ROOTS OF LANGUAGE.

In the last chapter my treatment of the classification and phylogeny of languages may have led the general reader to feel that philologists display extraordinary differences of opinion with regard to certain first principles of their science. I may, therefore, begin the present chapter by reminding such a reader that I have hitherto been concerned more with the differences of opinion than with the agreements. If one takes a general view of the progress of philological science since philology—almost in our own generation—first became a science, I think he must feel much more impressed by the amount of certainty which has been attained than by the amount of uncertainty which still remains. And the uncertainty which does remain is due rather to a backwardness of study than to differences of interpretation. When more is known about the structure and mutual relations of the polysynthetic tongues, it is probable that a better agreement will be arrived at touching the relation of their common type to that of isolating tongues on the one hand, and agglutinating on the other. But, be this as it may, even as matters stand at present, I think we have more reason to be surprised at the certainty which already attaches to the principles of philology, than at the uncertainty which occasionally arises in their applications to the comparatively unstudied branches of linguistic growth.

Furthermore, important as these still unsettled questions are from a purely philological point of view, they are not of any great moment from that of the evolutionist, as I have already observed. For, so long as it is universally agreed that all the language-groups have been products of a gradual development, it is, comparatively speaking, immaterial whether the groups all stand to one another in a relation of serial descent, or whether some of them stand to others in a relation of collateral descent. That is to say, the evolutionist is under no obligation to espouse either the monotypic or the polytypic theory of the origin of language. Therefore, it will make no material difference to the following discussion whether the reader feels disposed to follow the doctrine, that all languages must have originated in such monosyllabic isolations as we now meet with in a radical form of speech like the Chinese; that they all originated in such polysynthetic incapsulations as we now find in the numberless dialects of the American Indians; or, lastly, and as I myself think much more probably, that both these, and possibly other types of language-structure, are all equally primitive. Be these things as they may, my discussion will not be overshadowed by their uncertainty. For this uncertainty has reference only to the origin of the existing language-types as independent or genetically allied: it in no way affects the certainty of their subsequent evolution. Much as philologists may still differ upon the mutual relations of these several language-types, they all agree that “von der ersten Entstehung der Sprachwurzeln an bis zur Bildung der volkommenen Flexionssprachen, wie des Sanskrit, Griechischen, oder Deutschen, ist Alles in der Entwicklung der Sprache verstÄndlich.... Sobald nur die Wurzeln als die fertigen Bausteine der Sprache einmal da sind, lÄsst sich Schritt fÜr Schritt das Wachsthum des SprachgebÄudes verfolgen.”[165]

Therefore, having now said all that seems necessary to say on the question of language-types, I will pass on to consider the information that we possess on the subject of language-roots.

First, let us consider the number of roots out of which languages are developed—or, rather, let me say, the number of elementary constituents into which the researches of philologists have been able to reduce those languages which have been most closely studied. Of course the probability—nay, the certainty—is that the actual number of roots must in all cases be considerably less than philologists are now able to prove.

Chinese is composed of about five hundred separate words, each being a monosyllable. In actual use, these five hundred root-words are multiplied to over fifteen hundred by significant variety of intonation; but the entire structure of this still living language is made up of five hundred monosyllabic words. In the opinion of most philologists we have here a survival of the root stage of language; but in the opinion of some we have the remnants of erosion, or “phonetic decay.”[166] This difference of opinion, however, is not a matter of importance to us; and therefore I will not discuss it, further than to say that on account of it I will not hereafter draw upon the Chinese language for illustrations of “radical” utterance, except in so far as philologists of all schools would allow as legitimate.[167]

Hebrew has been reduced to about the same number of roots as Chinese—Renan stating it in round numbers at five hundred.[168] But without doubt this number would admit of being considerably reduced, if inquiries were sufficiently extended to the whole Semitic family.

According to Professor Skeat, English is entirely made up of 461 Aryan roots, in combination with about twenty modifying constants.[169] The remote progenitor, Sanskrit, has been estimated to present as many as 850 roots, or, according to Benfey, just about twice that number.[170] On the other hand, Max MÜller, as a result of more recent researches, professes to have reduced the total number of Sanskrit roots to 121.[171]

It is needless to give further instances. For these are enough to show that, even if we were to regard the analytic powers of comparative philology as adequate to resolve all the compounds of a language into its primitive elements the estimate of Pott would probably be high above the mark, when he states that on an average the roots of a language may be taken at a thousand.[172] Seeing that Chinese only contains in its whole vocabulary half that number of words, and that both Hebrew and English have similarly yielded each about five hundred radicals in the crucible of more modern research, I think we may safely reduce the general estimate of Pott by one-half, and probably would be nearer the truth if we were to do so by three-quarters, or more. At all events, we may be satisfied that the total number of radicals sufficient to feed the most luxuriant of languages is expressible in three figures; and this, as we shall presently see, is enough for all the purposes of my subsequent discussion.

Passing on now from the question of number to that of character, we have first to meet the question—What are these roots? Are they the actually primitive words of pre-historic languages, or are they what Max MÜller has aptly termed “phonetic types”? Here again we encounter a difference of opinion among philologists. Thus, for instance, Professor Whitney tells us that the Indo-European languages are all descended from an original monosyllabic tongue, and, therefore, that “our ancestors talked with one another in simple syllables, indicative of ideas of prime importance, but wanting all designation of their relations.”[173] On the other hand, it is objected to this view that “such a language is a sheer impossibility;”[174] that “there could be no hope of any mutual understanding” with a language restricted to such isolated and general terms, &c.[175] On this side of the question it is represented that “roots are the phonetic and significant types discovered by the analysis of the comparative philologist as common to a group of allied words;”[176] that “a root is the core of a group of allied words,”[177] “the naked kernel of a family of words.”[178] Or, to adopt a simile previously used in another connection, we may say that a root as now presented by the philologist is a composite photograph (or phonogram) of a number of words, all belonging to the same pre-historic language, and all closely allied in meaning.

The difference of authoritative teaching thus exhibited is not a matter of much importance for us. Nor, indeed, as we shall subsequently see, is it a difference so great as may at first sight appear. For even the phonetic-type theory does not doubt that all the aboriginal and unknown words, out of the composition of which a root is now extracted, must have been genetically allied with one another, and exhibited the closeness of their kinship by a close similarity of sound. Therefore, it does not make any practical difference whether we regard a root as itself a primitive word, which was used in some such way as the Chinese now use their monosyllabic terms; or whether we regard it as a generalized expression of a group of cognate words, all closely allied as to meaning. In fact, even so strong an adherent of the phonetic-type theory as Professor Max MÜller very clearly states this, where he says that, although “the mere root, qu root, may be denied the dignity of a word, as soon as a root is used for predication it becomes a word, whether outwardly it is changed or not.”[179]

Seeing, then, that this difference of opinion among philologists is not one of great importance for us, I will henceforth disregard it. And, as it will be conducive to brevity, if not also to clearness, I will speak of roots as archaic words, although by so doing I shall not intend to assume that they are more than phonetic types, or the nearest approach we can make to the words out of which they were generated.

We may next consider the kind of meanings which roots convey. Antecedently we might form various anticipations on this head, such as that they should be imitative of natural sounds, expressive of concrete ideas, and so forth. As a matter of fact, we find that they are not expressive of natural sounds; but, as far as we have now any means of judging, quite arbitrary. Moreover, they are not expressive of concrete or particular ideas; but always of abstract or general. Here, then, to begin with, we have two facts of apparently great importance. And they are both facts which, at first sight, seem to countenance the view that, in its last resort, comparative philology fails to testify to the natural origin of speech. But we must look into the matter more closely, and, in order to do this most fairly, I will quote from Professor Max MÜller the 121 roots into which he analyzes the Sanskrit language. This is the language which has been most carefully studied in the present connection, and of all its students Professor Max MÜller is least open to any suspicion of inclining to the side of “Darwinism.” The following is a list of what he calls “the 121 original concepts.”

1. Dig.
2. Plat, weave, sew, bind.
3. Crush, pound, destroy, waste, rub, smooth.
4. Sharpen.
5. Smear, colour, knead, harden.
6. Scratch.
7. Bite, eat.
8. Divide, share, eat.
9. Cut.

10. Gather, observe.
11. Stretch, spread.
12. Mix.
13. Scatter, strew.
14. Sprinkle, drip, wet.
15a. Shake, tremble, quiver, flicker.
15b. Shake, mentally, be angry, abashed, fearfully, etc.
16. Throw down, fall.
17. Fall to pieces.
18. Shoot, throw at.
19. Pierce, split.
20. Join, fight, check.
21. Tear.
22. Break, smash.
23. Measure.
24. Blow.
25. Kindle.
26. Milk, yield.
27. Pour, flow, rush.
28. Separate, free, leave, lack.
29. Glean.
30. Choose.
31. Cook, roast, boil.
32. Clean.
33. Wash.
34. Bend, bow.
35. Turn, roll.
36. Press, fix.
37. Squeeze.
38. Drive, thrust.
39. Push, stir, live.
40. Burst, gush, laugh, beam.
41. Dress.
42. Adorn.
43. Strip, remove.
44. Steal.
45. Check.
46. Fill, thrive, swell, grow strong.
47. Cross.
48. Sweeten.
49. Shorten.
50. Thin, suffer.
51. Fat, stick, love.
52. Lick.
53. Suck, nourish.
54. Drink, swell.
55. Swallow, sip.
56. Vomit.
57. Chew, eat.
58. Open, extend.
59. Reach, strive, rule, have.
60. Conquer, take by violence, struggle.
61. Perform, succeed.
62. Attack, hurt.
63. Hide, drive.
64. Cover, embrace.
65. Bear, carry.
66. Can, be strong.
67. Show.
68. Touch.
69. Strike.
70. Ask.
71. Watch, observe.
72. Lead.
73. Set.
74. Hold, wield.
75. Give, yield.
76. Cough.
77. Thirsty, dry.
78. Hunger.
79. Yawn.
80. Spue.
81. Fly.
82. Sleep.
83. Bristle, dare.
84. Be angry, harsh.
85. Breathe.
86. Speak.
87. Seek.
88. Hear.
89. Smell, sniff.
90. Sweat.
91. Seethe, boil.
92. Dance.
93. Leap.
94. Creep.
95. Stumble.
96. Stick.
97. Burn.
98. Dwell.
99. Stand.

100. Sink, lie, fail.
101. Swing.
102. Hang down, lean.
103. Rise up, grow.
104. Sit.
105. Toil.
106. Weary, waste, slacken.
107. Rejoice, please.
108. Desire, love.
109. Wake.
110. Fear.
111. Cool, refresh.
112. Stink.
113. Hate.
114. Know.
115. Think.
116. Shine.
117. Run.
118. Move, go.
119a. Noise, inarticulate.
119b. Noise, musical.
120. Do.
121. Be.

“These 121 concepts constitute the stock-in-trade with which I maintain that every thought that has ever passed through the mind of India, so far as it is known to us in its literature, has been expressed. It would have been easy to reduce that number still further, for there are several among them which could be ranged together under more general concepts. But I leave this further reduction to others, being satisfied as a first attempt with having shown how small a number of seeds may produce, and has produced, the enormous intellectual vegetation that has covered the soil of India from the most distant antiquity to the present day.”[180]

Now, the first thing which strikes one on reading this list is, that it unquestionably justifies the inference of its compiler, namely, “if the Science of Language has proved anything, it has proved that every term which is applied to a particular idea or object (unless it be a proper name) is already a general term.” But the next thing which immediately strikes one is that the list, surprisingly short as it is, nevertheless is much too long to admit of being interpreted as, in any intelligible sense of the words, an inventory of “original concepts”—unless by “original” we are to understand the ultimate results of philological analysis. That all these concepts are not “original” in the sense of representing the ideation of really primitive man, is abundantly proved by two facts.

The first is that fully a third of the whole number might be dispensed with, and yet leave no important blank in the already limited resources of the list for the purposes either of communication or reflection. To yawn, to spew, to vomit, to sweat, and so on, are not forms of activity of any such vital importance to the needs of a primitive community, as to demand priority of naming by any aboriginal framers of language. Moreover, as Professor Max MÜller himself elsewhere observes, “even these 121 concepts might be reduced to a much smaller number, if we cared to do so. Any one who examines them carefully, will see how easy it would have been to express to dig by to cut or to strike; to bite by to cut or to crush; to milk by to squeeze; to glean by to gather; to steal by to lift.... If we see how many special purposes can be served by one root, as I, to go, or Pas, to fasten, the idea that a dozen of roots might have been made to supply the whole wealth of our dictionary, appears in itself by no means so ridiculous as is often supposed.”[181]

Again, in the second place, a large proportional number of the words have reference to a grade of culture already far in advance of that which has been attained by most existing savages. “Many concepts, such as to cook, to roast, to measure, to dress, to adorn, belong clearly to a later phase of civilized life.”[182] It might have been suitably added that such “concepts” as to dig, to plant, to milk, &c., betoken a condition of pastoral life, which, as we know from abundant evidence, is representative of a comparatively high level of social evolution.[183] But if “many” of these concepts are thus unmistakably referable to semi-civilized as distinguished from savage life, what guarantee can we have that the remainder are “original”? Obviously we can have no such guarantee; but, on the contrary, find the very best, because intrinsic evidence, that they belong to a more or less high level of culture, far removed from that of primitive man. In other words, we must conclude that these 121 concepts are “original” only in the sense that they do not now admit of further analysis at the hands of comparative philologists: they are not original in the sense of bringing us within any measurable distance of the first beginnings of articulate speech.[184]

Nevertheless, they are of the utmost value and significance, in that they bring us down to a period of presumably restricted ideation, as compared with the enormous development since attained by various branches of this Indo-European stock—so far, at least, as the growth of language can be taken as a fair expression of such development. They are likewise of the highest importance as showing in how presumably short a period of time (comparatively speaking) so immense and divergent a growth may proceed from such a simple and germ-like condition of thought.[185] Lastly, they serve to show in a most striking manner that the ideas represented, although all of a general character, are nevertheless of the lowest degree of generality. Scarcely any of them present us with evidence of reflective thought, as distinguished from the naming of objects of sense-perception, or of the simplest forms of activity which are immediately cognizable as such.[186] In other words, few of these “original concepts” rise much higher in the scale of ideation than the level to which I have previously assigned what I have called “named recepts” or “pre-concepts.” A dumb animal, or an infant, presents a full receptual appreciation of the majority of actions which the catalogue includes; and, therefore, so that a society of human beings can speak at all (i.e. presents the power of naming their recepts), it is difficult to see how they could have avoided a denotation of the more important recepts which are here concerned.

Another most interesting feature of a general kind which the list presents is, that it is composed exclusively of verbs.[187] This peculiarity of the ultimate known roots of all languages, which shows them to have been expressive of actions and states as distinguished from objects and qualities, is a peculiarity on which Professor Max MÜller lays much stress. But the inference which he draws from the fact is clearly not justifiable. This inference is that, as every root expresses “the consciousness of repeated acts, such as scraping, digging, striking,” &c., the naming of actions, as distinguished from objects, “must be considered as the first step in the formation of concepts.” Now, in drawing this inference—and, indeed, throughout all his works as far as I remember—Professor Max MÜller has entirely overlooked two most important considerations. First, as already observed, that the roots in question are demonstrably very far from having been the original material of language as first coined by primitive man; and, next, that whatever this original material may have been, from the first there must have been a struggle for existence among the really primitive roots—only those surviving which were most fitted to survive as roots, i.e. as the parent stems of subsequent word-formations. Now, it appears to me obvious enough that archaic—though not necessarily aboriginal—words which were expressive of actions, would have stood a better chance of surviving as roots than those which may have been expressive of objects; first because they were likely to have been more frequently employed, and next because many of them must have lent themselves more readily to metaphorical extension—especially under a system of animistic thought.[188] And, if these things were so, there is nothing remarkable in words significant of actions having alone survived as roots.[189]

The consideration that it is only those words which were successful in the struggle for existence that can have become the progenitors of subsequent language—and therefore the only words that have been handed down to us as roots—has a still more important bearing upon another of Professor Max MÜller’s generalizations. From the fact that all his 121 Sanskrit roots are expressive of “general” ideas (by which term he of course includes what I call generic ideas), he concludes that from its very earliest origin speech must have been thus expressive of general ideas; or, in other words, that human language could not have begun by the naming of particulars: from the first it must have been concerned with the naming of “notions.” Now, of course, if any vestige of real evidence could be adduced to show that this “must have been” the case, most of the foregoing chapters of the present work would not have been written. For the whole object of these chapters has been to show, that on psychological grounds it is abundantly intelligible how the conceptual stage of ideation may have been gradually evolved from the receptual—the power of forming general, or truly conceptual ideas, from the power of forming particular and generic ideas. But if it could be shown—or even rendered in any degree presumable—that this distinctly human power of forming truly general ideas arose de novo with the first birth of articulate speech, assuredly my whole analysis would be destroyed: the human mind would be shown to present a quality different in origin—and, therefore, in kind—from all the lower orders of intelligence: the law of continuity would be interrupted at the terminal phase: an impassable gulf would be fixed between the brute and the man. As a matter of fact, however, there is not only no vestige of any such proof or even presumption; but, as we shall see in our two following chapters, there is uniform and overwhelming proof of precisely the opposite doctrine—proof, indeed, so uniform and overwhelming that it has long ago induced all other philologists to accept this opposite doctrine as one of the axioms of their science. Leaving, however, this proof to be adduced in its proper place, I have now merely to point out the futility of the evidence on which Professor Max MÜller relies.

This evidence consists merely in fact that the “121 original concepts,” which are embodied in the roots of Aryan speech, are expressive of “general ideas.” Now, this argument might be worth considering if there were the smallest reason to suppose that in these roots of Aryan speech we possess the aboriginal elements of language as first spoken by man. But as we well know that this is immeasurably far from being the case, the whole argument collapses. The mere fact that many words which have survived as roots are words expressive of general ideas, is no more than we might have antecedently expected. Remembering that it is a favourable condition to a word surviving as a root that it should prove itself a prolific parent of other words, obviously it is those words which were expressive of ideas presenting some degree of generality that would have had the best chance of thus coming down to us, even from the comparatively high level of culture which, as we have seen, is testified to by “the 121 original concepts.” Of course, as I have already said, the case would have been different if any one were free to suppose, even as a merely logical possibility, that this level of culture represented that of primitive man when he first began to employ articulate speech. But any such supposition is beyond the range of rational discussion. The 121 concepts themselves yield overwhelming evidence of belonging to a time immeasurably remote from that of any speechless progenitor of Homo sapiens; and in the enormous interval (whatever it may have been) many successive generations of words must certainly have flourished and died.[190]

These remarks are directed to the comparatively few instances of general ideas which, as a matter of fact, the list of “121 concepts” presents. As already observed, the great majority of these “concepts” exhibit no higher degree of “generality” than belongs to what I have called a “pre-concept,” i.e. a “named recept.” But precisely the same considerations apply to both. For, even supposing that a named recept was originally a word used only to designate a “particular” as distinguished from a “generic” idea, obviously it would have stood but a poor chance of surviving as a root unless it had first undergone a sufficient degree of extension to have become what I call receptually connotative. A proper name, for instance, could not, as such, become a root. Not until it had become extended to other persons or things of a like class could it have secured a chance of surviving as a root in the struggle for existence. As a matter of fact, I think it most probable—not only from general considerations, but also from a study of the spontaneous names first coined in “baby-language,”—that aboriginal speech was concerned simultaneously with the naming both of particular and of generic ideas—i.e. of individual percepts and of recepts. It will be remembered that in Chapter III., while treating of the Logic of Recepts, I dealt at some length with this subject. Here, therefore, it will be sufficient to quote the conclusion to which my analysis led.

“A generic idea is generic because the particular ideas of which it is composed present such obvious points of resemblance that they spontaneously fuse together in consciousness; but a general idea is general for precisely the opposite reason—namely, because the points of resemblance which it has seized are obscured from immediate perception, and therefore could never have fused together in consciousness but for the aid of intentional abstraction, or of the power of a mind knowingly to deal with its own ideas as ideas. In other words, the kind of classification with which recepts are concerned is that which lies nearest to the kind of classification with which all processes of so called perceptual inference depend—such as mistaking a bowl for a sphere. But the kind of classification with which concepts are concerned is that which lies furthest from this purely automatic grouping of perceptions. Classification there doubtless is in both cases; but in the one order it is due to the closeness of resemblances in an act of perception, while in the other it is due to their remoteness.”[191]

Of course it goes without saying that this “closeness of resemblances in an act of perception” may be due either to similarities of sense-perceptions themselves (as when the colour of a ruby is seen to resemble that of “pigeon’s blood”), or to frequency of their associations in experience (as when a sea-bird groups together in one recept the sundry sensations which go to constitute its perception of water, with its generic classification of water as a medium in which it is safe to dive). Now, if we remember these things, can we possibly wonder that the palÆontology of speech should prove early roots to have been chiefly expressive of “generic” as distinguished from “general” ideas on the one hand, or “particular” ideas on the other? By failing to observe this real distinction between classification as receptual and conceptual—i.e. as given immediately in the act of perception itself, or as elaborated of set purpose through the agency of introspective thought, Professor Max MÜller founds his whole argument on another and an unreal distinction: he everywhere regards the bestowing of a name as in itself a sufficient proof of conceptual thought, and therefore constitutes the faculty of denotation, equally with that of denomination, the distinctive criterion of a self-conscious mind. But, as we have now so repeatedly seen, such is certainly not the case. Actions and processes so habitual, or so immediately apparent to perception, as those with which the great majority of these “121 concepts” are concerned, do not betoken any order of ideation higher than the pre-conceptual, in virtue of which a young child is able to give expression to its higher receptual life prior to the advent of self-consciousness. Or, as Geiger tersely says:—“In enzelnen FÄllen ist die Entstehung von Gattungsbegriffe aus Mangel an Unterscheidung gleichwohl kaum zu bezweifeln.”[192]

Again, if we look to the still closer analogy furnished by savages, we meet with a still further corroboration of this view. For instance, Professor Sayce remarks that in “all savage and barbarous dialects, while individual objects of sense have a superabundance of names, general terms are correspondingly rare.” And he gives a number of remarkable illustrations.[193]

In view of these considerations, my only wonder is that these 120 root-words do not present better evidence of conceptual thought. I have already given my reasons for refusing to suppose that we have here to do with the “original” framers of spoken language; and looking to the comparatively high level of culture which the people in question must have reached, it seems remarkable that the root-words of their language should only in so few instances have risen above the level of pre-conceptual utterance.[194] This, however, only shows how comparatively small a part self-conscious reflection need play in the practical life of uncultured man: it does not show that the people in question were remarkably deficient in this distinctively human faculty. Archdeacon Farrar tells us that he has observed the whole conversational vocabulary of certain English labourers not to exceed a hundred words, and probably further observation would have shown that the great majority of these were employed without conceptual significance. Therefore, if these labourers had had to coin their own words, it is probable that, without exception, their language would have been destitute of any terms betokening more than a pre-conceptual order of ideation. Nevertheless, these men must have been capable, in however undeveloped a degree, of truly conceptual ideation: and this proves how unsafe it would be to argue from the absence of distinctively conceptual terms to the poverty of conceptual faculty among any people whose root-words may have come down to us—although, no doubt, in such a case we appear to be getting within a comparatively short distance of the origin of this faculty.

The point, however, now is that really aboriginal, and therefore purely denotative names, must certainly have been “generic” as well as “particular”: they must have been the names of recepts as well as of percepts, of actions as well as of objects and qualities. Moreover, it is equally certain that among this aboriginal assemblage of denotative names as particular and generic, only those belonging to the latter class could have stood much chance of surviving as roots. In other words, no aboriginal name could have survived as a root until it had acquired some greater or less degree of receptual and, therefore, of connotative value. Hence the fact that the ultimate result of the philological analysis of any language is that of reducing the language to a certain small number of roots, and the fact that all these roots are expressive of general and generic ideas,—these facts in themselves yield no support whatever to the doctrine, either that these roots were themselves the aboriginal elements of language, or, a fortiori, that the aboriginal elements of language were expressive of general ideas.[195]

And this conclusion involves another of scarcely less importance. A great deal of discussion has been expended over the question as to whether, or how far, aboriginal language was indebted to the principle of onomatopoeia, or the imitation by articulate names of sounds obviously distinctive of the objects or actions named. Of course, on evolutionary principles we should be strongly inclined to suppose that aboriginal language must have been largely assisted in its formation by such intentional imitation of natural sounds, seeing that of all forms of vocal expression they admit of most readily conveying an idea of the object or action named. And the same applies to the so-called interjectional element in word-formation, or the utilization as names of sounds which are naturally expressive of states of human feeling. On the other hand, contempt has been poured upon this theory as an adequate explanation of the first beginnings of articulate speech, on the ground that it is not supported either by history[196] or by the results of philogenetic inquiry.[197] It is, however, forgotten by those who argue on this side that names of onomatopoetic origin must always be, in the first instance, particular; that so long as they remain particular (as, for example, is the case with our word “cuckoo”), they cannot have much chance of surviving as roots; that in proportion as they increase their chances of survival as roots by becoming more general, they must do so by becoming more conventional; and, therefore, that the vast majority of roots, even if aboriginally they were of onomatopoetic origin, must necessarily have had that origin obscured.

In order to illustrate each and all of these general considerations, let us turn to the example of our own “baby-language.” The fact that such language presents so large an element of onomatopoeia in itself furnishes a strong presumption that what is now seen to constitute so important a principle in the infancy of the individual (notwithstanding the hereditary tendency to speak), must have constituted at least as important a principle in the infancy of the race. But the point now is, that if we mark the connotative extension of any such nursery word, we may find that just in proportion as it becomes general does its onomatopoetic origin become obscure. For instance, the late Mr. Darwin gave me the following particulars with regard to a grandchild of his own, who was then living in his house. I quote the account from notes taken at the time.

“The child, who was just beginning to speak, called a duck ‘quack’; and, by special association, it also called water ‘quack.’ By an appreciation of the resemblance of qualities, it next extended the term ‘quack’ to denote all birds and insects on the one hand, and all fluid substances on the other. Lastly, by a still more delicate appreciation of resemblance, the child eventually called all coins ‘quack,’ because on the back of a French sou it had once seen the representation of an eagle. Hence, to the child, the sign ‘quack,’ from having originally had a very specialized meaning, became more and more extended in its signification, until it now serves to designate such apparently different objects as ‘fly,’ ‘wine,’ and ‘coin.’”

Now, if any such process of extending or generalizing aboriginally onomatopoetic terms were to have taken place among the primitive framers of human speech, how hopeless would be the task of the philologist who should now attempt to find the onomatopoetic root! Yet, as above observed, not only may we be perfectly certain that such extensions of aboriginal onomatopoetic terms must have taken place, if any such terms were ever in existence at all (and this cannot be doubted), but also that it must have been almost a necessary condition to the survival of an onomatopoetic term as a root that such an extension of its meaning should have taken place. In other words, we can see very good reason to conclude that, as a rule, only those instances of primitive onomatopoeia can have survived as roots, which must long ago have had their onomatopoetic origin hopelessly obscured. So that nowhere so much as in this case should we be prepared to entertain the general principle of philological research, that, as Goethe graphically states it, the original meanings of words become gradually worn out, like the image and superscription of a coin.[198]

In view of such considerations, my only wonder is that this origin admits of being traced so often as it does, even as far back as the comparatively recent times when a pastoral people coined the terms which afterwards constituted the roots of Sanskrit. Kas, to cough; kshu, to sneeze; proth, to snort; ma, to bleat, and not a few others, are conceded, even by Professor Max MÜller, to be of obviously imitative origin. In the present connection, however, it is of interest to notice how this authority deals with such cases. He says:—“Not one of them is of any importance in helping us to account for real words in Sanskrit. Most of them have had no offspring at all, others have had a few descendants, mostly sterile. Their history shows clearly how far the influence of onomatopoeia may go, and if once we know its legitimate sphere, we shall be less likely to wish to extend it beyond its proper limits.”[199]

Now, under our present point of view we can see a very good reason why this element of sterility should have attached to these roots of Sanskrit whose onomatopoetic origin still admits of being clearly traced: it is just because they failed to be extended that their imitative source continues to be apparent.[200] But suppose, for the sake of illustration, that any one of them had been extended, and what would have happened? If ma, to bleat, had been metaphorically applied to the crying of a child, and had then become more and more habitually used in this new signification, while the original meaning became more and more obsolete, it might have taken the place of any such root as bhi, to fear; ish, to love, &c.; and in all the progeny of words which in this its conventional use it might subsequently have generated, no trace of imitative origin could now have been met with—any more than such an origin can be detected in the sound “quack,” as used by the above-mentioned child to designate a shilling.

Several other considerations to the same general effect might be adduced. But, to mention only some of the more important, Steinthal points out that imitative utterance differs widely even among different races of existing men, so that the onomatopoetic words of one race do not convey any imitative suggestion to the minds of another.[201] Similarly, Professor Sayce insists, “it is not necessary that the imitation of natural sounds should be an exact one; indeed, that it never can be: all that is wanted is that the imitation should be recognizable by those addressed. The same natural sound, consequently, may strike the ear of different persons very differently, and so be represented in articulate speech in a strangely varying manner.”[202] Another very good illustration of the same point is to be found in the names for a grasshopper in different languages. After giving a number, Archdeacon Farrar remarks that obviously they are “all imitative: yet how immensely varied by the fantasies of imitation! How is this to be explained? Simply by the fact to which it is so often necessary to recur, that words are not mere imitations, but subjective echoes and reproductions—repercussions which are modified both organically and ideally—which have moreover been immensely blurred and disintegrated by the lapse of ages.”[203]

But perhaps the best illustration that has been given of this point is in the different words which obtain in different languages as names for Thunder. Two independent treatises have been written on the subject, one by Grimm,[204] and the other by Pott.[205] While in nearly all the languages the principle of imitation is more or less clearly apparent, the greatest diversities occur among the resulting sounds.[206] In this connection, also, I may adduce yet one further consideration. In his Introduction to the Science of Language, Professor Sayce argues on several grounds that, when articulation first began, the articulate sounds were probably in large part dependent for their meaning on the gestures with which they were accompanied. Consequently, aboriginal root-words, even supposing that any such had come down to us, and that their origin were imitative, inasmuch as their imitative value may thus have in large part depended on appropriately accompanying gestures, their imitative source would long ago have become obscured.

In view of all these considerations, therefore, I cannot deem the merely negative evidence against the onomatopoetic origin of articulate sounds as of any value at all. Even if we had any reason to suppose that philological analysis were in possession of the really aboriginal commencements of spoken language, we should still be unable reasonably to conclude against their imitative origin, merely on the ground that in our greatly altered circumstances of life and of mind we are not now able to trace the imitations.

As a matter of fact, however, the evidence which we have on the subject is not all negative. On the contrary, there is an overwhelming body of actual and unquestionable proof of the imitative origin of very many words in all languages—especially those which are spoken by savages, and are known from their general structure to be in a comparatively undeveloped state. The evidence being much too copious for quotation, I must content myself with referring to the excellent and most forcible epitome which is given of it by Archdeacon Farrar in his works on the Origin of Language and Chapters on Language.[207] The foregoing remarks, therefore, which I have made on the negative side of the question, are merely intended to show that the element of onomatopoeia must have entered into the composition of aboriginal speech much more largely than philologists are now able to prove, notwithstanding that they have been able to prove how immensely important an element it has been in this respect. The only wonder is, that when so many causes have been at work in obscuring and corroding the originally imitative significance of words, this significance should still admit of being traced in all languages—even the most highly conventionalized—to the very large extent in which it does.

The hostility which Professor Max MÜller has displayed to the onomatopoetic theory of the origin of language is the more remarkable, because in his latest work he has enthusiastically embraced a special branch of this theory, which has been put forward by M. NoirÉ. This special branch of the onomatopoetic theory is that articulate sign-making had its origin in sounds which are made by bodies of men when engaged in some common occupation. When sailors row, soldiers march, builders co-operate in pulling or in lifting, &c., there is always a tendency to give vent to appropriate sounds, which the nature of the occupation usually breaks up into rhythmic periods. “These utterances, noises, shouts, hummings, or songs are a kind of natural reaction against the inward disturbance caused by muscular effort. They are the almost involuntary vibrations of the voice, corresponding to the more or less regular movements of our whole bodily frame.” The hypothesis, therefore, is that sounds thus naturally evolved, and differing with different occupations, would sooner or later come to be conventionally used as the names of these different occupations. And, if thus used habitually, they would be virtually the same as words, inasmuch as they would not merely admit of immediate understanding on the part of others, but, what is even of more importance, they would, by the mere fact of such conventional usage of names, elevate what had previously been but a receptual appreciation of an act into a pre-conceptual designation of it.

Now, I say that this hypothesis, whatever may be thought as to its probability, is clearly but a special branch of the general theory of onomatopoeia. So that primitive names were intentionally imitative of natural sounds, for all the purposes of onomatopoetic theory it makes no difference whether such sounds were made by natural objects or by man himself. Nor, of the natural sounds which were made by man himself, does it in any way affect this theory whether the naturally human sounds were “interjectional” only, “co-operative” only, or sometimes one and sometimes the other. If, following the example set by Professor Max MÜller, I may be allowed to designate NoirÉ’s special branch of the onomatopoetic theory as the Yeo-he-ho theory, it appears to me impossible to distinguish it in any essential particular from those other branches which are called by him the Bow-wow and Pooh-pooh theories—i.e. the imitative and the interjectional. Yet he has become as ardent a supporter of the one branch as he was a vehement opponent of the others.[208]

For my own part, I think it highly probable that there is an element of truth in the Yeo-he-ho theory, although I deem it in the last degree improbable that imitative sounds of this kind constituted the only source of aboriginal speech. At the most, it seems to me, this branch of onomatopoeia can be accredited with supporting but a small proportional part of aboriginal language-growth. Nevertheless, as already observed, I can have no doubt at all that the principle of onomatopoeia in all its branches has been the most important of all principles which were concerned in the first genesis of speech. That is to say, I fully agree with the almost unanimous voice of philological authority on this matter, which may be tersely expressed by allowing Professor Whitney to act as spokesman.

“Beyond all reasonable question, there was a positively long period of purely imitative signs, and a longer one of mixed imitative and traditional ones, the latter gradually gaining upon the former, before the present condition of things was reached, when the production of new signs by imitation is only sporadic and of the utmost rarity, and all language-signs besides are traditional, their increase in any community being solely caused by variation and combination, and by borrowing from other communities.”[209]

But now, having thus stated as emphatically as possible my acceptance of the theory of onomatopoeia, I have to express dissent from many of its more earnest advocates where they represent that it is necessarily the only theory to be entertained. In other words, I do not agree with the dogma that articulate speech cannot possibly have had any source, or sources, other than that which is supplied by vocal imitations.[210] For, on merely antecedent grounds, I can see no adequate reason for arbitrarily excluding the possibility of arbitrary invention. If even civilized children, who are not under the discipline of the “mother of invention,” will coin a language of their own in which the element of onomatopoeia is barely traceable;[211] and if uneducated deaf-mutes will spontaneously devise articulate sounds which are necessarily destitute of any imitative origin;[212] I do not see why it should be held antecedently impossible that primitive man can have found any other means of word-formation than that which is supplied by mimicry. Therefore, while I fully agree with Professor Wundt in holding that the question before us is one to be dealt with by psychology rather than philology (seeing that language cannot record the conditions of its own birth, and that so many causes have been at work to obliterate aboriginal onomatopoeia), I cannot follow him where he argues that on grounds of psychology there is no room for any other inference than that the principle of onomatopoeia in its widest sense must have constituted the sole origin of significant articulation.[213]

We have already seen that even the most imitative of vocalists, the talking birds, will invent wholly arbitrary sounds as denotative names,[214] and it would be psychologically absurd to suppose that they are superior to what primitive man must have been in the matter of finding expedients for semiotic utterance. Again, the clicks of Hottentots and Bushmen, whatever we suppose their origin to have been, certainly cannot have had that origin in onomatopoeia; and no less certainly, as Professor Sayce remarks, they still survive to show how the utterances of speechless man could be made to embody and convey ideas.[215] Lastly, on the general principle that the development of the individual furnishes information touching the development of the race, it is highly significant that the hitherto speechless child will spontaneously use arbitrary sounds (both articulate and otherwise) whereby to denotate habitual recepts. And even after it has begun to learn the use of actual words, arbitrary additions are frequently made to its vocabulary which defy any explanation at the hands of onomatopoeia—not only, as in the cases above alluded to, where they are left to themselves, but even in cases where they are in the closest contact with language as spoken by their elders. I could quote many instances of this fact; but it will be enough to refer to one already given on page 144 (foot-note). When, however, these spontaneous efforts are not controlled by constant association with elders, but fostered by children of about the same age being left much together, the remarkable consequence previously alluded to arises—namely, a newly devised language which depends but in small part upon the principle of onomatopoeia, and is therefore wholly unintelligible to all but its inventors.[216]

I have now briefly stated all the main facts and considerations which appear to me worth stating, both for and against the theory of onomatopoeia. And, having done this, I wish in conclusion to make it clear that the matter is not one which seriously affects the theory of evolution. To the philologist, no doubt, the question as to how far the element of onomatopoeia entered into the formation of aboriginal speech is a really important question, so that, as Geiger says, “Diess ist die gemeinsame Frage, und die antwort wird auf der einen Seite von einem inneren Zusammenhang zwischen je einem Laut und dem entsprechenden Begriffe, auf der andern aus WillkÜr und Uebereinkunft hergeleitet.”[217] But the question is one which the evolutionist may view with indifference. Whether words were all originally dependent on an inherent connection between every sound they made and the idea thereby expressed, or whether they were all due to arbitrary invention, in either case the evolutionist may see that they can equally well have come into existence as the natural products of a natural psychogenesis. And, a fortiori, as an evolutionist, he need not greatly concern himself with any further question as to the relative degrees in which imitation and invention may have entered into the composition of primitive speech.