language of the race. And, again, we must not neglect the way children have of creating new words never heard before, and often of attaching a sense to originally meaningless conglomerations of sound.

As for the languages of contemporary savages, we may in some instances take them as typical of more primitive languages than those of civilized nations, and therefore as illustrating a linguistic stage that is nearer to that in which speech originated. Still, inferences from such languages should be used with great caution, for it should never be forgotten that even the most backward race has many centuries of linguistic evolution behind it, and that the conditions therefore may, or must, be very different from those of primeval man. The so-called primitive languages will therefore in the following sections be only invoked to corroborate conclusions at which it is possible to arrive from other data.

The third and most fruitful source from which to gather information of value for our investigation is the history of language as it has been considered in previous chapters of this work. While the propounders of the theories of the origin of speech mentioned above made straight for the front of the lion’s den, we are like the fox in the fable, who noticed that all the traces led into the den and not a single one came out; we will therefore try and steal into the den from behind. They thought it logically correct, nay necessary, to begin at the beginning; let us, for variety’s sake, begin with languages accessible at the present day, and let us attempt from that starting-point step by step to trace the backward path. Perhaps in this way we may reach the very first beginnings of speech.

The method I recommend, and which I think I am the first to employ consistently, is to trace our modern twentieth-century languages as far back in time as history and our materials will allow us; and then, from this comparison of present English with Old English, of Danish with Old Norse, and of both with ‘Common Gothonic,’ of French and Italian with Latin, of modern Indian dialects with Sanskrit, etc., to deduce definite laws for the development of languages in general, and to try and find a system of lines which can be lengthened backwards beyond the reach of history. If we should succeed in discovering certain qualities to be generally typical of the earlier as opposed to the later stages of languages, we shall be justified in concluding that the same qualities obtained in a still higher degree in the earliest times of all; if we are able within the historical era to demonstrate a definite direction of linguistic evolution, we must be allowed to infer that the direction was the same even in those primeval periods for which we have no documents to guide us. But if the change witnessed in the evolution of modern speech out of older forms of speech is thus on a larger scale projected back into the childhood of mankind, and if by this process we arrive finally at uttered sounds of such a description that they can no longer be called a real language, but something antecedent to language—why, then the problem will have been solved; for transformation is something we can understand, while a creation out of nothing can never be comprehended by human understanding.

This, then, will be the object of the following rapid sketch: to search the several departments of the science of language for general laws of evolution—most of them have already been discussed at some length in the preceding chapters—then to magnify the changes observed, and thus to form a picture of the outer and inner structure of some sort of speech more primitive than the most primitive language accessible to direct observation.

XXI.—§ 4. Sounds.

First, as regards the purely phonetic side of language, we observe everywhere the tendency to make pronunciation more easy, so as to lessen the muscular effort; difficult combinations of sounds are discarded, those only being retained which are pronounced with ease (see Ch. XIV § 6 ff.). Modern research has shown that the Proto-Aryan sound-system was much more complicated than was imagined in the reconstructions of the middle of the nineteenth century. In most languages now only such sounds are used as are produced by expiration, while inbreathed sounds and clicks or suction-stops are not found in connected speech. In civilized languages we meet with such sounds only in interjections, as when an inbreathed voiceless l (generally with rhythmic variations of strength and corresponding small movements of the tongue) is used to express delight in eating and drinking, or when the click inadequately spelt tut is used to express impatience. In some very primitive South African languages, on the other hand, clicks are found as integral parts of words; and Bleek has rendered it probable that in former stages of these languages they were in more extensive use than now. We may perhaps draw the conclusion that primitive languages in general were rich in all kinds of difficult sounds.

The following point is of more far-reaching consequence. In some languages we find a gradual disappearance of tone or pitch accent; this has been the case in Danish, whereas Norwegian and Swedish have kept the old tones; so also in Russian as compared with Serbo-Croatian. In the works of old Indian, Greek and Latin grammarians we have express statements to the effect that pitch accent played a prominent part in those languages, and that the intervals used must have been comparatively greater than is usual in our modern languages. In modern Greek and in the Romanic languages the tone element has been obscured, and now ‘stress’ is heard on the syllable where the ancients noted only a high or a low tone. About the languages spoken nowadays by savage tribes we have generally very little information, as most of those who have made a first-hand study of such languages have not been trained to observe and to describe these delicate points; still, there is of late years an increasing number of observations of tone accents, for instance in African languages, which may justify us in thinking that tone plays an important part in many primitive languages.[107]

So much for word tones; now for the sentence melody. It is a well-known fact that the modulation of sentences is strongly influenced by the effect of intense emotions in causing stronger and more rapid raisings and sinkings of the tone. “All passionate language does of itself become musical—with a finer music than the mere accent; the speech of a man even in zealous anger becomes a chant, a song” (Carlyle). “The sounds of common conversation have but little resonance; those of strong feeling have much more. Under rising ill-temper the voice acquires a metallic ring.... Grief, unburdening itself, uses tones approaching in timbre to those of chanting; and in his most pathetic passages an eloquent speaker similarly falls into tones more vibratory than those common to him.... While calm speech is comparatively monotonous, emotion makes use of fifths, octaves, and even wider intervals” (H. Spencer).

Now, it is a consequence of advancing civilization that passion, or, at least, the expression of passion, is moderated, and we must therefore conclude that the speech of uncivilized and primitive men was more passionately agitated than ours, more like music or song. This conclusion is borne out by what we hear about the speech of many savages in our own days. European travellers very often record their impression of the speech of different tribes in expressions like these: “pronouncing whatever they spoke in a very singing manner,” “the singing tone of voice, in common conversation, was frequent,” “the speech is very much modulated and resembles singing,” “highly artificial and musical,” etc.

These facts and considerations all point to the conclusion that there once was a time when all speech was song, or rather when these two actions were not yet differentiated; but perhaps this inference cannot be established inductively at the present stage of linguistic science with the same amount of certainty as the statements I am now going to make as to the nature of primitive speech.

As we have seen above (Ch. XVII § 7), a great many of the changes going on regularly from century to century, as well as some of the sudden changes which take place now and then in the history of each language, result in the shortening of words. This is seen everywhere and at all times, and in consequence of this universal tendency we find that the ancient languages of our family, Sanskrit, Zend, etc., abound in very long words; the further back we go, the greater the number of sesquipedalia. We have seen also how the current theory, according to which every language started with monosyllabic roots, fails at every point to account for actual facts and breaks down before the established truths of linguistic history. Just as the history of religion does not pass from the belief in one god to the belief in many gods, but inversely from polytheism towards monotheism, so language proceeds from original polysyllabism towards monosyllabism: if the development of language took the same course in prehistoric as in historic times, we see, by projecting the teaching of history on a larger scale back into the darkest ages, that early words must have been to present ones what the plesiosaurus and gigantosaurus are to present-day reptiles. The outcome of this phonetic section is, therefore, that we must imagine primitive language as consisting (chiefly at least) of very long words, full of difficult sounds, and sung rather than spoken.

XXI.—§ 5. Grammar.

Can anything be stated about the grammar of primitive languages? Yes, I think so, if we continue backwards into the past the lines of evolution resulting from the investigations of previous chapters of this volume. Ancient languages have more forms than modern ones; forms originally kept distinct are in course of time confused, either phonetically or analogically, alike in substantives, adjectives and verbs.

A characteristic feature of the structure of languages in their early stages is that each form of a word (whether verb or noun) contains in itself several minor modifications which, in the later stages, are expressed separately (if at all), that is, by means of auxiliary verbs or prepositions. Such a word as Latin cantavisset unites in one inseparable whole the equivalents of six ideas: (1) ‘sing,’ (2) pluperfect, (3) that indefinite modification of the verbal idea which we term subjunctive, (4) active, (5) third person, and (6) singular. The tendency of later stages is towards expressing such modifications analytically; but if we accept the terms ‘synthesis’ and ‘analysis’ for ancient and recent stages, we must first realize that there exist many gradations of both: in no single language do we find either synthesis or analysis carried out with absolute purity and consistency. Everywhere we find a more or less. Latin is synthetic in comparison with French, French analytic in comparison with Latin; but if we were able to see the direct ancestor of Latin, say two thousand years before the earliest inscriptions, we should no doubt find a language so synthetic that in comparison with it Cicero’s would have to be termed highly analytic.

Secondly, we must not from the term ‘synthesis,’ which etymologically means ‘composition’ or ‘putting together,’ draw the conclusion that synthetic forms, such as we find, for instance, in Latin, consist of originally independent elements put together and thus in their turn presuppose a previous stage of analysis. Whoever does not share the usual opinion that all flexional forms have originated through coalescence of separate words, but sees as we have seen (in Ch. XIX) also the reverse process of inseparable portions of words gaining greater and greater independence, will perhaps do well to look out for a better and less ambiguous word than synthesis to describe the character of primitive speech. What in the later stages of languages is analyzed or dissolved, in the earlier stages was unanalyzable or indissoluble; ‘entangled’ or ‘complicated’ would therefore be better renderings of our impression of the first state of things.

XXI.—§ 6. Units.

But are the old forms really less dissoluble than their modern equivalents? This is repeatedly denied even by recent writers, on whom my words in Progress, p. 117, cannot have made much impression, if they have read them at all; and it will therefore be necessary to take up this cardinal point. Let me begin with quoting what others have said. “Historically considered, the Latin amat is really two words, as much as its English representative, the final t being originally a pronoun signifying ‘he,’ ‘she’ or ‘it,’ and it is only reasons of practical convenience that prevent us from writing am at or ama t as two and heloves as one word.... The really essential difference between amat and he loves is that in the former the pronominal element is expressed by a suffix, in the latter by a prefix” (Sweet PS 274, 1899). “It is purely accidental that the Latin form is not written am-av-it. To the unsophisticated Frenchman il a aimé is neither less nor more one unit than amavit to a Roman.... When the locution il a aimé sprang up, each element of it was still to some extent felt separately; but after it had become a fixed formula the elements were fused together into one whole. As a matter of fact, uneducated French people have not the least idea whether it is one or three words they speak” (Sütterlin WGS 11, 1902). “In some modern languages the personal pronoun is, just as in archaic Greek, beginning to be amalgamated with verbs so as to become a mere termination (sic: désinence; prefix must be what is meant): Fr. j’don’, tu-don’, il-don’ (je donne, tu donnes, il donne) and E. i-giv’, we giv’, you-giv’, they-giv’, correspond exactly to Gr. dido-mi, dido-si, dido-ti, only that the personal particle is in a different place” (Dauzat V 155, 1910). “If French were a savage language not yet reduced to writing, a travelling linguist, hearing the present tense of the verb aimer pronounced by the natives, would transcribe it in the following way: jèm, tu èm, ilèm, nouzémon, vouzémé, ilzèm. He would be struck particularly with the agglutination of the pronominal subject and the verb, and would never feel tempted to draw up a paradigm without pronouns: aime, aimes, aime, aimons, etc., in which traditional spelling makes us believe.... He would even, through a comparison of ilèm and ilzèm, be led to establish a tendency to incorporation, as the only sign of the plural is a z infixed in the verbal complex” (Bally LV 43, 1913).

In these utterances two questions are really mixed together, that of the origin of Aryan flexional forms and that of the actual status of some forms in various languages. As to the former question, we have seen (p. 383) how very uncertain it is that amat and didosi, etc., contain pronouns. As to the latter question, it is quite true that we should not let the usual spelling be decisive when it is asked whether we have one or two or three words; but all these writers strangely overlook the really important criteria which we possess in this matter. Bally’s traveller could only have arrived at his result by listening to grammar lessons in which the three persons of the verb were rattled off one after the other, for if he had taken his forms from actual conversation he would have come across numerous instances in which the forms occurred without pronouns, first in the imperative, aime, aimons, aimez, then in collocations like celui qui aime, ceux qui aiment, in which there is no infix to denote the plural; in le mari aime, les maris aiment, and innumerable similar groups there is neither pronoun nor infix. If he were at first inclined to take ilaaimé as one word, he would on further acquaintance with the language discover that the elements were often separated: il n’a pas aimé, il nous a toujours aimés, etc. Similarly with the English forms adduced: I never give, you always give. This is the crucial point: the French and English combinations are two (three) words because the elements are not always placed together; Lat. amat, amavit, are each of them only one word because they can never be divided, and in the same way we never find anything placed between am and o in the first person, amo. These forms are as inseparable as E. loves, but E. heloves is separable because both he and loves can stand alone, and can also, in certain combinations, though now rarely, be transposed: loves he. Some writers would compare French combinations like il te le disait with verbal forms in certain Amerindian languages, in which subject and direct and indirect object are alike ‘incorporated’ in a ‘polysynthetic’ verbal form; it is quite true that these French pronominal forms can never be used by themselves, but only in conjunction with a verb; still, the French pronouns are more independent of each other than the elements of some other more primitive languages. In the first place, this is shown by the possibility of varying the pronunciation: il te le disait may be either [itlədizɛ] or [itəldizɛ] or even more solemnly [iltələdizɛ]; secondly, by the regularity of these joined pronominal forms, for they are always the same, whatever the verb may be; and lastly, by their changing places in certain cases: te le disait-il? dis-le-lui, etc.

Nor can it be said that English forms like he’s = he is (or he has), I’d = I had (or I would), he’ll = he will show a tendency towards ‘entangling,’ for however closely together these forms are generally pronounced, each of them must be said to consist of two words, as is shown by the possibility of transposition (Is he ill?) and of intercalation of other words (I never had); it is also noteworthy that the same short forms of the verbs can be added to all kinds of words (the water’ll be ..., the sea’d been calm). In the forms don’t, won’t, can’t there is something like amalgamation of the verbal with the negative idea. Still, it is important to notice that the amalgamation only takes place with a few verbs of the auxiliary class. In saying ‘I don’t write’ the full verb is not touched by the fusion, and is even allowed to be unchanged in cases where it would have been inflected if no auxiliary had been used; compare I write, he writes, I wrote with the negative I don’t write, he doesn’t write, I didn’t write. It will be seen, especially if we take into account the colloquial or vulgar form for the third person, he don’t write, that the general movement here as elsewhere is really rather in the direction of ‘isolation’ than of fusion; for the verbal form write is stripped of all signs of person and tense, the person being indicated separately (if at all), and the tense sign being joined to the negation. So also in interrogative sentences; and if that tendency which can be observed in Elizabethan English had prevailed by using the combination I do write in positive statements, even where no special emphasis is intended, English verbs (except a few auxiliaries) would have been entirely stripped of those elements which to most grammarians constitute the very essence of a verb, namely, the marks of person, number, tense and mood, write being the universal form, besides the quasi-nominal forms writing and written.

Now, it is often said that the history of language shows a sort of gyration or movement in spirals, in which synthesis is followed by analysis, this by a new synthesis (flexion), and this again by analysis, and so forth. Latin amabo (which according to the old theory was once ama + some auxiliary) has been succeeded by amare habeo, which in its turn is fused into amerò, aimerai, and the latter form is now to some extent giving way to je vais aimer. But this pretended law of rotation is only arrived at by considering a comparatively small number of phenomena, and not by viewing the successive stages of the same language as wholes and drawing general inferences as to their typically distinctive characters (cf. above, p. 337). If for every two instances of new flexions springing up we see ten older ones discarded in favour of analysis or isolation, are we not entitled to the generalization that flexion or indissolubility tends to give way to analysis? We should beware of being under the same delusion as a man who, in walking over a mountainous country, thinks that he goes down just as many and just as long hills as he goes up, while on the contrary each ascent is higher than the preceding descent, so that finally he finds himself unexpectedly many thousand feet above the level from which he started.

The direction of movement is towards flexionless languages (such as Chinese, or to a certain extent Modern English) with freely combinable elements; the starting-point was flexional languages (such as Latin or Greek); at a still earlier stage we must suppose a language in which a verbal form might indicate not only six things, like cantavisset, but a still larger number, in which verbs were perhaps modified according to the gender (or sex) of the subject, as they are in Semitic languages, or according to the object, as in some Amerindian languages, or according to whether a man, a woman, or a person who commands respect is spoken to, as in Basque. But that amounts to the same thing as saying that the border-line between word and sentence was not so clearly defined as in more recent times; cantavisset is really nothing but a sentence-word, and the same holds good to a still greater extent of the sound conglomerations of Eskimo and some other North American languages. Primitive linguistic units must have been much more complicated in point of meaning, as well as much longer in point of sound, than those with which we are most familiar.

XXI.—§ 7. Irregularities.

Another point of great importance is this: in early languages we find a far greater number of irregularities, exceptions, anomalies, than in modern ones. It is true that we not unfrequently see new irregularities spring up, where the formations were formerly regular; but these instances are very far from counterbalancing the opposite class, in which words once irregularly inflected become regular, or are given up in favour of regularly inflected words, or in which anomalies in syntax are levelled. The tendency is more and more to denote the same thing by the same means in every case, to extend the ending, or whatever it is, that is used in a large class of words to express a certain modification of the central idea, until it is used in all other words as well.

Comparative linguistics did not attain a scientific character till the principle was established that the relationship of two languages had to be determined by a thoroughgoing conformity in the most necessary parts of language, namely (besides grammar proper) pronouns and numerals and the most indispensable of nouns and verbs. But if this domain of speech, by preserving religiously, as it were, the old tradition, affords infallible criteria of the near or remote relationship of different languages, may we not reasonably expect to find in the same domain some clue to the oldest grammatical system used by our ancestors? What sort of system, then, do we find there? We see such a declension as I, me, we, us: the several forms of the ‘paradigm’ do not at all resemble each other, as they do in more recently developed declensions. We find masculines and feminines, such as father, mother, man, wife, bull, cow; while such methods of derivation as are seen in count, countess, he-bear, she-bear, belong to a later time. We meet with degrees of comparison like good, better, ill, worse, while regular forms like happy, happier, big, bigger, prevail in all the younger strata of languages. We meet with verbal flexion such as appears in am, is, was, been, which forms a striking contrast to the more modern method of adding a mere ending while leaving the body of the word unchanged. In an interesting book, Vom Suppletivwesen der indogermanischen Sprachen (1899), H. Osthoff has collected a very great number of examples from the old Aryan languages of different stems supplementing each other, and has pointed out that this phenomenon is characteristic of the most necessary ideas occurring every moment in ordinary conversation: I take at random a few of the best-known of his examples: Fr. aller, je vais, j’irai, Lat. fero, tuli, Gr. horaō, opsomai, eidon, Lat. bonus, melior, optimus. Osthoff fully agrees with me that we have here a trait of primitive psychology: our remote ancestors were not able to see and to express what was common to these ideas; their minds were very unsystematic, and separated in their linguistic expressions things which from a logical point of view are closely related: much of their grammar, therefore, was really of a lexical character.

XXI.—§ 8. Savage Tribes.

If now it is asked whether the conclusions we have thus arrived at are borne out by a consideration of the languages of savage or primitive races nowadays, the answer is that these cannot be lumped together; there are among them many different types, even with regard to grammatical structure. But the more these languages are studied and the more accurately their structure is described, the more also students perceive intricacies and anomalies in their grammar. Gabelentz (Spr 386) says that the casual observer has no idea how manifold and how nicely circumscribed grammatical categories can be, even in the seemingly crudest languages, for ordinary grammars tell us nothing about that. P. W. Schmidt (Die Stellung der Pygmäenvölker, 1910, 129) says that whoever, from the low culture of the Andamanese, would expect to find their language very simple and poor in expressions would be strangely deceived, for its mechanism is highly complicated, with many prefixes and suffixes, which often conceal the root itself. Meinhof (MSA 136) mentions the multiplicity of plural formations in African languages. Vilhelm Thomsen, in speaking of the Santhal (Khervarian) language, says that its grammar is capable of expressing a multiplicity of nuances which in other languages must be expressed by clumsy circumlocutions; the native speakers go beyond what is necessary through requiring expressions for many subordinate notions, the language having, so to speak, only one fine gold-balance, on which everything, even the simplest and commonest things, must be weighed by the adding-up of a whole series of minutiæ. Curr speaks about the erroneous belief in the simplicity of Australian languages, which on the contrary have a great number of conjugations, etc. The extreme difficulty and complex structure of Eskimo and of many Amerindian languages is so notorious that no words need be wasted on them here. And the forms of the Basque verb are so manifold and intricate that we understand how Larramendi, in his legitimate pride at having been the first to reduce them to a system, called his grammar El Imposible Vencido, ‘The Impossible Overcome.’ At Béarn they have the story that the good God, wishing to punish the devil for the temptation of Eve, sent him to the Pays Basque with the command that he should remain there till he had mastered the language. At the end of seven years God relented, finding the punishment too severe, and called the devil to him. The devil had no sooner crossed the bridge of Castelondo than he found he had forgotten all that he had so hardly learned.

What is here said about the languages of wild tribes (and of the Basques, who are not exactly savages, but whose language is generally taken to have retained many primeval traits) is in exact keeping with everything that recent study of primitive man has brought to light: the life of the savage is regulated to the minutest details through ceremonies and conventionalities to be observed on every and any occasion; he is restricted in what he may eat and drink and when and how; and all these, to our mind, irrational prescriptions and innumerable prohibitions have to be observed with the most scrupulous, nay religious, care: it is the same with all the meticulous rules of his language.

XXI.—§ 9. Law of Development.

So far, then, from subscribing to Whitney’s dictum that “the law of simplicity of beginnings applies to language not less naturally and necessarily than to other instrumentalities” (G 226), we are drawn to the conclusion that primitive language had a superabundance of irregularities and anomalies, in syntax and word-formation no less than in accidence. It was capricious and fanciful, and displayed a luxuriant growth of forms, entangled one with another like the trees in a primeval forest. “Rien n’entre mieux dans les esprits grossiers que les subtilités des langues” (Tarde, Lois de l’imitation 285). Human minds in the early times disported themselves in long and intricate words as in the wildest and most wanton play. Nothing could be more beside the mark than to suppose that grammatical and logical categories were in primitive languages generally in harmony (as is supposed, e.g., by Sweet, New Engl. Grammar § 543): primitive speech cannot have been distinguished for logical consistency; nor, so far as we can judge, was it simple and facile: it is much more likely to have been extremely clumsy and unwieldy. Renan rightly reminds us of Turgot’s wise saying: “Des hommes grossiers ne font rien de simple. Il faut des hommes perfectionnés pour y arriver.”

We have seen in earlier chapters that the old theory of the three stages through which human language was supposed always to proceed, isolation, agglutination and flexion, was built up on insufficient materials; but while we feel tempted totally to reverse this system, we must be on our guard against establishing too rigid and too absolute a system ourselves. It would not do simply to reverse the order and say that flexion is the oldest stage, from which language tends through an agglutinative stage towards complete isolation, for flexion, agglutination and isolation do not include all possible structural types of speech. The possibilities of development are so manifold, and there are such innumerable ways of arriving at more or less adequate expressions for human thought, that it is next to impossible to compare languages of different families. Even, therefore, if it is probable that English, Finnish and Chinese are all simplifications of more complex languages, we cannot say that Chinese, for instance, at one time resembled English in structure and at some other time Finnish. English was once a flexional language, and is still so in some respects, while in others it is agglutinative, and in others again isolating, or nearly so. But we may perhaps give the following formula of what is our total impression of the whole preceding inquiry:

The evolution of language shows a progressive tendency from inseparable irregular conglomerations to freely and regularly combinable short elements.

The old system of historical linguistics may be likened to an enormous pyramid; only it is a pity that it should have as its base the small, square, strong, smart root word, and suspended above it the unwieldy, lumbering, ill-proportioned, flexion-encumbered sentence-vocable. Structures of this sort may with some adroitness be made to stand; but their equilibrium is unstable, and sooner or later they will inevitably tumble over.

XXI.—§ 10. Vocabulary.

On the lexical side of language we find a development parallel to that noticed in grammar; and, indeed, if we go deep enough into the question, we shall see that it is really the very same movement that has taken place. The more advanced a language is, the more developed is its power of expressing abstract or general ideas. Everywhere language has first attained to expressions for the concrete and special. In accounts of the languages of barbarous races we constantly come across such phrases as these: “The aborigines of Tasmania had no words representing abstract ideas; for each variety of gum-tree and wattle-tree, etc., they had a name; but they had no equivalent for the expression ‘a tree’; neither could they express abstract qualities, such as ‘hard, soft, warm, cold, long, short, round’”; or, The Mohicans have words for cutting various objects, but none to convey cutting simply. The Zulus have no word for ‘cow,’ but words for ‘red cow,’ ‘white cow,’ etc. (Sayce S 2. 5, cf. 1. 121). In Bakaïri (Central Brazil) “each parrot has its special name, and the general idea ‘parrot’ is totally unknown, as well as the general idea ‘palm.’ But they know precisely the qualities of each subspecies of parrot and palm, and attach themselves so much to these numerous particular notions that they take no interest in the common characteristics. They are choked in the abundance of the material and cannot manage it economically. They have only small coin, but in that they must be said to be excessively rich rather than poor” (K. v. d. Steinen, Unter den Naturvölkern Brasiliens, 1894, 81). The Lithuanians, like many primitive tribes, have many special, but no common names for various colours: one word for gray in speaking about wool and geese, one about horses, one about cattle, one about the hair of men and some animals, and in the same way for other colours (J. Schmidt, Kritik d. Sonantentheorie 37). Many languages have no word for ‘brother,’ but words for ‘elder brother’ and ‘younger brother’; others have different words according to whose (person and number) father or brother it is (see, e.g., the paradigm in Gabelentz Spr 421), and the same applies in many languages to names for various parts of the body. In Cherokee, instead of one word for ‘washing’ we find different words, according to what is washed: kutuwo ‘I wash myself,’ kulestula ‘I wash my head,’ tsestula ‘I wash the head of somebody else,’ kukuswo ‘I wash my face,’ tsekuswo ‘I wash the face of somebody else,’ takasula ‘I wash my hands or feet,’ takunkela ‘I wash my clothes,’ takutega ‘I wash dishes,’ tsejuwu ‘I wash a child,’ kowela ‘I wash meat’ (see, however, the criticism of Hewitt, Am. Anthropologist, 1893, 398). Primitive man did not see the wood for the trees.[108]

In some Amerindian languages there are distinct series of numerals for various classes of objects; thus in Kwakiatl and Tsimoshian (Sapir, Language and Environment 239); similarly the Melanesians have special words to denote a definite number of certain objects, e.g. a buku niu ‘two coconuts,’ a buru ‘ten coconuts,’ a koro ‘a hundred coconuts,’ a selavo ‘a thousand coconuts,’ a uduudu ‘ten canoes,’ a bola ‘ten fishes,’ etc. (Gabelentz, Die melan. Spr. 1. 23). In some languages the numerals are the same for all classes of objects counted, but require after them certain class-denoting words varying according to the character of the objects (in some respects comparable to the English twenty head of cattle, Pidgin piecey; cf. Yule and Burnell, Hobson-Jobson s.v. Numerical Affixes). This reminds one of the systems of weights and measures, which even in civilized countries up to a comparatively recent period varied not only from country to country, sometimes even from district to district, but even in the same country according to the things weighed or measured (in England stone and ton still vary in this way).

In old Gothonic poetry we find an astonishing abundance of words translated in our dictionaries by ‘sea,’ ‘battle,’ ‘sword,’ ‘hero,’ and the like: these may certainly be considered as relics of an earlier state of things, in which each of these words had its separate shade of meaning, which was subsequently lost and which it is impossible now to determine with certainty. The nomenclature of a remote past was undoubtedly constructed upon similar principles to those which are still preserved in a word-group like horse, mare, stallion, foal, colt, instead of he-horse, she-horse, young horse, etc. This sort of grouping has only survived in a few cases in which a lively interest has been felt in the objects or animals concerned. We may note, however, the different terms employed for essentially the same idea in a flock of sheep, a pack of wolves, a herd of cattle, a bevy of larks, a covey of partridges, a shoal of fish. Primitive language could show a far greater number of instances of this description, and, so far, had a larger vocabulary than later languages, though, of course, it lacked names for a great number of ideas that were outside the sphere of interest of uncivilized people.

There was another reason for the richness of the vocabulary of primitive man: his superstition about words, which made him avoid the use of certain words under certain circumstances—during war, when out fishing, during the time of the great cultic festivals, etc.—because he feared the anger of gods or demons if he did not religiously observe the rules of the linguistic tabu. Accordingly, in many cases he had two or more sets of words for exactly the same notions, of which later generations as a rule preserved only one, unless they differentiated these words by utilizing them to discriminate objects that were similar but not identical.

XXI.—§ 11. Poetry and Prose.

On the whole the development of languages, even in the matter of vocabulary, must be considered to have taken a beneficial course; still, in certain respects one may to some extent regret the consequences of this evolution. While our words are better adapted to express abstract things and to render concrete things with definite precision, they are necessarily comparatively colourless. The old words, on the contrary, spoke more immediately to the senses—they were manifestly more suggestive, more graphic and pictorial: while to express one single thing we are not unfrequently obliged to piece the image together bit by bit, the old concrete words would at once present it to the hearer’s mind as a whole; they were, accordingly, better adapted to poetic purposes. Nor is this the only point in which we see a close relationship between primitive words and poetry.

If by a mental effort we transport ourselves to a period in which language consisted solely of such graphic concrete words, we shall discover that, in spite of their number, they would not suffice, taken all together, to cover everything that needed expression; a wealth in such words is not incompatible with a certain poverty. They would accordingly often be required to do service outside of their proper sphere of application. That a figurative or metaphorical use of words is a factor of the utmost importance in the life of all languages is indisputable; but I am probably right in thinking that it played a more prominent part in old times than now. In the course of ages a great many metaphors have lost their freshness and vividness, so that nobody feels them to be metaphors any longer. Examine closely such a sentence as this: “He came to look upon the low ebb of morals as an outcome of bad taste,” and you will find that nearly every word is a dead metaphor.[109] But the better stocked a language is with those ex-metaphors which have become regular expressions for definite ideas, the less need there is for going out of one’s way to find new metaphors. The expression of thought therefore tends to become more and more mechanical or prosaic.

Primitive man, however, on account of the nature of his language, was constantly reduced to using words and phrases figuratively: he was forced to express his thoughts in the language of poetry. The speech of modern savages is often spoken of as abounding in similes and all kinds of figurative phrases and allegorical expressions. Just as in the literature transmitted to us poetry is found in every country to precede prose, so poetic language is on the whole older than prosaic language; lyrics and cult songs come before science, and Oehlenschläger is right when he sings (in N. Møller’s translation):