Chapter XI · THE FORMATION OF NEW GROUPS.

The effect of sound-change is to produce differences in language where none previously existed; but it likewise tends to cancel existing differences, and to cause forms originally distinct to resemble each other or actually to coincide. Now, symmetry and uniformity are clearly an aid to the memory, when attained by the abolition of useless and purposeless differences. It is, for instance, in English, far simpler to state, and far more easy to remember the statement, that all plurals are formed by adding s to the singular, than that some are formed in -n, or -en, or by such modifications as man, men; foot, feet; etc.: and it is therefore a gain to language when such forms as shoon, eyen, etc., disappear in favour of such forms as shoes, eyes, etc. On the other hand, the cancelling of such differences when they serve to mark different functions is naturally disadvantageous and tends to obscurity. When a sound which marked such a functional difference disappears, or when of two words or forms which had different meanings one becomes obsolete, and the other is employed to do service for both, it is clear that language cannot but be the loser by dispensing with an important aid to clearness and distinction. Thus, of the two forms mot and moste, the former has now disappeared, and the latter, in the form must, serves to indicate both the present and the past tense. The effect of this ambiguity is that where we wish to clearly indicate the past of must, we have to employ some idiom in which must has no place; as ‘was obliged to,’ ‘had to,’ ‘was constrained to,’ etc. Similarly, the loss of the plural s in very many French nouns (which s, though still written, is seldom sounded) would create ambiguity were it not that the difference of the article attached to the noun marks the difference, and to a large extent remedies the evil; cf. l’ami, les amis.

The remedy, however, for such obscurity is not always to be found in the context. Sometimes, indeed, the evil brings its own cure; changes arise which enable the necessary distinctions to be once more felt and maintained, creating new forms by analogy with other forms (see Chapter V.): but, on the other hand, it frequently occurs that the evil remains, and a confusion follows in the grouping of the words; which grouping, as we have seen, is all-important in the life history of the members of the group.

We must in this chapter endeavour to study some of the results of this confusion, and consequent re-arrangement in the groups; and to distinguish the cases where similarity caused by phonetic development affects the matter-groups from those where the modal-groups are influenced.

I. i. There are many cases where words connected neither by etymology nor by signification fall into the same form.

Still, in spite of this similarity in form, the words remain perfectly distinct in the linguistic consciousness of a speaker of ordinary intelligence. Such are, e.g.,—

1. a. Hale, in such a phrase as hale and hearty. This word is of Scandinavian94 origin (cf. Icelandic heill), and represents the Anglo-Saxon hál, to which word we owe the misspelt word whole. b. Hale, ‘to drag,’ found in Middle-English as halien.

2. a. Whole = A.S. hál; see above. b. Hole = A.S. hol, ‘a cave.’95

3. a. Grave (A.S. gráfan). b. Grave (Fr. grave, Lat. gravem).96

4. a. Cope (O.Fr. cape). b. Cope (Dutch koopen = to bargain, to chaffer, to buy, to vie with).

5. a. Stile (A.S. stigel). b. Stile (commonly misspelt style, Lat. stilum).

6. a. Well, adverb (A.S. wel). b. Well, noun (A.S. wella).

7. a. Arm (Lat. arma). b. Arm, the limb, cognate with Ger. arm.

8. a. Lay (A.S. lecgan). b. Lay (O.Fr. lais, ‘song’).

9. a. Pale (Fr. pal, Lat. pāum). b. Pale (Fr. pâle, Lat. pallidum).

10. a. Elder, the tree (A.S. ellarn). b. Elder, ‘older.’

It would, of course, be possible to extend this list to almost any length; but this would be useless for our purpose, which is to investigate solely those cases in which similarity causes confusion. This happens where the difference in origin and meaning is lost sight of. It is naturally impossible to draw a hard and fast line of demarcation between the case just discussed and that which we are about to exemplify, as one speaker may keep distinct what another may confuse or treat as identical. Still, no one, we may fairly say, unless he be a student of language, or unless he has been expressly informed, is aware that in a phrase like The ship is bound for London, the word bound employed by him has absolutely no connection with the past participle of the verb to bind. In the first case, bound is of Scandinavian origin, and meant originally ready, prepared; cf. the Icelandic verb búa, perf. part, búinn, ‘to prepare.’ Similarly, few ordinary speakers can explain, or indeed realise, the existence of the distinction in meaning between shed, ‘a hut’ (a doublet of shade), and shed in water-shed, when derived from the A.S. scéadan; or that between sheer, allied to Icelandic skærr, ‘bright,’ and sheer, akin to Dutch scheren, ‘to shave.’ Thus, again, many might suppose that some etymological connection existed between hide, ‘a skin’ (A.S. hýd, akin to Ger. haut), and hide, ‘to conceal’ (A.S. hídan); while others, when told that hide also served as the name for a certain measure of land, might naturally even suspect some allusion to the famous legend of the foundation of Byrsa or Carthage. The A.S. noun setl (a seat) and the verb settan survive both in the word settle and in to settle. In employing, however, the word in ‘to settle a dispute,’ we have a word of very different origin: the A.S. sacu, ‘a quarrel,’ ‘dispute,’ ‘lawsuit’ (surviving in ‘for my sake’, etc.), existed side by side with a verb sacan, ‘to strive,’ or ‘dispute:’ akin to this, we find saht, a substantive which owes its meaning, ‘reconciliation,’ to the development lawsuit, adjustment by lawsuit, etc. Again, derived from this we have the verb sahtlian, ‘to reconcile,’ which, at a later period, occurs in the forms saztlen and sattle.97 When this verb ceased to be understood, confusion with the other verb to settle = to fix, to arrange, arose, and the two forms ‘flowed together, just as two drops of rain running down a window-pane are very likely to run into one.’98 Another instance of this nature is discussed by Professor Skeat, s.v.; viz., sound = A.S. sund, akin to the Ger. (ge)sund; sound, ‘a strait of the sea,’ and sound’ M.E. soun, Anglo-Fr. soun or sun, Lat. sonum.

ii. Such forms, where phonetic development brought about merely a close resemblance without producing perfect similarity, and where, as a next step, one or other of the set of words underwent some change more or less violent in consequence of its supposed connection with the rest, are peculiarly instructive, proving as they do the confusion which arose in the minds of the speakers who thus combined what was distinct and unconnected. In these cases we have entered upon the domain of ‘popular etymology,’ to which we have already incidentally alluded.

It does not, however, always follow that the supposed connection in meaning—in other words, the coalescence of elements of different origin into a single material group, brings about the further change in form; at this period nothing but the linguistic consciousness of the speaker can decide whether the ‘popular etymology’ is or has been at work. Of course, as long as the etymology of the different words in the set is clearly understood by the speaker, there can be no question as to the connection, but when one or more of the members of the set is no longer understood in its historical bearings, it is possible for a new grouping to arise.

Let us take, as an instance, the word carousal. This bore originally the sense which it bears in the Parisian name of the Place du Carrousel, viz. a tournament or festival. It was confused with the word carouse (Ger. gar-aus = properly ‘quite out,’ i.e. ‘empty your glasses’); and at present our word carousal represents both. The Anglo-Saxon word bonda meant a boor, or householder. His tenure appears expressed in Low Latin by the word bondagium, and it is only to a supposed, but wholly erroneous connection with bond and the verb to bind, that our present word bondage owes its sense of servitude.

The Fr. sursis gave us, before its final s had ceased to be pronounced, our verb surcease, which most speakers now look on as a compound of cease (Fr. cesser).99 Wiseacre, really derived through the Dutch from the Ger. wízago (A.S. witega, ‘a prophet’), was already, while on its way to England, misunderstood in Holland, and taken to be a compound of wise. In Dutch, a verb wys-seggen and a noun wys-segger (‘to speak wisely’ and ‘a wise sayer’) were formed, and modern German as well possesses the word weissagen, ‘to prophesy.’ This wys-segger, when it reached England, could no longer be understood as a derivative from the verb secgan, which in English had already lost its guttural and had become (to) say; and thus popular etymology altered the second part of the supposed compound into the meaningless acre. The Fr. surlonge, the piece of meat ‘upon the loin’ (Lat. super, Fr. sur, and Lat. *lumbea, from lumbus, Fr. longe), became in English the surloyn in the time of Henry VI. This was no longer understood; the word was accepted as a compound with the word sir, and thus the fable was invented of the ‘merry monarch’ knighting the loin.100 The berfroit or belefreit of Old French is of German origin, and signifies a watchtower. The word had ceased to be understood, and its origin was forgotten; but, as many towers contained a bell or a peal of bells, a supposed connection with these bells caused the word to be changed into belfry. The spelling is affected in sovereign, where the g is due to a supposed connection with to reign (régner, regnare); the real derivation being from soverain (superaneum), and the word being correctly spelt sovran by Milton. Further instances are lance-knight (= lanz-knecht = landes knecht = ‘the knight, i.e. the man-of the land,’ ‘the servant of his country’); cray-fish (= écrévisse); shamefaced (really shamefast, like steadfast), etc.

In other cases of rarer occurrence than those which we have discussed, a significant part of a compound assumes the form of a mere derivative. This has occurred in the case of the word righteous, taken to be a derivative from some French adjective in -eux, Lat. -osus, though really due to right-wise, a compound like otherwise. It is natural that Proper nouns, where there is no connection or only a fanciful one between the word and its meaning, should be more liable to such transformations than others; so the Rose des quatre saisons appears as the quarter-sessions rose, the asparagus appears as sparrow grass, the ship Bellerophon becomes the Billy ruffan,101 the Pteroessa, the tearing hisser. We may perhaps add here a word like liquorice, which, though the name, rightly understood, is descriptive, has become a mere proper noun. Originally from liquiritia, itself a corrupt form of glykyrrhiza = ‘a sweet root,’ it has, as its spelling shows, become connected with liquor,102 while those who deemed this impossible preferred to explain the word as connected with to lick.103

II. Important, then, as the part played by phonetic development is in bringing about the formation of new material-groups, it has made its influence felt more widely still in the modal grouping of the various systems of inflection.

Here, again, two cases should be distinguished: (1) when forms which have had identical functions come to coincide: (2) when such coincidence occurs in the case of forms that have had different functions.

1. The cancelling of diversities in form or in inflection when such inflection indicated no difference in function must obviously on the whole be set down as a gain to language: simplicity is gained thereby without any loss in clearness. This gain, however, is only effected when the abolition is complete; should the abolition be partial only, simplification may be gained at the expense of a new confusion.

We have an example of such a complete process of cancelling in the terminations er and est in the comparative and superlative of adjectives. In Gothic the comparative was formed either with the suffix iz or ôz, the superlative with ist or with ōst; and, except, indeed, that the forms in iz and ist were more common than those in ôz and ôst, and that the latter are found only with stems in a, no rule can be given for their occurrence. Thus mānags (an a stem) has in its comparative managiz-a, superlative managists; alðeis (ja stem) alðiza, alðists; hardus (u stem), hardiza, hardists; but frôðs, frôdôza, frôdôsts; arms, armôza, armôsts.104 In Old High German there was a similar uncertainty. Here the z of Gothic appeared as r in the comparatives,105 and while salîg has for its comparative salîgôro and its superlative salîgôsto, we find (h)reini, (h)reiniro, (h)reinisto.106 In Anglo-Saxon we find already but a single termination for the comparative, viz. ra; but the two forms of superlative are still extant in ost and est; earm, earmra, earmost; heard, heardra, heardost; but eald, ieldra (with umlaut or modified vowel),107 ieldest. Our forms hard, harder, hardest; old, older, oldest; silly, sillier, silliest, etc., are clearly a further step in the right direction of simplicity in system.

The convergence is, however, not always complete: sometimes it happens that two systems coincide; and this coincidence may be (1) in ALL FORMS but only in SOME WORDS belonging to each system; or, again, (2) it may manifest itself in ALL WORDS but only in SOME FORMS; and, lastly, this coincidence may affect (3) only SOME WORDS in SOME FORMS of two converging systems.

In the case of (1) the convergence is complete and irrevocable, and words which formerly belonged to one system have simply parted company with it, and have definitely joined the other to which they were assimilated. In the cases, however, of (2) and (3), confusion must arise, and further development must be looked for. We find a good illustration of this confusion and of its development in the history of the Teutonic declensions. In the case of these, as of other Indo-European languages, the declensions differed as the stems of the words terminated in a consonant or a vowel; and amongst the latter, again, we must draw distinctions between the declension of stems in a, (o), i, and u. In the a declension, again, a subdivision arose for pure a, ja, wa, and long ā stems. These different terminations of the stems are, for instance, clearly preserved in Gothic dat. and acc. plur. dags, dagam, dagans; gasts, gastim, gastins; sunus, sunum, sununs; and (with Gothic ō instead of ā) gibā, gibōm, gibōs. In the oldest forms of Scandinavian, the so-called Ur-Norse, also, we find the vowels preserved in the nominative singular, holingar, erilar, etc., gastir, staldir, etc., haukoður, warur:108 but even in these, the oldest forms of the Teutonic dialects accessible to us, the various systems were confused; and it is the study of Comparative Grammar that we have to thank for the distinction between the different classes; and, again, it is only owing to the light shed on the subject by the comparison with Latin, Greek, and Sanskrit cognates, that we are enabled in some instances to decide to which of these classes any given word belongs. The ‘wearing down’ of the various terminations produced here identity, elsewhere close resemblance of many cases in many words, while in other cases the influence of the preceding letter made itself felt, and a difference in declension arose for the a stems: this difference depending on whether the a was preceded by a consonant i (j) or w. Where phonetic development had caused some of the cases to agree, other cases soon followed suit, and thus we find, for instance, that even in Gothic the entire singular of i declension has already become identical with that of the a stems:—

a stem. i stem. Sing. Nom. dags balgs Gen. dagis balgis Dat. daga balga Acc. dag balg Voc. dag balg Plur. Nom. dagôs balgeis Gen. dagê balgê Dat. dagam balgim Acc. dagans balgins.

As a consequence of this, numerous words which cognate languages prove to belong to the i declension are nevertheless entirely declined like a stems in Gothic; and even in the very few Gothic texts which we possess, and though these are derived from one source only, we meet with words evidencing the fact that Ulfilas himself (or, it may be, his copyist) was sometimes confused as to the declension usually followed by some word in his own language. Thus, in case of wêgs (a wave), we find norm plur. wêgôs, but dat. plur. wêgim; so too, the dat. plur. of aiws is aiwam, while the accus. is aiwins. In Old High German the coincidence in termination between these two schemes goes further, and extends over all cases; but since—in such words as had a, o, or u, in the preceding syllable—umlaut had been produced in the plural by the i of the stem, only those words whose stem vowel would not admit of umlaut or modification became throughout identical with the a declension. Where the reverse was the case, the words naturally remained distinct in the plural, and a further development arose; viz. that this umlaut in the plural began to be regarded as a sign of that number, and to be used for the purpose of marking it even in words whose etymology afforded no justification for the change, e.g. in hand, hände, which word originally belonged to the u declension. See also our remarks in Chapter V. pp. 87 and foll.

2. So far, in every case which we have discussed, we have had to do with similarity arising from phonetic development of forms with identical functions: one or more cases of one system converged with the same cases in another system. Often, however, this same phonetic development creates a similarity between forms which were originally distinct and served distinct purposes; and we have a good instance of this in our personal pronouns, and one which is instructive as to the consequences of this phenomenon:— The Gothic ik meina mis mik ðu ðeina ðus ðuk weis unsara uns uns jus izwara izwis izwis already shows no difference in the forms of accusative and dative plural; but in Anglo-Saxon we find that a further stage has been reached:— In ic mín mé mé ðú ðín ðé ðé wé úser ús ús gé eówer eów eów we see (though separate forms for accusative still occur) that dative and accusative have become identical throughout, and so it is in the modern language with— I mine me thou thine thee we our us ye (you) your you The double form of the nominative ye (you), and more especially the history of the pronoun for the third person, illustrate one of the consequences of such coincidence, viz. that the language-producing community becomes accustomed to use the same form for certain sets of functions, and transfers this similarity to cases which it would not reach—or, at least, has not yet reached—by the aid of phonetic development alone. Let us consider first the pronoun of the third person. In Anglo-Saxon we find— Sing. Masc. Fem. Neuter. Nom. hé heó hit Gen. his hire his Dat. him hire him Acc. hine hí hit. The forms which we now use for the plural are derived from a different stem,109 which in Anglo-Saxon gave us the following plural for all three genders:— Nom. ðá Gen. ðára, or ðǽra Dat. ðǽm Acc. ðá and here we find distinct forms for dative and accusative, the latter of which has now disappeared, so that here, too (as in the case of the other personal pronouns), we use one form only (the original dative form) for both dative and accusative. But we have only reached this stage after a period of confusion and uncertainty, during which the historically correct form of the accusative and the new form (that of the old dative) strove for permanence.

It is the very marked difference between ic (I) and me (accus.), ðu (thou) and ðe, we and us, which has protected the members of these pairs from becoming identical in form, notwithstanding the important fact that such a process had long since identified the nominative and accusative of all nouns and adjectives. To this influence, indeed, ye and you (both of which, when unemphatic, become ye, where e is pronounced as in the before a consonant) have succumbed.

Not only in this way, moreover, does such convergence of forms with different functions show its effect: it also causes the ordinary speaker to lose sight of such difference in function altogether. As students of Latin, and especially teachers of that language, know by sad experience, it is extremely hard for the untrained English mind to realise the function of the accusative case; and the difference between this case and the dative may be fairly described as non-existent for the Englishman who has not learnt it from the study of other languages. This, again, influences syntax, so that a phrase like I showed him the room can be turned in the passive into The room was shown (to) him, etc., or He was shown the room, etc.