Primitive Culture/Chapter 6

by Edward Burnett Tylor

 

CHAPTER VI.

EMOTIONAL AND IMITATIVE LANGUAGE (continued).

Imitative Words — Human actions named from sound — Animals' names from cries, &c. — Musical Instruments — Sounds reproduced — Words modified to adapt sound to sense — Reduplication — Graduation of vowels to express distance and difference — Children's Language — Sound-words as related to Sense-words — Language an original product of the lower Culture.

From the earliest times of language to our own day, it is unlikely that men ever quite ceased to be conscious that some of their words were derived from imitation of the common sounds heard about them. In our own modern English, for instance, results of such imitation are evident; flies buzz, bees hum, snakes hiss, a cracker or a bottle of ginger-beer pops, a cannon or a bittern booms. In the words for animals and for musical instruments in the various languages of the world, the imitation of their cries and tones is often to be plainly heard, as in the names of the hoopoe, the ai-ai sloth, the kaka parrot, the Eastern tomtom, which is a drum, the African ulule, which is a flute, the Siamese khong-bong, which is a wooden harmonicon, and in like manner through a host of other words. But these evident cases are far from representing the whole effects of imitation on the growth of language. They form, indeed, the easy entrance to a philological region, which becomes less penetrable the farther it is explored.

The operations of which we see the results before us in the actual languages of the world seem to have been somewhat as follows. Men have imitated their own emotional utterances or interjections, the cries of animals, the tones of musical instruments, the sounds of shouting, howling, stamping, breaking, tearing, scraping, with others which are all day coming to their ears, and out of these imitations many current words indisputably have their source. But these words, as we find them in use, differ often widely, often beyond all recognition, from the original sounds they sprang from. In the first place, man's voice can only make a very rude copy of most sounds his ear receives; his possible vowels are very limited in their range compared with natural tones, and his possible consonants still more helpless as a means of imitating natural noises. Moreover, his voice is only allowed to use a part even of this imperfect imitative power, seeing that each language for its own convenience restricts it to a small number of set vowels and consonants, to which the imitative sounds have to conform, thus becoming conventionalized into articulate words with further loss of imitative accuracy. No class of words have a more perfect imitative origin than those which simply profess to be vocal imitations of sound. How ordinary alphabets to some extent succeed and to some extent fail in writing down these sounds may be judged from a few examples. Thus, the Australian imitation of a spear or bullet striking is given as toop; to the Zulu, when a calabash is beaten, it says boo; the Karens hear the flitting ghosts of the dead call in the wailing voice of the wind, re, re, ro, ro; the old traveller, Pietro della Valle, tells how the Shah of Persia sneered at Timur and his Tartars, with their arrows that went ter ter; certain Buddhist heretics maintained that water is alive, because when it boils it says chichitá, chitichitá, a symptom of vitality which occasioned much theological controversy as to drinking cold and warm water. Lastly, sound-words taken up into the general inventory of a language have to follow its organic changes, and in the course of phonetic transition, combination, decay, and mutilation, to lose ever more and more .their original shape. To take a single example, the French huer 'to shout' (Welsh hwa) may be a perfect imitative verb; yet when it passes into modern English hue and cry, our changed pronunciation of the vowel destroys all imitation of the call. Now to the language-makers all this Was of little account. They merely wanted recognized words to express recognized thought, and no doubt arrived by repeated trials at systems which were found practically to answer this purpose. But to the modern philologist, who is attempting to work out the converse of the problem, and to follow backward the course of words to original imitative sound, the difficulty is most embarrassing. It is not only that thousands of words really derived from such imitation may now by successive change have lost all safe traces of their history; such mere deficiency of knowledge is only a minor evil. What is far worse is that the way is thrown open to an unlimited number of false solutions, which yet look on the face of them fully as like truth as others which we know historically to be true. One thing is clear, that it is of no use to resort to violent means, to rush in among the words of language, explaining them away right and left as derived each from some remote application of an imitative noise. The advocate of the Imitative Theory who attempts this, trusting in his own powers of discernment, has indeed taken in hand a perilous task, for, in fact, of all judges of the question at issue, he has nourished and trained himself up to become the very worst. His imagination is ever suggesting to him what his judgment would like to find true; like a witness answering the questions of the counsel on his own side, he answers in good faith, but with what bias we all know. It was thus with De Brosses, to whom this department of philology owes so much. It is nothing to say that he had a keen ear for the voice of Nature; she must have positively talked to him in alphabetic language, for he could hear the sound of hollowness in the sk of σκάπτω 'to dig,' of hardness in the cal of callosity, the noise of insertion of a body between two others in the tr of trans, intra. In enquiries so liable to misleading fancy, no pains should be spared in securing impartial testimony, and it fortunately happens that there are available sources of such evidence, which, when thoroughly worked, will give to the theory of imitative words as near an approach to accuracy as has been attained to in any other wide philological problem. By comparing a number of languages, widely apart in their general system and materials, and whose agreement as to the words in question can only be accounted for by similar formation of words from similar suggestion of sound, we obtain groups of words whose imitative character is indisputable. The groups here considered consist in general of imitative words of the simpler kind, those directly connected with the special sound they are taken from, but their examination to some extent admits of words being brought in, where the connexion of the idea expressed with the sound imitated is more remote. This, lastly, opens the far wider and more difficult problem, how far imitation of sounds is the primary cause of the great mass of words in the vocabularies of the world, between whose sound and sense no direct connexion appears.

Words which express human actions accompanied with sound form a very large and intelligible class. In remote and most different languages, we find such forms as pu, puf, bu, buf, fu, fuf, in use with the meaning of puffing, fuffing; or blowing; Malay puput; Tongan buhi; Maori pupui; Australian bobun, bwa-bun; Galla bufa, afufa; Zulu futa, punga, pupuza (fu, pu, used as expressive particles); Quiché puba; Quichua puhuni; Tupi ypeû; Finnish puhkia; Hebrew puach; Danish puste; Lithuanian puciu; and in numbers of other languages;^[1] here, grammatical adjuncts apart, the significant force lies in the imitative syllable. Savages have named the European musket when they saw it, by the sound pu, describing not the report, but the puff of smoke issuing from the muzzle. The Society Islanders supposed at first that the white men blew through the barrel of the gun, and they called it accordingly pupuhi, from the verb puhi to blow, while the New Zealanders more simply called it a pu. So the Amaxosa of South Africa call it umpu, from the imitative sound pu! The Chinook Jargon of North-West America uses the phrase mamook poo (make poo) for a verb 'to shoot,' and a six-chambered revolver is called tohum poo, i.e., a 'six-poo.' When a European uses the word puff to denote the discharge of a gun, he is merely referring to the smoke blown out, as he would speak of a puff of wind, or even a powder-puff or a puff-ball; and when a pistol is called in colloquial German a puffer, the meaning of the word matches that used for it in French Argot, a 'soufflant.' It has often been supposed that the puff imitates the actual sound, the bang of the gun, and this has been brought forward to show by what extremely different words one and the same sound may be imitated, but this is a mistake.^[2] These derivations of the name of the gun from the notion of blowing correspond with those which give names to the comparatively noiseless blow-tube of the bird-hunter, called by the Indians of Yucatan a pub, in South America by the Chiquitos a pucuna, by the Cocamas a puna. Looking into vocabularies of languages which have such verbs 'to blow,' it is usual to find with them other words apparently related to them, and expressing more or less distant ideas. Thus Australian poo-yu, puyu 'smoke;' Quichua puhucuni 'to light a fire,' punquini 'to swell,' puyu, puhuyu 'a cloud;' Maori puku 'to pant,' puka 'to swell;' Tupi púpú, pupúre 'to boil;' Galla bube 'wind,' bubiza 'to cool by blowing;' Kanuri (root fu) fungin 'to blow, swell,' furúdu 'a stuffed pad or bolster,' &c., bubute 'bellows' (bubute fungin 'I blow the bellows'); Zulu (dropping the prefixes) puku, pukupu 'frothing, foam,' whence pukupuku 'an empty frothy fellow,' pupuma 'to bubble, boil,' fu 'a cloud,' fumfu 'blown about like high grass in the wind,' whence fumfuta 'to be confused, thrown into disorder,' futo 'bellows,' fuba 'the breast, chest,' then figuratively 'bosom, conscience.'

The group of words belonging to the closed lips, of which mum, mumming, mumble are among the many forms belonging to European languages,^[3] are worked out in like manner among the lower races — Vei mu mu 'dumb'; Mpongwe imamu 'dumb'; Zulu momata (from moma, 'a motion with the mouth as in mumbling') 'to move the mouth or lips,' mumata 'to close the lips as with a mouthful of water,' mumuta, mumuza 'to eat mouthfuls of corn, &c., with the lips shut;' Tahitian mamu 'to be silent,' omumu 'to murmur;' Fijian, nomo, nomo-nomo 'to be silent;' Chilian, ñomn 'to be silent;' Quiché, mem 'mute,' whence memer 'to become mute;' Quichua, amu 'dumb, silent,' amullini 'to have something in the mouth,' amullayacuni simicta 'to mutter, to grumble.' The group represented by Sanskrit t'hût'hû 'the sound of spitting,' Persian thu kerdan (make thu) 'to spit,' Greek (Greek characters), may be compared with Chinook mamook toh, tooh, (make toh, tooh); Chilian tuvcùtun (make tuv); Tahitian tutua; Galla twu; Yoruba tu. Among the Sanskrit verb-roots, none carries its imitative nature more plainly than kshu 'to sneeze;' the following analogous forms are from South America: — Chilian, echiun; Quichua, achhini; and from various languages of Brazilian tribes, techa-ai, haitschu, atchian, natschun, aritischune, &c. Another imitative verb is well shown in the Negro-English dialect of Surinam, njam 'to eat' (pron. nyam), njam-njam 'food' ('en hem njanjam ben de sprinkhan nanga boesi-honi' — 'and his meat was locusts and wild honey'). In Australia the imitative verb 'to eat' reappears as g'nam-ang. In Africa the Susu language has nimnim, 'to taste,' and a similar formation is observed in the Zulu nambita 'to smack the lips after eating or tasting, and thence to be tasteful, to be pleasant to the mind.' This is an excellent instance of the transition of mere imitative sound to the expression of mental emotion, and it corresponds with the imitative way in which the Yakama language, in speaking of little children or pet animals, expresses the verb 'to love' as nem-no-sha (to make n'm-n'). In more civilized countries these forms are mostly confined to baby-language. The Chinese child's word for eating is nam, in English nurseries nim is noticed as answering the same purpose, and the Swedish dictionary even recognizes namnam 'a tid-bit.'

As for imitative names of animals derived from their cries or noises, they are to be met with in every language from the Australian twonk 'frog,' the Yakama rol-rol 'lark,' to the Coptic eeiō 'ass,' the Chinese maou 'cat,' and the English cuckoo and peewit. Their general principle of formation being acknowledged, their further philological interest turns mostly on cases where corresponding words have thus been formed independently in distant regions, and those where the imitative name of the creature, or its habitual sound, passes to express some new idea suggested by its character. The Sanskrit name of the kâka crow reappears in the name of a similar bird in British Columbia, the káh-káh; a fly is called by the natives of Australia a bumberoo, like Sanskrit bambharâli 'fly,' Greek (Greek characters), and our bumble-bee. Analogous to the name of the tse-tse fly, the terror of African travellers, is ntsintsi, the word for 'fly' among the Basutos, which also, by a simple metaphor, serves to express the idea of 'a parasite.' Mr. H. W. Bates's description seems to settle the dispute among naturalists, whether the toucan had its name from its cry or not. He speaks of its loud, shrill, yelping cries having 'a vague resemblance to the syllables tocáno, tocáno, and hence the Indian name of this genus of birds.' Granting this, we can trace this sound-word into a very new meaning; for it appears that the bird's monstrous bill has suggested a name for a certain large-nosed tribe of Indians, who are accordingly called Tucanos.^[4] The cock, gallo quiquiriqui, as the Spanish nursery-language calls him, has a long list of names from various languages which in various ways imitate his crowing; in Yoruba he is called koklo, in Ibo okoko, akoka, in Zulu kuku, in Finnish kukko, in Sanskrit kukkuta, and so on. He is mentioned in the Zend-Avesta in a very curious way, by a name which elaborately imitates his cry, but which the ancient Persians seem to have held disrespectful to their holy bird, who rouses men from sleep to good thought, word, and work: —

'The bird who bears the name of Parôdars, O holy Zarathustra; Upon whom evil-speaking men impose the name Kabrkataç.'^[5]

The crowing of the cock (Malay kâluruk, kukuk) serves to mark a point of time, cockcrow. Other words originally derived from such imitation of crowing have passed into other curiously transformed meanings: Old French cocart 'vain;' modern French coquet 'strutting like a cock, coquetting, a coxcomb;' cocarde 'a cockade' (from its likeness to a cock's comb); one of the best instances is coquelicot, a name given for the same reason to the wild poppy, and even more distinctly in Languedoc, where cacaracá means both the crowing and the flower. The hen in some languages has a name corresponding to that of the cock, as in Kussa kukuduna 'cock,' kukukasi 'hen;' Ewe koklo-tsu 'cock,' koklo-no 'hen;' and her cackle (whence she has in Switzerland the name of gugel, güggel) has passed into language as a term for idle gossip and chatter of women, caquet, caqueter, gackern, much as the noise of a very different creature seems to have given rise not only to its name, Italian cicala, but to a group of words represented by cicalar 'to chirp, chatter, talk sillily.' The pigeon is a good example of this kind, both for sound and sense. It is Latin pipio, Italian pippione, piccione, pigione, modern Greek πιπίνιον, French pipion (old), pigeon; its derivation is from the young bird's peep, Latin pipire, Italian pipiare, pigiolare, modern Greek πιπνίζω, to chirp; by an easy metaphor, a pigeon comes to mean 'a silly young fellow easily caught,' to pigeon 'to cheat,' Italian pipione 'a silly gull, one that is soon caught and trepanned,' pippionare 'to pigeon, to gull one.' In an entirely different family of languages, Mr. Wedgwood points out a curiously similar process of derivation; Magyar pipegni, pipelni 'to peep or cheep;' pipe, pipök 'a chicken, gosling;' pipe-ember (chicken-man), 'a silly young fellow, booby.'^[6] The derivation of Greek (Greek characters), Latin bos, Welsh bu, from the ox's lowing, or booing as it is called in the north country, has been much debated. With an excessive desire to make Sanskrit answer as a general Indo-European type, Bopp connected Sanskrit go, old German chuo, English cow, with these words, on the unusual and forced assumption of a change from guttural to labial.^[7] The direct derivation from sound, however, is favoured by other languages, Cochin-Chinese bo, Hottentot bou. The beast may almost answer for himself in the words of that Spanish proverb which remarks that people talk according to their nature: 'Habló el buey, y dijó bu!' 'The ox spoke, and he said boo!'

Among musical instruments with imitative names are the following: — the shee-shee-quoi, the mystic rattle of the Red Indian medicine-man, an imitative word which reappears in the Darien Indian shak-shak, the shook-shook of the Arawaks, the Chinook shugh (whence shugh-opoots, rattletail, i.e., 'rattlesnake;') — the drum, called ganga in Haussa, gañgañ in the Yoruba country, gunguma by the Gallas, and having its analogue in the Eastern gong; — the bell, called in Yakama (N. Amer.) kwa-lal-kwa-lal, in Yalof (W. Afr.) walwal, in Russian kolokol. The sound of the horn is imitated in English nurseries as toot-toot, and this is transferred to express the 'omnibus' of which the bugle is the signal: with this nursery word is to be classed the Peruvian name for the 'shell-trumpet,' pututu, and the Gothic thuthaurn (thut-horn), which is even used in the Gothic Bible for the last trumpet of the day of judgement, — 'In spêdistin thuthaúrna. thuthaúrneith auk jah daúthans ustandand' (1 Cor. xv. 52). How such imitative words, when thoroughly taken up into language, suffer change of pronunciation in which the original sound-meaning is lost, may be seen in the English word tabor, which we might not recognize as a sound-word at all, did we not notice that it is French tabour, a word which in the form tambour obviously belongs to a group of words for drums, extending from the small rattling Arabic tubl to the Indian dundhubi and the tombe, the Moqui drum made of a hollowed log. The same group shows the transfer of such imitative words to objects which are like the instrument, but have nothing to do with its sound; few people who talk of tambour-work, and fewer still who speak of a footstool as a tabouret, associate these words with the sound of a drum, yet the connexion is clear enough. When these two processes go on together, and a sound-word changes its original sound on the one hand, and transfers its meaning to something else on the other, the result may soon leave philological analysis quite helpless, unless by accident historical evidence is forthcoming. Thus with the English word pipe. Putting aside the particular pronunciation which we give the word, and referring it back to its mediæval Latin or French sound in pipa, pipe, we have before us an evident imitative name of a musical instrument, derived from a familiar sound used also to represent the chirping of chickens, Latin pipire, English to peep, as in the translation of Isaiah viii. 19: 'Seek ... unto wizards that peep, and that mutter.' The Algonquin Indians appear to have formed from this sound pib (with a grammatical suffix) their name for the pib-e-gwun or native flute. Now just as tuba, tubus, 'a trumpet' (itself very likely an imitative word) has given a name for any kind of tube, so the word pipe has been transferred from the musical instrument to which it first belonged, and is used to describe tubes of various sorts, gas-pipes, water-pipes, and pipes in general. There is nothing unusual in these transitions of meaning, which are in fact rather the rule than the exception. The chibouk was originally a herdsman's pipe or flute in Central Asia. The calumet, popularly ranked with the tomahawk and the mocassin among characteristic Red Indian words, is only the name for a shepherd's pipe (Latin calamus) in the dialect of Normandy, corresponding with the chalumeau of literary French; for when the early colonists in Canada saw the Indians performing the strange operation of smoking, 'with a hollow piece of stone or wood like a pipe,' as Jacques Cartier has it, they merely gave to the native tobacco-pipe the name of the French musical instrument it resembled. Now changes of sound and of sense like this of the English word pipe must have been in continual operation in hundreds of languages where we have no evidence to follow them by, and where we probably may never obtain such evidence. But what little we do know must compel us to do justice to the imitation of sound as a really existing process, capable of furnishing an indefinitely large supply of words for things and actions which have no necessary connexion at all with that sound. Where the traces of the transfer are lost, the result is a stock of words which are the despair of philologists, but are perhaps none the less fitted for the practical use of men who simply want recognized symbols for recognized ideas.

The claim of the Eastern tomtom to have its name from a mere imitation of its sound seems an indisputable one; but when it is noticed in what various languages the beating of a resounding object is expressed by something like tum, tumb, tump, tup, as in Javan tumbuk, Coptic tmno, 'to pound in a mortar,' it becomes evident that the admission involves more than at first sight appears. In Malay, timpa, tampa, is 'to beat out, hammer, forge;' in the Chinook Jargon tum-tum is 'the heart,' and by combining the same sound with the English word 'water,' a name is made for 'waterfall,' tum-wâta. The Gallas of East Africa declare that a box on the ear seems to them to make a noise like tub, for they call its sound tubdjeda, that is, 'to say tub.' In the same language, tuma is 'to beat,' whence tumtu, 'a workman, especially one who beats, a smith.' With the aid of another imitative word, bufa 'to blow,' the Gallas can construct this wholly imitative sentence, tumtun bufa bufti, 'the smith blows with bellows,' as an English child might say, 'the tumtum puffs the puffer.' This imitative sound seems to have obtained a footing among the Aryan verb-roots, as in Sanskrit tup, tubh 'to smite,' while in Greek, tup, tump, has the meaning of 'to beat, to thump,' producing for instance (Greek characters), tympanum, 'a drum or tomtom.' Again, the verb to crack has become in modern English as thorough a root-word as the language possesses. The mere imitation of the sound of breaking has passed into a verb to break; we speak of a cracked cup or a cracked reputation without a thought of imitation of sound; but we cannot yet use the German krachen or French craquer in this way, for they have not developed in meaning as our word has, but remain in their purely imitative stage. There are two corresponding Sanskrit words for the saw, kra-kara, kra-kacha, that is to say, the ‘kra-maker, kra-crier;' and it is to be observed that all such terms, which expressly state that they are imitations of sound, are particularly valuable evidence in these enquiries, for whatever doubt there may be as to other words being really derived from imitative sound, there can, of course, be none here. Moreover, there is evidence of the same sound having given rise to imitative words in other families of language, Dahoman kra-kra, 'a watchman's rattle;' Grebo grikâ 'a saw;' Aino chacha to saw;' Malay graji 'a saw,' karat 'to gnash the teeth,' karot 'to make a grating noise;' Coptic khrij 'to gnash the teeth,' khrajrej 'to grate.' Another form of the imitation is given in the descriptive Galla expression cacakdjeda, i.e., 'to say cacak,' 'to crack, krachen.' With this sound corresponds a whole family of Peruvian words, of which the root seems to be the guttural cca, coming from far back in the throat; ccallani, 'to break,' ccatatani, 'to gnash the teeth,' ccacñiy, 'thunder,' and the expressive words for 'a thunder-storm,' ccaccaccahay, which carries the imitative process so much farther than such European words as thunder-clap, donner-klapf. In Maori, pata is 'to patter as water dropping, drops of rain.' The Manchu language describes the noise of fruits falling from the trees as pata pata (so Hindustani bhadbhad); this is like our word pat, and we should say in the same manner that the fruit comes pattering down, while French patatra is a recognized imitation of something falling. Coptic potpt is 'to fall,' and the Australian badbadin (or patpatin) is translated into almost literal English as pitpatting. On the strength of such non-Aryan languages, are we to assign an imitative origin to the Sanskrit verb-root pat, 'to fall,' and to Greek (Greek characters)?

Wishing rather to gain a clear survey of the principles of language-making than to plunge into obscure problems, it is not necessary for me to discuss here questions of intricate detail. The point which continually arises is this, — granted that a particular kind of transition from sound to sense is possible in the abstract, may it be safely claimed in a particular case? In looking through the vocabularies of the world, it appears that most languages offer words which, by obvious likeliness or by their correspondence with similar forms elsewhere, may put forward a tolerable claim to be considered imitative. Some languages, as Aztec or Mohawk, offer singularly few examples, while in others they are much more numerous. Take Australian cases: walle, 'to wail;' bung-bung-ween, 'thunder;' wirriti, 'to blow, as wind;' wirrirriti, 'to storm, rage, as in fight;' wirri, bwirri, 'the native throwing club,' seemingly so called from its whir through the air; kurarriti, 'to hum, buzz;' kurrirrurriri, 'round about, unintelligible,' &c.; pitata, 'to knock, pelt, as rain,' pitapitata, 'to knock;' wiiti, 'to laugh, rejoice' as in our own 'Turnament of Tottenham' : —

'"We te he!" quoth Tyb, and lugh, "Ye er a dughty man!"'

The so-called Chinook Jargon of British Columbia is a language crowded with imitative words, sometimes adopted from the native Indian languages, sometimes made on the spot by the combined efforts of the white man and the Indian to make one another understand. Samples of its quality are hóh-hoh, 'to cough,' kó-ko, 'to knock,' kwa-lal-kwa-lal, 'to gallop,' muck-a-muck, 'to eat,' chak-chak, 'the bald eagle' (from its scream), mamook tsish (make tsish), 'to sharpen on the grindstone.' It has been remarked by Prof. Max Müller that the peculiar sound made in blowing out a candle is not a favourite in civilized languages, but it seems to be recognized here, for no doubt it is what the compiler of the vocabulary is doing his best to write down when he gives mamook poh (make poh) as the Chinook expression for 'to blow out or extinguish as a candle.' This jargon is in great measure of new growth within the last seventy or eighty years, but its imitative words do not differ in nature from those of the more ordinary and old-established languages of the world. Thus among Brazilian tribes there appear Tupi cororóng, cururuc, 'to snore' (compare Coptic kherkher, Quichua ccorcuni (ccor)), whence it appears that an imitation of a snore may perhaps serve the Carajas Indians to express 'to sleep' as arourou-cré, as well as the related idea of 'night,' roou. Again Pimenteira ebaung, 'to bruise, beat,' compares with Yoruba gba, 'to slap,' gbã (gbang) 'to sound loudly, to bang,' and so forth. Among African languages, the Zulu seems particularly rich in imitative words. Thus bibiza, 'to dribble like children, drivel in speaking' (compare English bib); babala, 'the larger bush-antelope' (from the baa of the female); boba, 'to babble, chatter, be noisy,' bobi, 'a babbler;' boboni, 'a throstle' (cries bo! bo! compare American bobolink); bomboloza, 'to rumble in the bowels, to have a bowel-complaint;' bubula, 'to buzz like bees,' bubulela, 'a swarm of bees, a buzzing crowd of people;' bubuluza, 'to make a blustering noise, like frothing beer or boiling fat.' These examples, from among those given under one initial letter in one dictionary of one barbaric language, may give an idea of the amount of the evidence from the languages of the lower races bearing on the present problem.

For the present purpose of giving a brief series of examples of the sort of words in which imitative sound seems fairly traceable, the strongest and most manageable evidence is of course found among such words as directly describe sounds or what produces them, such as cries of and names for animals, the terms for action accompanied by sound, and the materials and objects so acted upon. In further investigation it becomes more and more requisite to isolate the sound-type or root from the modifications and additions to which it has been subjected for grammatical and phonetical adaptation. It will serve to give an idea of the extent and intricacy of this problem, to glance at a group of words in one European language, and notice the etymological network which spreads round the German word klapf, in Grimm's dictionary, klappen, klippen, klopfen, kläffen, klimpern, klampern, klateren, kloteren, klitteren, klatzen, klacken, and more, to be matched with allied forms in other languages. Setting aside the consideration of grammatical inflexion, it belongs to the present subject to notice that man's imitative faculty in language is by no means limited to making direct copies of sound and shaping them into words. It seizes upon ready-made terms of whatever origin, alters and adapts them to make their sound fitting to their sense, and pours into the dictionaries a flood of adapted words of which the most difficult to analyse are those which are neither altogether etymological nor altogether imitative, but partly both. How words, while preserving, so to speak, the same skeleton, may be made to follow the variation of sound, of force, of duration, of size, an imitative group more or less connected with the last will show — crick, creak, crack, crash, crush, crunch, craunch, scrunch, scraunch. It does not at all follow that because a word suffers such imitative and symbolic changes it must be, like this, directly imitative in its origin. What, for instance, could sound more imitative than the name of that old-fashioned cannon for throwing grape-shot, the patterero? Yet the etymology of the word appears in the Spanish form pedrero, French perrier; it means simply an instrument for throwing stones (piedra, pierre), and it was only when the Spanish word was adopted in England that the imitative faculty caught and transformed it into an apparent sound-word, resembling the verb to patter. The propensity of language, especially in slang, to make sense of strange words by altering them into something with an appropriate meaning has been often dwelt upon by philologists, but the propensity to alter words into something with an appropriate sound has produced results immensely more important. The effects of symbolic change of sound acting upon verb-roots seem almost boundless. The verb to waddle has a strong imitative appearance, and so in German we can hardly resist the suggestion that imitative sound has to do with the difference between wandern and wandeln; but all these verbs belong to a family represented by Sanskrit vad, to go, Latin vado, and to this root there seems no sufficient ground for assigning an imitative origin, the traces of which it has at any rate lost if it ever had them. Thus, again, to stamp with the foot, which has been claimed as an imitation of sound, seems only a 'coloured' word. The root sta, 'to stand,' Sanskrit sthâ, forms a causative stap, Sanskrit stâdpay, 'to make to stand,' English to stop, and a foot-step is when the foot comes to a stand, a foot-stop. But we have Anglo-Saxon stapan, stæpan, steppan, English to step, varying to express its meaning by sound in to staup, to stamp, to stump, and to stomp, contrasting in their violence or clumsy weight with the foot on the Dorset cottage-sill in Barnes's poem: —

'Where love do seek the maïden's evenèn vloor, Wi' stip-step light, an tip-tap slight Ageän the door.'

By expanding, modifying, or, so to speak, colouring, sound is able to produce effects closely like those of gesture-language, expressing length or shortness of time, strength or weakness of action, and then passing into a further stage to describe greatness or smallness of size or of distance, and thence making its way into the widest fields of metaphor. And it does all this with a force which is surprising when we consider how childishly simple are the means employed. Thus the Bachapin of Africa call a man with the cry héla! but according as he is far or farther off the sound of the hêela! hê-ê-la! is lengthened out. Mr. Macgregor in his 'Rob Roy on the Jordan,' graphically describes this method of expression, '"But where is Zalmouda?" ... Then with rough eagerness the strongest of the Dowana faction pushes his long forefinger forward, pointing straight enough — but whither? and with a volley of words ends, Ah-ah-a-a-a——a-a. This strange expression had long before puzzled me when first heard from a shepherd in Bashan. ... But the simple meaning of this long string of "ah's" shortened, and quickened, and lowered in tone to the end, is merely that the place pointed to is a "very great way off."' The Chinook Jargon, as usual representing primitive developments of language, uses a similar device in lengthening the sound of words to indicate distance. The Siamese can, by varying the tone-accent, make the syllable non, 'there,' express a near, indefinite, or far distance, and in like manner can modify the meaning of such a word as ny, 'little.' In the Gaboon, the strength with which such a word as mpolu, 'great,' is uttered serves to show whether it is great, very great, or very very great, and in this way, as Mr. Wilson remarks in his Mpongwe Grammar, 'the comparative degrees of greatness, smallness, hardness, rapidity, and strength, &c., may be conveyed with more accuracy and precision than could readily be conceived.' In Madagascar ratchi means 'bad,' but râtchi is 'very bad.' The natives of Australia, according to Oldfield, show the use of this process in combination with that of symbolic reduplication: among the Watchandie tribe jir-rie signifies 'already or past,' jir-rie jir-rie indicates 'a long time ago,' while jie-r-rie jirrie (the first syllable being dwelt on for some time) signifies 'an immense time ago.' Again, boo-rie is 'small,' boo-rie-boo-rie 'very small,' and b-o-rie boorie 'exceedingly small.' Wilhelm von Humboldt notices the habit of the southern Guarani dialect of South America of dwelling more or less time on the suffix of the perfect tense, yma, y—ma, to indicate the length or shortness of the distance of time at which the action took place; and it is curious to observe that a similar contrivance is made use of among the aboriginal tribes of India, where the Ho language forms a future tense by adding á to the root, and prolonging its sound, kajee 'to speak,' Amg kajēēá 'I will speak.' As might be expected, the languages of very rude tribes show extremely well how the results of such primitive processes pass into the recognized stock of language. Nothing could be better for this than the words by which one of the rudest of living races, the Botocudos of Brazil, express the sea. They have a word for a stream, ouatou, and an adjective which means great, ijipakijiou; thence the two words 'stream-great,' a little strengthened in the vowels, will give the term for a river, ouatou-ijiipakiiijou, as it were, 'stream-grea-at,' and this, to express the immensity of the ocean, is amplified into ouatou-iijipakiijou-ou-ou-ou-ou-ou. Another tribe of the same family works out the same result more simply; the word ouatou, 'stream/ becomes ouatou-ou-ou-ou, 'the sea.' The Chavantes very naturally stretch the expression rom-o-wodi, 'I go a long way,' into rom-o-o-o-o-wodi, ' I go a very long way indeed,' and when they are called upon to count beyond five they say it is ka-o-o-oki, by which they evidently mean it is a very great many. The Cauixanas in one vocabulary are described as saying lawauugabi for four, and drawling out the same word for five, as if to say 'a long four,' in somewhat the same way as the Aponegicrans, whose word for six is itawuna, can expand this into a word for seven, itawuūna, obviously thus meaning a 'long six.' In their earlier and simpler stages nothing can be more easy to comprehend than these, so to speak, pictorial modifications of words. It is true that writing, even with the aid of italics and capitals, ignores much of this symbolism in spoken language, but every child can see its use and meaning, in spite of the efforts of book-learning and school-teaching to set aside whatever cannot be expressed by their imperfect symbols, nor controlled by their narrow rules. But when we try to follow out to their full results these methods, at first so easy to trace and appreciate, we soon find them passing out of our grasp. The language of the Sahaptin Indians shows us a process of modifying words which is far from clear, and yet not utterly obscure. These Indians have a way of making a kind of disrespectful diminutive by changing the n in a word to l; thus twinwt means 'tailless,' but to indicate particular smallness, or to express contempt, they make this into twilwt, pronounced with an appropriate change of tone; and again, wana means 'river,' but this is made into a diminutive wala by 'changing n into l, giving the voice a different tone, putting the lips out in speaking, and keeping them suspended around the jaw.' Here we are told enough about the change of pronunciation to guess at least how it could convey the notions of smallness and contempt. But it is less easy to follow the process by which the Mpongwe language turns an affirmative into a negative verb by 'an intonation upon, or prolongation of the radical vowel,' tŏnda, to love, tŏ̱nda, not to love; tŏndo, to be loved, tŏ̱ndo, not to be loved. So Yoruba, bába, 'a great thing,' bàba, 'a small thing,' contrasted in a proverb, 'Baba bo, baba molle' — 'A great matter puts a smaller out of sight.' Language is, in fact, full of phonetic modifications which justify a suspicion that symbolic sound had to do with their production, though it may be hard to say exactly how.

Again, there is the familiar process of reduplication, simple or modified, which produces such forms as murmur, pitpat, helterskelter. This action, though much restricted in literary dialects, has such immense scope in the talk of children and savages that Professor Pott's treatise on it^[8] has become incidentally one of the most valuable collections of facts ever made with relation to early stages of language. Now up to a certain point any child can see how and why such doubling is done, and how it always adds something to the original idea. It may make superlatives or otherwise intensify words, as in Polynesia loa 'long,' lololoa 'very long'; Mandingo ding 'a child,' dingding 'a very little child.' It makes plurals, as Malay raja-raja 'princes,' orang-orang 'people.' It adds numerals, as Mosquito walwal 'four' (two-two), or distributes them, as Coptic ouai ouai 'singly' (one-one). These are cases where the motive of doubling is comparatively easy to make out. As an example of cases much more difficult to comprehend may be taken the familiar reduplication of the perfect tense, Greek γέγραφα from γράφω, Latin momordi from mordeo, Gothic haihald from haldan, 'to hold.' Reduplication is habitually used in imitative words to intensify them, and still more, to show that the sound is repeated or continuous. From, the immense mass of such words we may take as instances the Botocudo hou-hou-hou-gitcha 'to suck' (compare Tongan hūhū 'breast'), kiaku-käck-käck, 'a butterfly'; Quichua chiuiuiuiñichi 'wind whistling in the trees'; Maori haruru 'noise of wind'; hohoro 'hurry'; Dayak kakakkaka 'to go on laughing loud'; Aino shiriushiriukanni 'a rasp'; Tamil murumuru 'to murmur'; Akra ewiewiewiewie 'he spoke repeatedly and continually'; and so on, throughout the whole range of the languages of the world.

The device of conveying different ideas of distance by the use of a graduated scale of vowels seems to me one of great philological interest, from the suggestive hint it gives of the proceedings of the language-makers in most distant regions of the world, working out in various ways a similar ingenious contrivance of expression by sound. A typical series is the Javan: iki 'this' (close by); ika 'that' (at some distance); iku 'that' (farther off). It is not likely that the following list nearly exhausts the whole number of cases in the languages of the world, for about half the number have been incidentally noted down by myself without any especial search, but merely in the course of looking over vocabularies of the lower races.^[9]

Javan . . . iki, this; ika, that (intermediate); iku, that.

Malagasy . . ao, there (at a short distance); eo, there (at a shorter distance); io, there (close at hand).

atsy, there (not far off); etsy, there (nearer); itsy, this or these.

Japanese . . ko, here; ka, there.

korera, these; karera, they (those).

Canarese . . ivanu, this; uvanu, that (intermediate); avanu, that.

Tamul . . î, this; â, that.

Rajmahali . . îh, this; âh, that.

Dhimal . . . isho, ita, here; usho, uta, there.

iti, idong, this; uti, udong, that [of things and persons respectively].

Abchasian . . abri, this; ubri, that.

Ossetic . . . am, here; um, there.

Magyar . . . ez, this; az, that.

Zulu . . . apa, here; apo, there.

lesi, leso, lesiya; abu, abo, abuya; &c.= this, that, that (in the distance). Yoruba . . . na, this; ni, that. Fernandian . . olo, this; ole, that. Tumale . . . re, this; ri, that. ngi, I; ngo, thou; ngu, he.

Greenlandish . . uv, here, there (where one points to) ; iv, there, up there [found in comp.]. Sujelpa (Coleville Ind.), a-^a, this ; t^t, that. Sahaptin . . kina, here ; n<z, there. Mutsun . . . ne, here ; nu, there. Tarahumara . . ibe, here ; abe, there. Guarani . . . tide, ne, thou ; ndi, ni, he. Botocudo . . ati, I ; oti, thou, you, (prep.) to. Carib . . . ne, thou ; ni, he. Chilian . . . tva, vacbi, this ; tvey, veycbi, that.

It is obvious on inspection of this list of pronouns and adverbs that they have in some way come to have their vowels contrasted to match the contrast of here and there, this and that. Accident may sometimes account for such cases. For instance it is well known to philologists that our own this and that are pronouns partly distinct in their formation, thi-s being probably two pronouns run together, but yet the Dutch neuters dit 'this,' and dat 'that,' have taken the appearance of a single form with contrasted vowels.^[10] But accident cannot account for the frequency of such words in pairs, and even in sets of three, in so many different languages. There must have been some common intention at work, and there is evidence that some of these languages do resort to a change of sound as a means of expressing change of distance. Thus the language of Fernando Po can not only express 'this' and 'that' by olo, ole, but it can even make a change of the pronunciation of the vowel distinguish between o boehe 'this month,' and oh boehe, 'that month.' In the same way the Grebo can make the difference between 'I' and 'thou,' 'we,' and 'you,' 'solely by the intonation of the voice, which the final h of the second persons mâh and ăh is intended to express.'

mâ di, I eat; mâh di, thou eatest; ă di, we eat; ăh di, ye eat.

The set of Zulu demonstratives which express the three distances of near, farther, farthest, are very complex, but a remark as to their use shows how thoroughly symbolic sound enters into their nature. The Zulus not only say nansi, 'here is,' nanso, 'there is,' nansiya, 'there is in the distance,' but they even express the greatness of this distance by the emphasis and prolongation of the ya. If we could discern a similar gradation of the vowels to express a corresponding gradation of distance throughout our list, the whole matter would be easier to explain; but it is not so, the i-words for instance, are sometimes nearer and sometimes farther off than the a-words. We can only judge that, as even children can see that a scale of vowels makes a most expressive scale of distances, many pronouns and adverbs in use in the world have probably taken their shape under the influence of this simple device, and thus there have arisen sets of what we may call contrasted or 'differential' words.

How the differencing of words by change of vowels may be used to distinguish between the sexes, is well put in a remark of Professor Max Müller's: 'The distinction of gender ... is sometimes expressed in such a manner that we can only explain it by ascribing an expressive power to the more or less obscure sound of vowels. Ukko, in Finnic, is an old man; akka, an old woman. ... In Mandshu chacha is mas. ... cheche, femina. Again, ama, in Mandshu, is father; eme, mother; amcha, father-in-law, emche, mother-in-law.'^[11] The Coretú language of Brazil has another curiously contrasted pair of words tsáackö, 'father,' tsaacko 'mother,' while the Carib has baba for father, and bibi for mother, and the Ibu of Africa has nna for father and nne for mother. This contrivance of distinguishing the male from the female by a difference of vowels is however but a small part of the process of formation which can be traced among such words as those for father and mother. Their consideration leads into a very interesting philological region, that of 'Children's Language.'

If we set down a few of the pairs of words which stand for 'father' and 'mother' in very different and distant languages — papa and mama; Welsh, tad (dad) and mam; Hungarian, atya and anya; Mandingo, fa and ba; Lummi (N. America), man and tan; Catoquina (S. America), payú and nayú; Watchandie (Australia), amo and ago — their contrast seems to lie in their consonants, while many other pairs differ totally, like Hebrew ab and im; Kuki, p'ha and noo; Kayan, amay and inei; Tarahumara, nono and jeje. Words of the class of papa and mama, occurring in remote parts of the world, were once freely used as evidence of a common origin of the languages in which they were found alike. But Professor Buschmann's paper on 'Nature-Sound,' published in 1853,^[12] effectually overthrew this argument, and settled the view that such coincidence might arise again and again by independent production. It was clearly of no use to argue that Carib and English were allied because the word papa, 'father,' belongs to both, or Hottentot and English because both use mama for 'mother,' seeing that these childish articulations may be used in just the opposite way, for the Chilian word for mother is papa, and the Tlatskanai for father is mama. Yet the choice of easy little words for 'father' and 'mother' does not seem to have been quite indiscriminate. The immense list of such words collected by Buschmann shows that the types pa and ta, with the similar forms ap and at, preponderate in the world as names for 'father,' while ma and na, am and an, preponderate as names for 'mother.' His explanation of this state of things as affected by direct symbolism choosing the hard sound for the father, and the gentler for the mother, has very likely truth in it, but it must not be pushed too far. It cannot be, for instance, the same principle of symbolism which leads the Welshmen to say tad for 'father' and mam for 'mother,' and the Indian of British Columbia to say maan, 'father' and taan, ' mother,' or the Georgian to say mama 'father' and deda 'mother.' Yet I have not succeeded in finding anywhere our familiar papa and mama exactly reversed in one and the same language; the nearest approach to it that I can give is from the island of Meang, where mama meant 'father, man,' and babi, 'mother, woman.'^[13]

Between the nursery words papa and mama and the more formal father and mother there is an obvious resemblance in sound. What, then, is the origin of these words father and mother? Up to a certain point their history is clear. They belong to the same group of organized words with vater and mutter, pater and mater, πατήρ and μήτηρ, pitar and mâtar, and other similar forms through the Indo-European family of languages. There is no doubt that all these pairs of names are derived from an ancient and common Aryan source, and when they are traced back as far as possible towards that source, they appear to have sprung from a pair of words which may be roughly called patar and matar, and which were formed by adding tar, the suffix of the actor, to the verb-roots pa and ma. There being two appropriate Sanskrit verbs pâ and mâ, it is possible to etymologize the two words as patar, 'protector,' and matar, 'producer.' Now this pair of Aryan words must have been very ancient, lying back at the remote common source from which forms parallel to our English father and mother passed into Greek and Persian, Norse and Armenian, thus holding fixed type, through the eventful course of Indo-European history. Yet, ancient as these words are, they were no doubt preceded by simpler rudimentary words of the children's language, for it is not likely that the primitive Aryans did without baby-words for father and mother until they had an organized system of adding suffixes to verb-roots to express such notions as 'protector' or 'producer.' Nor can it be supposed that it was by mere accident that the root-words thus chosen happened to be the very sounds pa and ma, whose types so often occur in the remotest parts of the world as names for 'father' and 'mother.' Prof. Adolphe Pictet makes shift to account for the coincidence thus: he postulates first the pair of forms pâ and mâ as Aryan verb-roots of unknown origin, meaning 'to protect' and 'to create,' next another pair of forms pa and ma, children's words commonly used to denote father and mother, and lastly he combines the two by supposing that the root-verbs pâ and mâ were chosen to form the Indo-European words for parents, because of their resemblance to the familiar baby-words already in use. This circuitous process at any rate saves those sacred monosyllables, the Sanskrit verb-roots, from the disgrace of an assignable origin. Yet those who remember that these verb-roots are only a set of crude forms in use in one particular language of the world at one particular period of its development, may account for the facts more simply and more thoroughly. It is a fair guess that the ubiquitous pa and ma of the children's language were the original forms; that they were used in an early period of Aryan speech as indiscriminately substantive and verb, just as our modern English, which so often reproduces the most rudimentary linguistic processes, can form from the noun 'father' a verb 'to father;' and that lastly they became verb-roots, whence the words patar and matar were formed by the addition of the suffix.^[14]

The baby-names for parents must not be studied as though they stood alone in language. They are only important members of a great class of words, belonging to all times and countries within our experience, and forming a children's language, whose common character is due to its concerning itself with the limited set of ideas in which little children are interested, and expressing these ideas by the limited set of articulations suited to the child's first attempts to talk. This peculiar language is marked quite characteristically among the low savage tribes of Australia; mamman 'father,' ngangan 'mother,' and by metaphor 'thumb,' 'great toe' (as is more fully explained in jinnamamman 'great toe,' i.e. foot's father), tammin 'grandfather or grandmother,' bab-ba 'bad, foolish, childish,' bee-bee, beep 'breast,' pappi 'father,' pappa 'young one, pup, whelp,' (whence is grammatically formed the verb papparniti 'to become a young one, to be born.' Or if we look for examples from India, it does not matter whether we take them from non-Hindu or Hindu languages, for in baby-language all races are on one footing. Thus Tamil appâ 'father,' ammâ 'mother,' Bodo aphâ 'father,' âyâ 'mother;' the Kocch group nânâ and nâni 'paternal grandfather and grandmother,' mâmâ 'uncle,' dâdâ 'cousin,' may be set beside Sanskrit tata 'father,' nanâ 'mother,' and the Hindustani words of the same class, of which some are familiar to the English ear by being naturalized in Anglo-Indian talk, bâbâ 'father,' bâbû 'child, prince, Mr.,' bîbî 'lady,' dadâ 'nurse' (âyâ 'nurse' seems borrowed from Portuguese). Such words are continually coming fresh into existence everywhere, and the law of natural selection determines their fate. The great mass of the nana's and dada's of the nursery die out almost as soon as made. Some few take more root and spread over large districts as accepted nursery words, and now and then a curious philologist makes a collection of them. Of such, many are obvious mutilations of longer words, as French faire dodo 'to sleep' (dormir), Brandenburg wiwi, a common cradle lullaby (wiegen). Others, whatever their origin, fall, in consequence of the small variety of articulations out of which they must be chosen, into a curiously indiscriminate and unmeaning mass, as Swiss bobo 'a scratch;' bambam 'all gone;' Italian bobò 'something to drink,' gogo 'little boy,' for dede 'to play.' These are words quoted by Pott, and for English examples nana 'nurse,' tata! 'good-bye!' may serve. But all baby-words, as this very name proves, do not stop short even at this stage of publicity. A small proportion of them establish themselves in the ordinary talk of grown-up men and women, and when they have once made good their place as constituents of general language, they may pass on by inheritance from age to age. Such examples as have been here quoted of nursery words give a clue to the origin of a mass of names in the most diverse languages, for father, mother, grandmother, aunt, child, breast, toy, doll, &c. The negro of Fernando Po who uses the word bubboh for 'a little boy,' is on equal terms with the German who uses bube; the Congo-man who uses tata for 'father' would understand how the same word could be used in classic Latin for 'father,' and in mediæval Latin for 'pedagogue;' the Carib and the Caroline Islander agree with the Englishman that papa is a suitable word to express 'father,' and then it only remains to carry on the word, and make the baby-language name the priests of the Eastern Church and the great Papa of the Western. At the same time the evidence explains the indifference with which, out of the small stock of available materials, the same sound does duty for the most different ideas; why mama means here 'mother,' there 'father,' there 'uncle,' maman here 'mother,' there 'father-in-law,' dada here 'father,' there 'nurse,' there 'breast,' tata here 'father,' there 'son.' A single group of words may serve to show the character of this peculiar region of language: Blackfoot Indian ninnah 'father;' Greek νέννος 'uncle,' νέννα 'aunt;' Zulu nina, Sangir nina, Malagasy nini 'mother;' Javan nini 'grandfather or grandmother;' Vayu nini 'paternal aunt;' Darien Indian ninah 'daughter;' Spanish niño, niña 'child;' Italian ninna 'little girl;' Milanese ninin 'bed;' Italian ninnare 'to rock the cradle.'

In this way a dozen easy child's articulations, ba's and na's, ti's and de's, pa's and ma's, serve almost as indiscriminately to express a dozen child's ideas as though they had been shaken in a bag and pulled out at random to express the notion that came first, doll or uncle, nurse or grandfather. It is obvious that among words cramped to such scanty choice of articulate sounds, speculations as to derivation must be more than usually unsafe. Looked at from this point of view, children's language may give a valuable lesson to the philologist. He has before him a kind of language, formed, under peculiar conditions, and showing the weak points of his method of philological research, only exaggerated into extraordinary distinctness. In ordinary language, the difficulty of connecting sound with sense lies in great measure in the inability of a small and rigid set of articulations to express an interminable variety of tones and noises. In children's language, a still more scanty set of articulations fails yet more to render these distinctly. The difficulty of finding the derivation of words lies in great measure in the use of more or less similar root-sounds for most heterogeneous purposes. To assume that two words of different meanings, just because they sound somewhat alike, must therefore have a common origin, is even in ordinary language the great source of bad etymology. But in children's language the theory of root-sounds fairly breaks down. Few would venture to assert, for instance, that papa and pap have a common derivation or a common root. All that we can safely say of connexion between them is that they are words related by common acceptance in the nursery language. As such, they are well marked in ancient Rome as in modern England: papas 'nutricius, nutritor,' pappus 'senex;' 'cum cibum et potum buas ac papas dicunt, et matrem mammam, patrem tatam (or papam).'^[15]

From children's language, moreover, we have striking proof of the power of consensus of society, in establishing words in settled use without their carrying traces of inherent expressiveness. It is true that children are intimately acquainted with the use of emotional and imitative sound, and their vocal intercourse largely consists of such expression. The effects of this are in some degree discernible in the class of words we are considering. But it is obvious that the leading principle of their formation is not to adopt words distinguished by the expressive character of their sound, but to choose somehow a fixed word to answer a given purpose. To do this, different languages have chosen similar articulations to express the most diverse and opposite ideas. Now in the language of grown-up people, it is clear that social consensus has worked in the same way. Even if the extreme supposition be granted, that the ultimate origin of every word of language lies in inherently expressive sound, this only partly affects the case, for it would have to be admitted that, in actual languages, most words have so far departed in sound or sense from this originally expressive stage, that to all intents and purposes they might at first have been arbitrarily chosen. The main principle of language has been, not to preserve traces of original sound-signification for the benefit of future etymologists, but to fix elements of language to serve as counters for practical reckoning of ideas. In this process much original expressiveness has no doubt disappeared beyond all hope of recovery.

Such are some of the ways in which vocal sounds seem to have commended themselves to the mind of the word-maker as fit to express his meaning, and to have been used accordingly. I do not think that the evidence here adduced justifies the setting-up of what is called the Interjectional and Imitative Theory as a complete solution of the problem of original language. Valid as this theory proves itself within limits, it would be incautious to accept a hypothesis which can perhaps satisfactorily account for a twentieth of the crude forms in any language, as a certain and absolute explanation of the nineteen-twentieths whose origin remains doubtful. A key must unlock more doors than this, to be taken as the master-key. Moreover, some special points which have come under consideration in these chapters tend to show the positive necessity of such caution in theorizing. Too narrow a theory of the application of sound to sense may fail to include the varied devices which the languages of different regions turn to account. It is thus with the distinction in meaning of a word by its musical accent, and the distinction of distance by graduated vowels. These are ingenious and intelligible contrivances, but they hardly seem directly emotional or imitative in origin. A safer way of putting the theory of a natural origin of language is to postulate the original utterance of ideas in what may be called self-expressive sounds, without defining closely whether their expression lay in emotional tone, imitative noise, contrast of accent or vowel or consonant, or other phonetic quality. Even here, exception of unknown and perhaps enormous extent must be made for sounds chosen by individuals to express some notion, from motives which even their own minds failed to discern, but which sounds nevertheless made good their footing in the language of the family, the tribe, and the nation. There may be many modes even of recognizable phonetic expression, unknown to us as yet. So far, however, as I have been able to trace them here, such modes have in common a claim to belong not exclusively to the scheme of this or that particular dialect, but to wide-ranging principles of formation of language. Their examples are to be drawn with equal cogency from Sanskrit or Hebrew, from the nursery-language of Lombardy, or the half-Indian, half-European jargon of Vancouver's Island; and wherever they are found, they help to furnish groups of sound-words — words which have not lost the traces of their first expressive origin, but still carry their direct significance plainly stamped upon them. In fact, the time has now come for a substantial basis to be laid for Generative Philology. A classified collection of words with any strong claim to be self-expressive should be brought together out of the thousand or so of recognized languages and dialects of the world. In such a Dictionary of Sound-Words, half the cases cited might very likely be worthless, but the collection would afford the practical means of expurgating itself; for it would show on a large scale what particular sounds have manifested their fitness to convey particular ideas, by having been repeatedly chosen among different races to convey them.

Attempts to explain as far as may be the primary formation of speech, by tracing out in detail such processes as have been here described, are likely to increase our knowledge by sure and steady steps wherever imagination does not get the better of sober comparison of facts. But there is one side of this problem of the Origin of Language on which such studies have by no means an encouraging effect. Much of the popular interest in such matters is centred in the question, whether the known languages of the world have their source in one or many primæval tongues. On this subject the opinions of the philologists who have compared the greatest number of languages are utterly at variance, nor has any one brought forward a body of philological evidence strong and direct enough to make anything beyond mere vague opinion justifiable. Now such processes as the growth of imitative or symbolic words form a part, be it small or large, of the Origin of Language, but they are by no means restricted to any particular place or period, and are indeed more or less in activity now. Their operation on any two dialects of one language will be to introduce in each a number of new and independent words, and words even suspected of having been formed in this direct way become valueless as proof of genealogical connexion between the languages in which they are found. The test of such genealogical connexion must, in fact, be generally narrowed to such words or grammatical forms as have become so far conventional in sound and sense, that we cannot suppose two tribes to have arrived at them independently, and therefore consider that both must have inherited them from a common source. Thus the introduction of new sound-words tends to make it practically of less and less consequence to a language what its original stock of words at starting may have been; and the philologist's extension of his knowledge of such direct formations must compel him to strip off more and more of any language, as being possibly of later growth, before he can set himself to argue upon such a residuum as may have come by direct inheritance from times of primæval speech.

In concluding this survey, some general considerations suggest themselves as to the nature and first beginnings of language. In studying the means of expression among men in stages of mental culture far below our own, one of our first needs is to clear our minds of the kind of superstitious veneration with which articulate speech has so commonly been treated, as though it were not merely the principal but the sole means of uttering thought. We must cease to measure the historical importance of emotional exclamations, of gesture-signs, and of picture-writing, by their comparative insignificance in modern civilized life, but must bring ourselves to associate the articulate words of the dictionary in one group with cries and gestures and pictures, as being all of them means of manifesting outwardly the inward workings of the mind. Such an admission, it must be observed, is far from being a mere detail of scientific classification. It has really a most important bearing on the problem of the Origin of Language. For as the reasons are mostly dark to us, why particular words are currently used to express particular ideas, language has come to be looked upon as a mystery, and either occult philosophical causes have been called in to explain its phenomena, or else the endowment of man with the faculties of thought and utterance has been deemed insufficient, and a special revelation has been demanded to put into his mouth the vocabulary of a particular language. In the debate which has been carried on for ages over this much-vexed problem, the saying in the 'Kratylos' comes back to our minds again and again, where Sokrates describes the etymologists who release themselves from their difficulties as to the origin of words by saying that the first words were divinely made, and therefore right, just as the tragedians, when they are in perplexity, fly to their machinery and bring in the gods.^[16] Now I think that those who soberly contemplate the operation of cries, groans, laughs, and other emotional utterances, as to which some considerations have been here brought forward, will admit that, at least, our present crude understanding of this kind of expression would lead us to class it among the natural actions of man's body and mind. Certainly, no one who understands anything of the gesture-language or of picture-writing would be justified in regarding either as due to occult causes, or to any supernatural interference with the course of man's intellectual development. Their cause evidently lies in natural operations of the human mind, not such as were effective in some long-past condition of humanity and have since disappeared, but in processes existing amongst us, which we can understand and even practise for ourselves. When we study the pictures and gestures with which savages and the deaf-and-dumb express their minds, we can mostly see at a glance the direct relation between the outward sign and the inward thought which it makes manifest. We may see the idea of 'sleep' shown in gesture by the head with shut eyes, leant heavily against the open hand; or the idea of 'running' by the attitude of the runner, with chest forward, mouth half open, elbows and shoulders well back; or 'candle' by the straight forefinger held up, and as it were blown out; or 'salt' by the imitated act of sprinkling it with thumb and finger. The figures of the child's picture-book, the sleeper and the runner, the candle and the salt-cellar, show their purport by the same sort of evident relation between thought and sign. We so far understand the nature of these modes of utterance, that we are ready ourselves to express thought after thought by such means, so that those who see our signs shall perceive our meaning.

When, however, encouraged by our ready success in making out the nature and action of these ruder methods, we turn to the higher art of speech, and ask how such and such words have come to express such and such thoughts, we find ourselves face to face with an immense problem, as yet but in small part solved. The success of investigation has indeed been enough to encourage us to push vigorously forward in the research, but the present explorations have not extended beyond corners and patches of an elsewhere unknown field. Still the results go far to warrant us in associating expression by gestures and pictures with articulate language as to principles of original formation, much as men associate them in actual life by using gesture and word at once. Of course, articulate speech, in its far more complex and elaborate development, has taken up devices to which the more simple and rude means of communication offer nothing comparable. Still, language, so far as its constitution is understood, seems to have been developed like writing or music, like hunting or fire-making, by the exercise of purely human faculties in purely human ways. This state of things by no means belongs exclusively to rudimentary philological operations, such as the choosing expressive sounds to name corresponding ideas by. In the higher departments of speech, where words already existing are turned to account to express new meanings and shade off new distinctions, we find these ends attained by contrivances ranging from extreme dexterity down to utter clumsiness. For a single instance, one great means of giving new meaning to old sound is metaphor, which transfers ideas from hearing to seeing, from touching to thinking, from the concrete of one kind to the abstract of another, and can thus make almost anything in the world help to describe or suggest anything else. What the German philosopher described as the relation of a cow to a comet, that both have tails, is enough and more than enough for the language-maker. It struck the Australians, when they saw a European book, that it opened and shut like a mussel-shell, and they began accordingly to call books 'mussels' (mūyūm). The sight of a steam engine may suggest a whole group of such transitions in our own language; the steam passes along 'fifes' or 'trumpets,' that is, pipes or tubes, and enters by 'folding-doors' or valves, to push a 'pestle' or piston up and down in a 'roller' or cylinder, while the light pours from the furnace in 'staves' or 'poles,' that is, in rays or beams. The dictionaries are full of cases compared with which such as these are plain and straightforward. Indeed, the processes by which words have really come into existence may often enough remind us of the game of 'What is my thought like?' When one knows the answer, it is easy enough to see what junketting and cathedral canons have to do with reeds; Latin juncus 'a reed,' Low Latin juncata, 'cheese made in a reed-basket,' Italian giuncata 'cream cheese in a rush frail,' French joncade and English junket, which are preparations of cream, and lastly junketting parties where such delicacies are eaten; Greek κάννη, 'reed, cane,' κανῶν, 'measure, rule,' thence canonicus, 'a clerk under the ecclesiastical rule or canon.' But who could guess the history of these words, who did not happen to know these intermediate links?

Yet there is about this process of derivation a thoroughly human artificial character. When we know the whole facts of any case, we can generally understand it at once, and see that we might have done the same ourselves had it come in our way. And the same thing is true of the processes of making sound-words detailed in these chapters. Such a view is, however, in no way inconsistent with the attempt to generalize upon these processes, and to state them as phases of the development of language among mankind. If certain men under certain circumstances produce certain results, then we may at least expect that other men much resembling these and placed under roughly similar circumstances will produce more or less like results; and this has been shown over and over again in these pages to be what really happens. Now Wilhelm von Humboldt's view that language is an 'organism' has been considered a great step in philological speculation; and so far as it has led students to turn their minds to the search after general laws, no doubt it has been so. But it has also caused an increase of vague thinking and talking, and thereby no small darkening of counsel. Had it been meant to say that human thought, language, and action generally, are organic in their nature, and work under fixed laws, this would be a very different matter; but this is distinctly not what is meant, and the very object of calling language an organism is to keep it apart from mere human arts and contrivances. It was a hateful thing to Humboldt's mind to 'bring down speech to a mere operation of the understanding.' 'Man,' he says, 'does not so much form language, as discern with a kind of joyous wonder its developments, coming forth as of themselves.' Yet, if the practical shifts by which words are shaped or applied to fit new meanings are not devised by an operation of the understanding, we ought consistently to carry the stratagems of the soldier in the field, or the contrivances of the workman at his bench, back into the dark regions of instinct and involuntary action. That the actions of individual men combine to produce results which may be set down in those general statements of fact which we call laws, may be stated once again as one of the main propositions of the Science of Culture. But the nature of a fact is not altered by its being classed in common with others of the same kind, and a man is not less the intelligent inventor of a new word or a new metaphor, because twenty other intelligent inventors elsewhere may have fallen on a similar expedient.

The theory that the original forms of language are to be referred to a low or savage condition of culture among the remotely ancient human race, stands in general consistency with the known facts of philology. The causes which have produced language, so far as they are understood, are notable for that childlike simplicity of operation which befits the infancy of human civilization. The ways in which sounds are in the first instance chosen and arranged to express ideas, are practical expedients at the level of nursery philosophy. A child of five years old could catch the meaning of imitative sounds, interjectional words, symbolism of sex or distance by contrast of vowels. Just as no one is likely to enter into the real nature of mythology who has not the keenest appreciation of nursery tales, so the spirit in which we guess riddles and play at children's games is needed to appreciate the lower phases of language. Such a state of things agrees with the opinion that such rudimentary speech had its origin among men while in a childlike intellectual condition, and thus the self-expressive branch of savage language affords valuable materials for the problem of primitive speech. If we look back in imagination to an early period of human intercourse, where gesture and self-expressive utterance may have had a far greater comparative importance than among ourselves, such a conception introduces no new element into the problem, for a state of things more or less answering to this is described among certain low savage tribes. If we turn from such self-expressive utterance, to that part of articulate language which carries its sense only by traditional and seemingly arbitrary custom, we shall find no contradiction to the hypothesis. Sound carrying direct meaning may be taken up as an element of language, keeping its first significance recognizable to nations yet unborn. But it may far more probably become by wear of sound and shift of sense an expressionless symbol, such as might have been chosen in pure arbitrariness — a philological process to which the vocabularies of savage dialects bear full witness. In the course of the development of language, such traditional words with merely an inherited meaning have in no small measure driven into the background the self-expressive words, just as the Eastern figures 2, 3, 4, which are not self-expressive, have driven into the background the Roman numerals II, III, IIII, which are — this, again, is an operation which has its place in savage as in cultivated speech. Moreover, to look closely at language as a practical means of expressing thought, is to face evidence of no slight bearing on the history of civilization. We come back to the fact, so full of suggestion, that the languages of the world represent substantially the same intellectual art, the higher nations indeed gaining more expressive power than the lowest tribes, yet doing this not by introducing new and more effective central principles, but by mere addition and improvement in detail. The two great methods of naming thoughts and stating their relation to one another, viz., metaphor and syntax, belong to the infancy of human expression, and are as thoroughly at home in the language of savages as of philosophers. If it be argued that this similarity in principles of language is due to savage tribes having descended from higher culture, carrying down with them in their speech the relics of their former excellence, the answer is that linguistic expedients are actually worked out with as much originality, and more extensively if not more profitably, among savages than among cultured men. Take for example the Algonquin system of compounding words, and the vast Esquimaux scheme of grammatical inflexion. Language belongs in essential principle both to low grades and high of civilization; to which should its origin be attributed? An answer may be had by comparing the methods of language with the work it has to do. Take language all in all over the world, it is obvious that the processes by which words are made and adapted have far less to do with systematic arrangement and scientific classification, than with mere rough and ready ingenuity and the great rule of thumb. Let any one whose vocation it is to realize philosophical or scientific conceptions and to express them in words, ask himself whether ordinary language is an instrument planned for such purposes. Of course it is not. It is hard to say which is the more striking, the want of scientific system in the expression of thought by words, or the infinite cleverness of detail by which this imperfection is got over, so that he who has an idea does somehow make shift to get it clearly in words before his own and other minds. The language by which a nation with highly developed art and knowledge and sentiment must express its thoughts on these subjects, is no apt machine devised for such special work, but an old barbaric engine added to and altered, patched and tinkered into some sort of capability. Ethnography reasonably accounts at once for the immense power and the manifest weakness of language as a means of expressing modern educated thought, by treating it as an original product of low culture, gradually adapted by ages of evolution and selection, to answer more or less sufficiently the requirements of modern civilization.

1. Mpongwe punjina; Basuto foka; Carib phoubäe; Arawac appüdün (ignem sufflare). Other cases are given by Wedgwood, 'Or. of Lang.' p. 83.
2. See Wedgwood, 'Dic.' Introd. p. viii.
3. See Wedgwood, Dic., s.v. 'mum,' &c.
4. Bates, 'Naturalist on the Amazons,' 2nd ed., p. 404; Markham in 'Tr. Eth. Soc.,' vol. iii. p. 143.
5. 'Avesta,' Farg. xviii. 34-5.
6. Wedgwood, Dic., s.v. 'pigeon;' Diez, 'Etym. Wörterb.,' s.v. 'piccione.'
7. Bopp, 'Gloss. Sanscr.,' s.v. 'go.' See Pott, 'Wurzel-Wörterb. der Indo-Germ. Spr.,' s.v. 'gu,' 'Zählmethode,' p. 227.
8. Pott, 'Doppelung (Reduplication, Gemination) als eines der wichtigsten Bildungsmittel der Sprache,' 1862. Frequent use has been here made of this work.
9. For authorities see especially Pott, 'Doppelung,' p. 30, 47-49; W. v. Humboldt, 'Kawi-Spr.' vol. ii. p. 36; Max Müller in Bunsen, 'Philos. of Univ. Hist.' vol. i. p. 329; Latham, 'Comp. Phil.' p. 200; and the grammars and dictionaries of the particular languages. The Guarani and Carib on authority of D'Orbigny, 'L'Homme Américain,' vol. ii. p. 268; Dhimal of Hodgson, 'Abor. of India,' p. 69, 79, 115; Colville Ind. of Wilson in 'Tr. Eth. Soc.' vol. iv. p. 331; Botocudo of Martius, 'Gloss. Brasil.'
10. Also Old High German diz and daz.
11. Max Müller, l.c.
12. J. C. E. Buschmann, 'Ueber den Naturlaut,' Berlin, 1853; and in 'Abh. der K. Akad. d. Wissensch,' 1852. An English trans. in 'Proc. Philological Society,' vol. vi. See De Brosses, 'Form. des L.,' vol. i. p, 211.
13. One family of languages, the Athapascan, contains both appá and mama as terms for 'father,' in the Tahkali and Tlatskanai.
14. See Pott, 'Indo-Ger. Wurzelwörterb.' s.v. 'pâ'; Böhtlingk and Roth, 'Sanskrit-Wörterb.' s.v. mâtar; Pictet, 'Origines Indo-Europ.,' part ii. p. 349; Max Müller, 'Lectures,' 2nd series, p. 212.
15. Facciolati, 'Lexicon;' Varro, ap. Nonn., ii. 97.
16. Plato, 'Cratylus' 90.