Language: Its Nature, Development and Origin

XVIII.—§ 1. Nominal Forms.

In the flexion of substantives and adjectives we see phenomena corresponding to those we have just been considering in the verbs. The ancient languages of our family have several forms where modern languages content themselves with fewer; forms originally kept distinct are in course of time confused, either through a phonetic obliteration of differences in the endings or through analogical extension of the functions of one form. The single form good is now used where OE. used the forms god, godne, gode, godum, godes, godre, godra, goda, godan, godena; Ital. uomo or French homme is used for Lat. homo, hominem, homini, homine—nay, if we take the spoken form into consideration, Fr. [ɔm] corresponds not only to these Latin forms, but also to homines, hominibus. Where the modern language has one or two cases, in an earlier stage it had three or four, and still earlier seven or eight. The difficulties inherent in the older system cannot, however, be measured adequately by the number of forms each word is susceptible of, but are multiplied by the numerous differences in the formation of the same case in different classes of declension; sometimes we even find anomalies which affect one word only.

Those who would be inclined to maintain that new irregularities may and do arise in modern languages which make up for whatever earlier irregularities have been discarded in the course of the historical development will do well to compile a systematic list of all the flexional forms of two different stages of the same languages, arranged exactly according to the same principles: this is the only way in which it is possible really to balance losses and profits in a language. This is what I have done in my Progress in Language § 111 ff. (reprinted in ChE § 9 ff.), where I have contrasted the case systems of Old and Modern English: the result is that the former system takes 7 (+ 3) pages, and the latter only 2 pages. Those pages, with their abbreviations and tabulations, do not, perhaps, offer very entertaining reading, but I think they are more illustrative of the real tendencies of language than either isolated examples or abstract reasonings, and they cannot fail to convince any impartial reader of the enormous gain achieved through the changes of the intervening nine hundred years in the general structure of the English language.

For our general purposes it will be worth our while here to quote what Friedrich Müller (Gr i. 2. 7) says about a totally different language: “Even if the Hottentot distinguishes ‘he,’ ‘she’ and ‘it,’ and strictly separates the singular from the plural number, yet by his expressing ‘he’ and ‘she’ by one sound in the third person, and by another in the second, he manifests that he has no perception at all of our two grammatical categories of gender and number, and consequently those elements of his language that run parallel to our signs of gender and number must be of an entirely different nature.” Fr. Müller should not perhaps throw too many stones at the poor Hottentots, for his own native tongue is no better than a glass house, and we might with equal justice say, for instance: “As the Germans express the plural number in different manners in words like gott—götter, hand—hände, vater—väter, frau—frauen, etc., they must be entirely lacking in the sense of the category of number.” Or let us take such a language as Latin; there is nothing to show that dominus bears the same relation to domini as verbum to verba, urbs to urbes, mensis to menses, cornu to cornua, fructus to fructūs, etc.; even in the same word the idea of plurality is not expressed by the same method for all the cases, as is shown by a comparison of dominus—domini, dominum—dominos, domino—dominis, domini—dominorum. Fr. Müller is no doubt wrong in saying that such anomalies preclude the speakers of the language from conceiving the notion of plurality; but, on the other hand, it seems evident that a language in which a difference so simple even to the understanding of very young children as that between one and more than one can only be expressed by a complicated apparatus must rank lower than another language in which this difference has a single expression for all cases in which it occurs. In this respect, too, Modern English stands higher than the oldest English, Latin or Hottentot.

XVIII.—§ 2. Irregularities Original.

It was the belief of the older school of comparativists that each case had originally one single ending, which was added to all nouns indifferently (e.g. -as for the genitive sg.), and that the irregularities found in the existing oldest languages were of later growth; the actually existing forms were then derived from the supposed unity form by all kinds of phonetic tricks and dodges. Now people have begun to see that the primeval language cannot have been quite uniform and regular (see, for instance, Walde in Streitberg’s Gesch., 2. 194 ff.). If we look at facts, and not at imagined or reconstructed forms, we are forced to acknowledge that in the oldest stages of our family of languages not only did the endings present the spectacle of a motley variety, but the kernel of the word was also often subject to violent changes in different cases, as when it had in different forms different accentuation and (or) different apophony, or as when in some of the most frequently occurring words some cases were formed from one ‘stem’ and others from another, for instance, the nominative from an r stem and the oblique cases from an n stem. In the common word for ‘water’ Greek has preserved both stems, nom. hudōr, gen. hudatos, where a stands for original [ən]. Whatever the origin of this change of stems, it is a phenomenon belonging to the earlier stages of our languages, in which we also sometimes find an alteration between the r stem in the nominative and a combination of the n and the r stems in the other cases, as in Lat. jecur ‘liver,’ jecinoris; iter ‘voyage,’ itineris, which is supposed to have supplanted itinis, formed like feminis from femur. In the later stages we always find a simplification, one single form running through all cases; this is either the nominative stem, as in E. water, G. wasser (corresponding to Gr. hudōr), or the oblique case-stem, as in the Scandinavian forms, Old Norse vatn, Swed. vatten, Dan. vand (corresponding to Gr. hudat-), or finally a contaminated form, as in the name of the Swedish lake Vättern (Noreen’s explanation), or in Old Norse and Dan. skarn ‘dirt,’ which has its r from a form like the Gr. skōr, and its n from a form like the Gr. genitive skatos (older [skəntos]). The simplification is carried furthest in English, where the identical form water is not only used unchanged where in the older languages different case forms would have been used (‘the water is cold,’ ‘the surface of the water,’ ‘he fell into the water,’ ‘he swims in the water’), but also serves as a verb (‘did you water the flowers?’), and as an adjunct as a quasi-adjective (‘a water melon,’ ‘water plants’).

In most cases irregularities have been done away with in the way here indicated, one of the forms (or stems) being generalized; but in other cases it may have happened, as Kretschmer supposes (in Gercke and Norden, Einleit. in die Altertumswiss., I, 501) that irregular flexion caused a word to go out of use entirely; thus in Modern Greek hêpar was supplanted by sukōti,[84] phréar by pēgadi, húdōr by neró, oûs by aphtí (= ōtíon), kúōn by skullí; this possibly also accounts for commando taking the place of Lat. jubeo.

Some scholars maintain that the medieval languages were more regular than their modern representatives; but if we look more closely into what they mean, we shall see that they are not speaking of any regularity in the sense in which the word has here been used—the only regularity which is of importance to the speakers of the language—but of the regular correspondence of a language with some earlier language from which it is derived. This is particularly the case with E. Littré, who, in his essays on L’Histoire de la Langue Française, was full of enthusiasm for Old French, but chiefly for the fidelity with which it had preserved some features of Latin. There was thus the old distinction of two cases: nom. sg. murs, acc. sg. mur, and in the plural inversely nom. mur and acc. murs, with its exact correspondence with Latin murus, murum, pl. muri, muros. When this ‘règle de l’s’ was discovered, and the use or omission of s, which had hitherto been looked upon as completely arbitrary in Old French, was thus accounted for, scholars were apt to consider this as an admirable trait in the old language which had been lost in modern French, and the same view obtained with regard to the case distinction found in other words, such as OFr. nom. maire, acc. majeur, or nom. emperere, acc. emperëur, corresponding to the Latin forms with changing stress, májor, majórem, imperátor, imperatórem, etc. But, however interesting such things may be to the historical linguist, there is no denying that to the users of French the modern simpler flexion is a gain as compared with this more complex system. “Des sprachhistorikers freud ist des sprachbrauchers leid,” as Schuchardt somewhere shrewdly remarks.

XVIII.—§ 3. Syntax.

There were also in the old languages many irregularities in the syntactic use of the cases, as when some verbs governed the genitive and others the dative, etc. Even if it may be possible in many instances to account historically for these uses, to the speakers of the languages they must have appeared to be mere caprices which had to be learned separately for each verb, and it is therefore a great advantage when they have been gradually done away with, as has been the case, to a great extent, even in a language like German, which has retained many old case forms. Thus verbs like entbehren, vergessen, bedürfen, wahrnehmen, which formerly took the genitive, are now used more and more with the simple accusative—a simplification which, among other things, makes the construction of sentences in the passive voice easier and more regular.

The advantage of discarding the old case distinctions is seen in the ease with which English and French speakers can say, e.g., ‘with or without my hat,’ or ‘in and round the church,’ while the correct German is ‘mit meinem hut oder ohne denselben’ and ‘in der kirche und um dieselbe’; Wackernagel writes: “Was in ihm und um ihn und über ihm ist.” When the prepositions are followed by a single substantive without case distinction, German, of course, has the same simple construction as English, e.g. ‘mit oder ohne geld,’ and sometimes even good writers will let themselves go and write ‘um und neben dem hochaltare’ (Goethe), or ‘Ihre tochter wird meine frau mit oder gegen ihren willen’ (these examples from Curme, German Grammar 191). Cf. also: ‘Ich kann deinem bruder nicht helfen und ihn unterstützen.’

Many extremely convenient idioms unknown in the older synthetic languages have been rendered possible in English through the doing away with the old case distinctions, such as: Genius, demanding bread, is given a stone after its possessor’s death (Shaw) (cf. my ChE § 79) | he was offered, and declined, the office of poet-laureate (Gosse) | the lad was spoken highly of | I love, and am loved by, my wife | these laws my readers, whom I consider as my subjects, are bound to believe in and to obey (Fielding) | he was heathenishly inclined to believe in, or to worship, the goddess Nemesis (id.) | he rather rejoiced in, than regretted, his bruise (id.) | many a dun had she talked to, and turned away from her father’s door (Thackeray) | their earthly abode, which has seen, and seemed almost to sympathize in, all their honour (Ruskin).

XVIII.—§ 4. Objections.

Against my view of the superiority of languages with few case distinctions, Arwid Johannson, in a very able article (in IF I, see especially p. 247 f.), has adduced a certain number of ambiguous sentences from German:

According to Johannson, these passages show the disadvantages of doing away with formal distinctions, for the sentences would have been clear if each separate case had had its distinctive sign; “the greater the wealth of forms, the more intelligible the speech.” And they show, he says, that such ambiguities will occur, even where the strictest rules of word order are observed. I shall not urge that this is not exactly the case in the last sentence if die schwester and dein weib are to be taken as accusatives, for then an should have been placed at the very end of the sentence; nor that, in the last sentence but one, the mention of George Eliot as the ‘konkurrent’ of Fernan Caballero seems to show a partiality to the Spanish authoress on the part of the writer of the sentence, so that the reader is prepared to take welche as the nominative case; freilich would seem to point in the same direction. But these, of course, are only trifling objections; the essential point is that we must grant the truth of Johannson’s contention that we have here a flaw in the German language; the defects of its grammatical system may and do cause a certain number of ambiguities. Neither is it difficult to find the reasons of these defects by considering the structure of the language in its entirety, and by translating the sentences in question into a few other languages and comparing the results.

First, with regard to the formal distinctions between cases, the really weak point cannot be the fewness of these endings, for in that case we should expect the same sort of ambiguities to be very common in English and Danish, where the formal case distinctions are considerably fewer than in German; but as a matter of fact such ambiguities are more frequent in German than in the other two languages. And, however paradoxical it may seem at first sight, one of the causes of this is the greater wealth of grammatical forms in German. Let us substitute other words for the ambiguous ones, and we shall see that the amphibology will nearly always disappear, because most other words will have different forms in the two cases, e.g.:

It is this very fact that countless sentences of this sort are perfectly clear which leads to the employment of similar constructions even where the resulting sentence is by no means clear; but if all, or most, words were identical in the nominative and the dative, like gott, or in the dative and genitive, like der Indierin, constructions like those used would be impossible to imagine in a language meant to be an intelligible vehicle of thought. And so the ultimate cause of the ambiguities is the inconsistency in the formation of the several cases. But this inconsistency is found in all the old languages of the Aryan family: cases which in one gender or with one class of stems are kept perfectly distinct, are in others identical. I take some examples from Latin, because this is perhaps the best known language of this type, but Gothic or Old Slavonic would show inconsistencies of the same kind. Domini is genitive singular and nominative plural (corresponding to, e.g., verbi and verba); verba is nominative and accusative pl. (corresponding to domini and dominos); domino is dative and ablative; dominæ gen. and dative singular and nominative plural; te is accusative and ablative; qui is singular and plural; quæ singular fem. and plural fem. and neuter, etc. Hence, while patres filios amant or patres filii amant are perfectly clear, patres consules amant allows of two interpretations; and in how many ways cannot such a proposition as Horatius et Virgilius poetæ Varii amici erant be construed? Menenii patris munus may mean ‘the gift of father Menenius,’ or ‘the gift of Menenius’s father’; expers illius periculi either ‘free from that danger’ or ‘free from (sharing) that person’s danger’; in an infinitive construction with two accusatives, the only way to know which is the subject and which the object is to consider the context, and that is not always decisive, as in the oracular response given to the Æacide Pyrrhus, as quoted by Cicero from Ennius: “Aio te, Æacida, Romanos vincere posse.” Such drawbacks seem to be inseparable from the structure of the highly flexional Aryan languages; although they are not logical consequences of a wealth of forms, yet historically they cling to those languages which have the greatest number of grammatical endings. And as we are here concerned not with the question how to construct an artificial language (and even there I should not advise the adoption of many case distinctions), but with the valuation of natural languages as actually existing in their earlier and modern stages, we cannot accept Johannson’s verdict: “The greater the wealth of forms, the more intelligible the speech.”

XVIII.—§ 5. Word Order.

If the German sentences quoted above are ambiguous, it is not only on account of the want of clearness in the forms employed, but also on account of the German rules of word order. One rule places the verb last in subordinate sentences, and in two of the sentences there would be no ambiguity in principal sentences: Die deutsche zunge klingt und singt gott im himmel lieder; or, Die deutsche zunge klingt, und gott im himmel singt lieder | Sie erreicht freilich nicht die spanische dichterin; or, Die spanische dichterin erreicht sie freilich nicht. In one of the remaining sentences the ambiguity is caused by the rule that the verb must be placed immediately after an introductory subjunct: if we omit doch the sentence becomes clear: Die gesellschaft der Indierin würde lästig gewesen sein, or, Die gesellschaft würde der Indierin lästig gewesen sein. Here, again we see the ill consequences of inconsistency of linguistic structure; some of the rules for word position serve to show grammatical relations, but in certain cases they have to give way to other rules, which counteract this useful purpose. If you change the order of words in a German sentence, you will often find that the meaning is not changed, but the result will be an unidiomatic construction (bad grammar); while in English a transposition will often result in perfectly good grammar, only the meaning will be an entirely different one from the original sentence. This does not amount to saying that the German rules of position are useless and the English ones all useful, but only to saying that in English word order is utilized to express difference of meaning to a far greater extent than in German.

One critic cites against me “one example, which figures in almost every Rhetoric as a violation of clearness: And thus the son the fervid sire address’d,” and he adds: “The use of a separate form for nominative and accusative would clear up the ambiguity immediately.” The retort is obvious: no doubt it would, but so would the use of a natural word order. Word order is just as much a part of English grammar as case-endings are in other languages; a violation of the rules of word order may cause the same want of intelligibility as the use of dominum instead of dominus would in Latin. And if the example is found in almost every English Rhetoric, I am glad to say that equally ambiguous sentences are very rare indeed in other English books. Even in poetry, where there is such a thing as poetic licence, and where the exigencies of rhythm and rime, as well as the fondness for archaic and out-of-the-way expressions, will often induce deviations from the word order of prose, real ambiguity will very seldom arise on that account. It is true that it has been disputed which is the subject in Gray’s line:

but then it does not matter much, for the ultimate understanding of the line must be exactly the same whether the air holds stillness or stillness holds the air. In ordinary language we may find similar collocations, but it is worth saying with some emphasis that there can never be any doubt as to which is the subject and which the object. The ordinary word order is, Subject-Verb-Object, and where there is a deviation there must always be some special reason for it. This may be the wish, especially for the sake of some contrast, to throw into relief some member of the sentence. If this is the subject, the purpose is achieved by stressing it, but the word order is not affected. But if it is the object, this may be placed in the very beginning of the sentence, but in that case English does not, like German and Danish, require inversion of the verb, and the order consequently is, Object-Subject-Verb, which is perfectly clear and unambiguous. See, for instance, Dickens’s sentence: “Talent, Mr. Micawber has; capital, Mr. Micawber has not,” and the following passage from a recent novel: “Even Royalty had not quite their glow and glitter; Royalty you might see any day, driving, bowing, smiling. The Queen had a smile for every one; but the Duchess no one, not even Lizzie, ever saw.” Thus, also, in Shakespeare’s:

and in Longfellow’s translation from Logau:

The reason for deviating from the order, Subject-Verb-Object, may again be purely grammatical: a relative or an interrogative pronoun must be placed first; but here, too, English grammar precludes ambiguity, as witness the following sentences: This picture, which surpasses Mona Lisa | This picture, which Mona Lisa surpasses | What picture surpasses Mona Lisa? | What picture does not Mona Lisa surpass? In German (dieses bild, welches die M. L. übertrifft, etc.) all four sentences would be ambiguous, in Danish the two last would be indistinguishable; but English shows that a small number of case forms is not incompatible with perfect clearness and perspicuity. If the famous oracular answer (Henry VI, 2nd Part, I. 4. 33), “The Duke yet liues, that Henry shall depose,” is ambiguous, it is only because it is in verse, where you expect inversions: in ordinary prose it could be understood only in one way, as the word order would be reversed if Henry was meant as the object.

XVIII.—§ 6. Gender.

Besides case distinctions the older Aryan languages have a rather complicated system of gender distinctions, which in many instances agrees with, but in many others is totally independent of, and even may be completely at war with, the natural distinction between male beings, female beings and things without sex. This grammatical gender is sometimes looked upon as something valuable for a language to possess; thus Schroeder (Die formale Unterscheidung 87) says: “The formal distinction of genders is decidedly an enormous advantage which the Aryan, Semitic and Egyptian languages have before all other languages.” Aasen (Norsk Grammatik 123) finds that the preservation of the old genders gives vividness and variety to a language; he therefore, in constructing his artificial Norwegian ‘landsmaal,’ based it on those dialects which made a formal distinction between the masculine and feminine article. But other scholars have recognized the disadvantages accruing from such distinctions; thus Tegnér (SM 50) regrets the fact that in Swedish it is impossible to give such a form to the sentence ‘sin make må man ej svika’ as to make it clear that the admonition is applicable to both husband and wife, because make, ‘mate,’ is masculine, and maka feminine. In Danish, where mage is common to both sexes, no such difficulty arises. Gabelentz (Spr 234) says: “Das grammatische geschlecht bringt es weiter mit sich dass wir deutschen nie eine frauensperson als einen menschen und nicht leicht einen mann als eine person bezeichnen.”

As a matter of fact, German gender is responsible for many difficulties, not only when it is in conflict with natural sex, as when one may hesitate whether to use the pronoun es or sie in reference to a person just mentioned as das mädchen or das weib, or er or sie in reference to die schildwache, but also when sexless things are concerned, and er might be taken as either referring to the man or to der stuhl or to der wald just mentioned, etc. In France, grammarians have disputed without end as to the propriety or not of referring to the (feminine) word personnes by means of the pronoun ils (see Nyrop, Kongruens 24, and Gr. iii. § 712): “Les personnes que vous attendiez sont tous logés ici.” As a negative pronoun personne is now frankly masculine: ‘personne n’est malheureux.’ With gens the old feminine gender is still kept up when an adjective precedes, as in les bonnes gens, thus also toutes les bonnes gens, but when the adjective has no separate feminine form, schoolmasters prefer to say tous les honnêtes gens, and the masculine generally prevails when the adjective is at some distance from gens, as in the old school-example, Instruits par l’expérience, toutes les vieilles gens sont soupçonneux. There is a good deal of artificiality in the strict rules of grammarians on this point, and it is therefore good that the Arrêté ministériel of 1901 tolerates greater liberty; but conflicts are unavoidable, and will rise quite naturally, in any language that has not arrived at the perfect stage of complete genderlessness (which, of course, is not identical with inability to express sex-differences).

Most English pronouns make no distinction of sex: I, you, we, they, who, each, somebody, etc. Yet, when we hear that Finnic and Magyar, and indeed the vast majority of languages outside the Aryan and Semitic world, have no separate forms for he and she, our first thought is one of astonishment; we fail to see how it is possible to do without this distinction. But if we look more closely we shall see that it is at times an inconvenience to have to specify the sex of the person spoken about. Coleridge (Anima Poetæ 190) regretted the lack of a pronoun to refer to the word person, as it necessitated some stiff and strange construction like ‘not letting the person be aware wherein offence had been given,’ instead of ‘wherein he or she has offended.’ It has been said that if a genderless pronoun could be substituted for he in such a proposition as this: ‘It would be interesting if each of the leading poets would tell us what he considers his best work,’ ladies would be spared the disparaging implication that the leading poets were all men. Similarly there is something incongruous in the following sentence found in a German review of a book: “Was Maria und Fritz so zueinander zog, war, dass jeder von ihnen am anderen sah, wie er unglücklich war.” Anyone who has written much in Ido will have often felt how convenient it is to have the common-sex pronouns lu (he or she), singlu, altru, etc. It is interesting to see the different ways out of the difficulty resorted to in actual language. First the cumbrous use of he or she, as in Fielding TJ 1. 174, the reader’s heart (if he or she have any) | Miss Muloch H. 2. 128, each one made his or her comment.[85] Secondly, the use of he alone: If anybody behaves in such and such a manner, he will be punished (cf. the wholly unobjectionable, but not always applicable, formula: Whoever behaves in such and such a manner will be punished). This use of he has been legalized by the Act 13 and 14 Vict., cap. 21. 4: “That in all acts words importing the masculine gender shall be deemed and taken to include females.” Third, the sexless but plural form they may be used. If you try to put the phrase, ‘Does anybody prevent you?’ in another way, beginning with ‘Nobody prevents you,’ and then adding the interrogatory formula, you will perceive that ‘does he’ is too definite, and ‘does he or she’ too clumsy; and you will therefore naturally say (as Thackeray does, P 2. 260), “Nobody prevents you, do they?” In the same manner Shakespeare writes (Lucr. 125): “Everybody to rest themselves betake.” The substitution of the plural for the singular is not wholly illogical; for everybody is much the same thing as ‘all men,’ and nobody is the negation of ‘all men’; but the phenomenon is extended to cases where this explanation will not hold good, as in G. Eliot, M. 2. 304, I shouldn’t like to punish any one, even if they’d done me wrong. (For many examples from good writers see my MEG. ii. 5, 56.)

The English interrogative who is not, like the quis or quæ of the Romans, limited to one sex and one number, so that our question ‘Who did it?’ to be rendered exactly in Latin, would require a combination of the four: Quis hoc fecit? Quæ hoc fecit? Qui hoc fecerunt? Quæ hoc fecerunt? or rather, the abstract nature of who (and of did) makes it possible to express such a question much more indefinitely in English than in any highly flexional language; and indefiniteness in many cases means greater precision, or a closer correspondence between thought and expression.

XVIII.—§ 7. Nominal Concord.

We have seen in the case of the verbs how widely diffused in all the old Aryan languages is the phenomenon of Concord. It is the same with the nouns. Here, as there, it consists in secondary words (here chiefly adjectives) being made to agree with principal words, but while with the verbs the agreement was in number and person, here it is in number, case and gender. This is well known in Greek and Latin; as examples from Gothic may here be given Luk. 1. 72, gamunan triggwos weihaizos seinaizos, ‘to remember His holy covenant,’ and 1. 75, allans dagans unsarans, ‘all our days.’ The English translation shows how English has discarded this trait, for there is nothing in the forms of (his), holy, all and our, as in the Gothic forms, to indicate what substantive they belong to.

Wherever the same adjectival idea is to be joined to two substantives, the concordless junction is an obvious advantage, as seen from a comparison of the English ‘my wife and children’ with the French ‘ma femme et mes enfants,’ or of ‘the local press and committees’ with ‘la presse locale et les comités locaux.’ Try to translate exactly into French or Latin such a sentence as this: “What are the present state and wants of mankind?” (Ruskin). Cf. also the expression ‘a verdict of wilful murder against some person or persons unknown,’ where some and unknown belong to the singular as well as to the plural forms; Fielding writes (TJ 3. 65): “Some particular chapter, or perhaps chapters, may be obnoxious.” Where an English editor of a text will write: “Some (indifferently singular and plural) word or words wanting here,” a Dane will write: “Et (sg.) eller flere (pl.) ord (indifferent) mangler her.” These last examples may be taken as proof that it might even in some cases be advantageous to have forms in the substantives that did not show number; still, it must be recognized that the distinction between one and more than one rightly belongs to substantival notions, but logically it has as little to do with adjectival as with verbal notions (cf. above, Ch. XVII § 11). In ‘black spots’ it is the spots, but not the qualities of black, that we count. And in ‘two black spots’ it is of course quite superfluous to add a dual or plural ending (as in Latin duo, duæ) in order to indicate once more what the word two denotes sufficiently, namely, that we have not to do with a singular. Compare, finally, E. to the father and mother, Fr. au père et à la mère, G. zu dem vater und der mutter (zum vater und zur mutter).

If it is admitted that it is an inconvenience whenever you want to use an adjective to have to put it in the form corresponding in case, number and gender to its substantive, it may be thought a redeeming feature of the language which makes this demand that, on the other hand, it allows you to place the adjective at some distance from the substantive, and yet the hearer or reader will at once connect the two together. But here, as elsewhere in ‘energetics,’ the question is whether the advantage counter-balances the disadvantage; in other words, whether the fact that you are free to place your adjective where you will is worth the price you pay for it in being always saddled with the heavy apparatus of adjectival flexions. Why should you want to remove the adjective from the substantive, which naturally must be in your thought when you are thinking of the adjective? There is one natural employment of the adjective in which it has very often to stand at some distance from the substantive, namely, when it is predicative; but then the example of German shows the needlessness of concord in that case, for while the adjunct adjective is inflected (ein guter mensch, eine gute frau, ein gutes buch, gute bücher) the predicative is invariable like the adverb (der mensch ist gut, die frau ist gut, das buch ist gut, die bücher sind gut). It is chiefly in poetry that a Latin adjective is placed far from its substantive, as in Vergil: “Et bene apud memores veteris stat gratia facti” (Æn. IV. 539), where the form shows that veteris is to be taken with facti (but then, where does bene belong? it might be taken with memores, stat or facti). In Horace’s well-known aphorism: “Æquam memento rebus in arduis servare mentem,” the flexional form of æquam allows him to place it first, far from mentem, and thus facilitates for him the task of building up a perfect metrical line; but for the reader it would certainly be preferable to have had æquam mentem together at once, instead of having to hold his attention in suspense for five words, till finally he comes upon a word with which to connect the adjective. There is therefore no economizing of the energy of reader or hearer. Extreme examples may be found in Icelandic skaldic poetry, in which the poets, to fulfil the requirements of a highly complicated metrical system, entailing initial and medial rimes, very often place the words in what logically must be considered the worst disorder, thereby making their poem as difficult to understand as an intricate chess-problem is to solve—and certainly coming short of the highest poetical form.

XVIII.—§ 8. The English Genitive.

If we compare a group of Latin words, such as opera virorum omnium bonorum veterum, with a corresponding group in a few other languages of a less flexional type: OE. ealra godra ealdra manna weorc; Danish alle gode gamle mænds værker; Modern English all good old men’s works, we perceive by analyzing the ideas expressed by the several words that the Romans said really: ‘work,’ plural, nominative or accusative + ‘man,’ plural, masculine, genitive + ‘all,’ plural, genitive + ‘good,’ plural, masculine, genitive + ‘old,’ plural, masculine, genitive. Leaving opera out of consideration, we find that plural number is expressed four times, genitive case also four times, and masculine gender twice;[86] in Old English the signs of number and case are found four times each, while there is no indication of gender; in Danish the plural number is marked four times and the case once. And finally, in Modern English, we find each idea expressed once only; and as nothing is lost in clearness, this method, as being the easiest and shortest, must be considered the best. Mathematically the different ways of rendering the same thing might be represented by the formulas: anx + bnx + cnx = (an + bn + cn)x = (a + b + c)nx.

This unusual faculty of ‘parenthesizing’ causes Danish, and to a still greater degree English, to stand outside the definition of the Aryan family of languages given by the earlier school of linguists, according to which the Aryan substantive and adjective can never be without a sign indicating case. Schleicher (NV 526) says: “The radical difference between Magyar and Indo-Germanic (Aryan) words is brought out distinctly by the fact that the postpositions belonging to co-ordinated nouns can be dispensed with in all the nouns except the last of the series, e.g. a jó embernek, ‘dem guten menschen’ (a for az, demonstrative pronoun, article; jó, good; ember, man, -nek, -nak, postposition with pretty much the same meaning as the dative case), for az-nak (annak) jó-nak ember-nek, as if in Greek you should say το ἀγαθο ἀνθρώπῳ. An attributive adjective preceding its noun always has the form of the pure stem, the sign of plurality and the postposition indicating case not being added to it. Magyars say, for instance, Hunyady Mátyás magyar király-nak (to the Hungarian king Mathew Hunyady), -nak belonging here to all the preceding words. Nearly the same thing takes place where several words are joined together by means of ‘and.’”

Now, this is an exact parallel to the English group genitive in cases like ‘all good old men’s works,’ ‘the King of England’s power,’ ‘Beaumont and Fletcher’s plays,’ ‘somebody else’s turn,’ etc. The way in which this group genitive has developed in comparatively recent times may be summed up as follows (see the detailed exposition in my ChE ch. iii.): In the oldest English -s is a case-ending, like all others found in flexional languages; it forms together with the body of the noun one indivisible whole, in which it is sometimes impossible to tell where the kernel of the word ends and the ending begins (compare endes from ende and heriges from here); only some words have this ending, and in others the genitive is indicated in other ways. As to syntax, the meaning or function of the genitive is complicated and rather vague, and there are no fixed rules for the position of the genitive in the sentence.

In course of time we witness a gradual development towards greater regularity and precision. The partitive, objective, descriptive and some other functions of the genitive become obsolete; the genitive is invariably put immediately before the word it belongs to; irregular forms disappear, the s ending alone surviving as the fittest, so that at last we have one definite ending with one definite function and one definite position.

In Old English, when several words belonging together were to be put in the genitive, each of them had to take the genitive mark, though this was often different in different words, and thus we had combinations like anes reades mannes, ‘a red man’s’ | þære godlican lufe, ‘the godlike love’s’ | ealra godra ealdra manna weorc, etc. Now the s used everywhere is much more independent, and may be separated from the principal word by an adverb like else or by a prepositional group like of England, and one s is sufficient at the end even of a long group of words. Here, then, we see in the full light of comparatively recent history a giving up of the old flexion with its inseparability of the constituent elements of the word and with its strictness of concord; an easier and more regular system is developed, in which the ending leads a more independent existence and may be compared with the ‘agglutinated’ elements of such a language as Magyar or even with the ‘empty words’ of Chinese grammar. The direction of this development is the direct opposite of that assumed by most linguists for the development of languages in prehistoric times.

XVIII.—§ 9. Bantu Concord.

One of the most characteristic traits of the history of English is thus seen to be the gradual getting rid of concord as of something superfluous. Where concord is found in our family of languages, it certainly is an heirloom from a primitive age, and strikes us now as an outcome of a tendency to be more explicit than to more advanced people seems strictly necessary. It is on a par with the ‘concord of negatives,’ as we might term the emphasizing of the negative idea by seemingly redundant repetitions. In Old English it was the regular idiom to say: nan man nyste nan þing, ‘no man not-knew nothing’; so it was in Chaucer’s time: he neuere yet no vileynye ne sayde In all his lyf unto no manner wight; and it survives in the vulgar speech of our own days: there was niver nobody else gen (gave) me nothin’ (George Eliot); whereas standard Modern English is content with one negation: no man knew anything, etc. That concord is really a primitive trait (though not, of course, found equally distributed among all ‘primitive peoples’) will be seen also by a rapid glance at the structure of the South African group of languages called Bantu, for here we find not only repetition of negatives, but also other phenomena of concord in specially luxuriant growth.

I take the following examples chiefly from W. H. I. Bleek’s excellent, though unfortunately unfinished, Comparative Grammar, though I am well aware that expressions like si-m-tanda (we love him) “are never used by natives with this meaning without being determined by some other expression” (Torrend, p. 7). The Zulu word for ‘man’ is umuntu; every word in the same or a following sentence having any reference to that word must begin with something to remind you of the beginning of umuntu. This will be, according to fixed rules, either mu or u, or w or m. In the following sentence, the meaning of which is ‘our handsome man (or woman) appears, we love him (or her),’ these reminders (as I shall term them) are printed in italics: