We have next to look at the way in which language grows by combining its words to form new ones. To see this, words have to be noticed not as they stand by themselves, but as they come together in actual speaking. Language consists of sentences, and a sentence is made up of words, each word being a distinct spoken sound carrying a distinct meaning. The simplest notion of a sentence may be had from such a language as Chinese, where it can be taken apart into words which are each a single syllable. Thus kou chi shi jin sse, that is “dog sow eat man food” means that dogs and sows eat the food of men. The class of languages which can be taken to pieces in this perfect way are called analytic or isolating. In most languages of the world, however, which are more or less synthetic or compounding, the tendency is not so strong to keep words separate, and they are apt to attach themselves together. To bring clearly before our minds how the joining or compounding of words takes place, let us notice rather more closely than usual one of our English sentences. On listening, it will appear that the spoken words have not really breaks between them as in writing, but the syllables run on continuously till the speaker pauses, and what marks a word is, not its being really separated, but its having an emphasis, or stress (as it is called by Mr. Sweet). Now, from time to time, certain words may be noticed becoming actually fixed together. How this joining gradually takes place we sometimes try to show by writing them differently, as hard ware, hard-ware, hardware; or steam ship, steam-ship, steamship. On listening to such joined words, it is found that one of the two has lost its stress, the whole compound having now but one stress. This is how in talking English our minds give a sign by our voices that two words have become one. The next step is when the sound of one of the part-words becomes slurred or broken down, as in the end-words of waterman, wrongful. Or both the simple words may have broken down, as in boatswain and coxswain, where writing keeps up the original meaning of the swain in charge of the boat or cock-boat, but in actual speaking the words have shrunk to what may be spelt bōsun, coxun. Now this process of forming a new word by (so to speak) welding together two or more old ones, is one of the chief acts by which word-makers, ancient and modern, have furnished themselves with more manageable terms, which again as the meanings of the separate parts were less cared for, were cut shorter in speaking. When this has not gone too far, philologists can still get back to the original elements of such words, discerning the fourteen night in fortnight; the unus and decem in undecim, shrunk still farther in French onze; the jus, dico, in Latin judex, which in English comes down to judge.
As examples how word-compounding goes on in unfamiliar tongues, may be taken the Malay term for “arrow,” which is anak-panah, or “child-(of-the)-bow;” and the native Australian term for “unanimous,” which is gurdugynyul, or “heart-one-come.” To show how such compound words become shortened, take the Mandingo word for “sister,” inbadingmuso, which is made up of mi bado dingo muso, meaning “my-mother-child-female.” The natives of Vancouver’s Island gave to a certain long-bearded Englishman the name Yakpus; this appears to have come from yakhpekukselkous, made up of words signifying “long-face-hair-man,” which in speaking had been cut down to yakpus. No one who did not happen to be told the history of this word could ever have guessed it. This is an important lesson in the science of language, for it is likely that tens of thousands of words in the languages of the world may have come into the state in which we find them by the shortening of long compound words, and when this has been done recklessly as in the last example, and the history lost, all reasonable hope is gone of ever getting back to the original form and meaning. Nor does this process of contraction affect only compound words, but it may act on a whole sentence, fusing it as it were into one word. Here the synthetic or compounding principle reaches its height. As a contrast to the analytic Chinese sentence given at page 139, to show the perfect distinctness of their words, we may take a sentence of an African language to show how utterly that distinctness may be lost. When a Grebo negro wishes to express that he is very angry, he says in his metaphorical way “it has raised a bone in my breast.” His full words for expressing this would be e ya mu kra wudi, but in speaking he runs them together so that what he actually utters is yamukroure. Where such breaking down has gone on unchecked, it is easy to see how the language of a barbaric tribe may alter so much in a few generations as hardly to be recognised. Indeed, any one who will attend to how English words run together in talking may satisfy himself that his own language would undergo rapid changes like those of barbaric tongues, were it not for the schoolmaster and the printer, who insist on keeping our words fixed and separate.
The few examples here given of new words made by compounding old ones may serve to illustrate the great principle that such combination, far from being a mere source of confusion, has been one of the great means of building up language. Especially, one of the great discoveries in modern philology is how grammatical formation and inflexion has partly come about by a kind of word-compounding. It must have seemed to the old scholars a mysterious and arbitrary proceeding that Latin should have fixed upon a set of meaningless affixes to inflect and make into different parts of speech ago, agis, agit, agere, agens, actum, actor, actio, activus, activè, &c. But the mystery to some extent disappeared when it was noticed how in modern languages the running together of words produced something of the kind. Thus the hood of womanhood, priesthood, which is now a mere grammatical suffix, was in old English a word of itself, hâd, meaning form, order, state; and the suffix -ly was once the distinct word “like,” as is seen by Anglo-Saxon saying cwên-lic, “queen-like,” where modern English says queenly. In Chaucer’s English it is seen how the pronoun thou had dwindled into a mere verb-ending,
“He pokyd Johan, and seyde, Slepistow?
Herdistow ever slik a sang er now?”
In English the future tense of the verb to give is “I will give,” or, colloquially, “I’ll give.” Here writing separates what speaking joins, but the modern French future tense donnerai, donneras, is the verb donner with the auxiliary verb ai, as, both spoken and written on to it, so that “je donnerai” is a phrase like “I have to give.” The plural donnerons, donnerez, can no longer be thus taken to pieces, for the remains of the auxiliary verb have passed into meaningless grammatical affixes ons, ez. There is reason to suppose that many of the affixes of Greek and Latin grammar arose in this way by distinct words combining together and then shrinking. Not that it would be safe to assert that all affixes came into existence in this particular way. As was pointed out in the last chapter, men wanting to utter a thought are clever enough to catch up in very far-fetched ways a sound to express it. Thus the prefix ge, which German uses to make past participles with, seems to have originally signified “with” or “together,” which sense it still retains in such words as gespiele, “playfellow;” but by a curious shifting of purpose it came to serve as a means of forming participles, as spielen, to play, gespielt, played. It was so used also in Anglo-Saxon, as clypian, to call, geclypod, called, which word in its later form yclept still keeps up among us a trace of the old grammatical device. Philologists have to keep their eyes open to this power which language-makers have of using sounds for some new purpose they were not intended for. Thus, in English, the change of vowels in foot, feet, and in find, found, now serves as a means of declining the noun and conjugating the verb. But history happens to show that the vowel change was not originally made with this intention at all. The Anglo-Saxon declension proves that the vowel was not then a sign of number in the noun; it was singular fôt, fôtes, fêt, plural fêt, fôta, fôtum. Nor was it a sign of tense in the Anglo-Saxon verb, where the perfect of findan, to find, had different vowels in its singular, ic fand, I found, and its plural, we fundon, we found. It was the later Englishmen who, knowing nothing of the real reasons which brought about the variation of the vowels, took to using them to mark singular from plural, and present from perfect.
It is the work of grammarians in examining any language to take all its combined words to pieces as far as possible. Greek and Latin grammars now teach how to analyze words by stripping off their affixes, so as to get down to the real part or root, which is generally a simple sound expressing a simple notion. A root is best understood by considering it to have been once a separate word, as it would be in such a language as English. Even in languages where the roots seldom appear without some affix attached, they may stand by themselves as imperative, like Latin dic! say! Turkish sev! love! But in many languages roots can only be found as imaginary forms, by comparing a group of words and getting at the common part belonging to them all. Thus in Latin it appears from gnosco, gnotus, &c., that there must be a root gno which carries the thought of knowing. Going on to Greek, there is found in gignōskō, gnōsis, gnōmē, &c., the same root gno with the same meaning. Turning next to Sanskrit, a similar sound, jnâ, appears as the root-form for knowing. In this way, by comparing the whole set of Aryan or Indo-European languages, it appears that there must have been in ancient times a word something like gna, meaning to know, which is to be traced not only in Sanskrit, Greek, and Latin, but in many other languages of the family, as Russian znat, English know. A few more such Aryan roots, which the reader recognises at once in well-known languages, are sta, to stand, sad, to sit, ga, to go, i, to go, ma, to measure, da, to give, vid, to see, rag, to rule, mar to die. These simple sounds seem to have already become fixed to carry their meanings in the remote ages when the ancestors of the Aryan peoples wandered with their herds on the highlands of Central Asia. It is not needful to tell the student of anthropology how interesting it is to arrive thus at the earliest known root-words of any family. But it should at the same time be noticed that even in the earliest of these sets of roots, we seldom come to anything like an actual origin or beginning. Some few may indeed have been taken direct from the natural language, for instance ru, to roar, and if this was so here is a real origin. But most roots, to whatever languages of the world they may belong, are like the group given above, where it is impossible to say confidently how their sound came to express their meaning. Unless this can be done, it is safest not to take such roots as really primitive formations, for they may have a long lost history of the utmost change. How this may happen, our own language has a useful lesson to teach. Imagine one who knows no language but English trying to get at its roots. To him the verb to roll might seem a root-word, a primitive element of language; indeed it actually has been fancied a natural sound imitating the act of rolling. Yet any philologist would tell him that English roll is a comparatively modern form, which came through a long series of earlier stages; it was borrowed from French rolle, roller, now rôle, rouler, all from Latin rotulus, diminutive of rota, a wheel, even this coming from a more ancient verb and signifying a runner or goer. Still more adventurous is the history of another English word which has now all the parts of a verb, to check, checking, checked, besides such forms as a check in one’s course, the check-string to stop the coachman, the check-valve to stop the water in a pipe. This word check has all the simplicity of sound and sense which might belong to an original root-word. Yet strange to say, it is really the Persian word shah, meaning “king,” which came to Europe with the game of chess as the word of challenge to the king, and thence by a curious metaphor passed into a general word for stopping anybody or anything. For all that is known, many root-words among the Greeks or Jews, or even the simple-looking monosyllables of the Chinese, may during præhistoric ages have travelled as far from their real origin as these English verbs. Thus the roots from which language grows may often be themselves sprung as it were from yet earlier seeds or cuttings, grown at home or imported from abroad, and though in our time words mostly come from the ancient roots, the power of striking new roots is not yet dead.
Having now, in such a broad way as suits the present purpose, looked at the formation of words, something may be said as to how language contrives to show the relations among the words of a sentence. This is done by what grammarians call syntax, concord, and government. It has been seen ([p. 119]) that the gesture-language, though wanting in grammatical forms, has a strongly marked syntax. The deaf-mute’s signs must follow one another in proper order, otherwise they may convey a wrong meaning or seem nonsense. So, in spoken languages which do not inflect their words, such as the Chinese, syntax is the main part of grammar; thus li ping = sharp weapons, ping li = weapons (are) sharp; chi kuo = to govern the kingdom, but kuo chi = the kingdom is governed. This seems quite natural to us, for modern English has come far towards the Chinese plan of making the sense of the sentence depend on the order of the words, thus marking the difference between rank of families and families of rank, or between men kill lions and lions kill men. In Latin it is very different, where words can be put about with such freedom, that the English reader may be hardly able to make sense of one of Tacitus’ sentences without fresh sorting the words into some order he can think them in. Especially in Latin verses there is often hardly more syntax than if the words were nonsense-syllables arranged only to scan. The sense has to be made out from the grammatical inflections, as where it is seen that in “vile potabis modicis Sabinum cantharis,” the cheapness has to do with the wine and the smallness with the mugs. It is because so many of the inflections have disappeared from English, that the English translation has to obtain a proper understanding by stricter order of words. Where the meaning of sentences depends on order or syntax, that order must be followed, but it must be borne in mind that this order differs in different languages. For a single instance, in Malay, where orang = man and utan = forest, savages and apes are called orang utan, which is just opposite to the English construction “forest man.”
Every one who can construe Greek and Latin sees what real service is done by government and agreement in showing how the words of a sentence hang together, what quality is stated of what thing, or who is asserted to act on what. But even Greek and Latin have changed so much from their earlier state, that they often fail to show the scholar clearly what they mean to do, and why. It is useful to make acquaintance with the languages of ruder nations, which show government and agreement in earlier and plainer stages of growth. One great object of grammatical construction is to make it quite clear which of two nouns concerned is subject and which object, for instance, whether it was a chief who killed a bear, or a bear who killed a chief. A particle properly attached will do this, as when the Algonquin Indians put on the syllable un both to noun and verb, in a way which we may try to translate by the pronoun him, thus:—