It is, however, not always the derivative which, in its isolation, assumes the modified signification. The primitive may change, and the derivative remain stationary. Thus the English shop, as a place for retail trade, has been displaced in America by store, while shop comes to have the value of work-shop, machine-shop, etc. Yet the derivative shopping, a much-used word in America, retains a reminiscence of the older value of shop.
To return for a moment to the example which we gave from German: the verb reiten (pronounced with a vowel sound closely resembling that of i in to ride) and the noun ritter (i nearly like i in rid, or, more correctly, like ee of need, but shortened), show a gradation of vowel-sound, of the same nature and origin as that in such pairs as write, wrote; sing, sang; give, gave. This change in vowel-sound without doubt co-operated in effecting the isolation, and so facilitated the change in meaning in the one form; a change in which the other did not participate. Thus, speaking generally, phonetic development, by creating numerous meaningless distinctions, loosens the modal and material groups, and serves to forward isolation of meaning. Thus, again, the special meaning which we now attach to the verb to rear would have been more likely to transfer itself to the primary verb to rise, or—vice versâ—the meaning of the primary to rise would have almost certainly prevented the special development of to rear, if the etymological connection had not been obscured by the phonetic development which we formulate as Verner’s law, i.e. if the grouping had not been loosened.
It is, moreover, clear that if, from whatever cause, an interchange of certain sounds becomes less frequent in a language, those words which do preserve that interchange become ipso facto more strongly separated. Thus, e.g., the umlaut, i.e. the change of u (sounded as oo) to ü (sounded as u in French, the Devonshire u; more like English ee than like English u), or of a (a as in father) to ä (sound much like a in fate, but without the ee sound which in English follows it), etc., is in German so common that in no case is its presence or absence alone sufficient to effect the isolation of any form from its related group. In English, this interchange has almost completely disappeared, and the few traces of it which we preserve in the plural formation (foot, feet; tooth, teeth; mouse, mice; man, men, etc.) are only preserved as so-called ‘irregularities,’ and no longer form a model or pattern for other formations. Hence in English, where, besides umlaut, we have difference in function (e.g. adjective and noun), the isolation has often been complete. Thus, no ordinary speaker groups the adjective foul with the noun filth; and the connection, though still felt, between long and length, broad and breadth, is undoubtedly less clearly felt than between, e.g., long and longer, or broad and to broaden, high and height: similarly, the difference in vowel between weal and wealth, (to) heal and health, has facilitated isolation of these forms.
If phonetic development were the only agent in the history of language, we see that, shortly, an infinite variety of forms, absolutely unconnected, or at best but loosely connected, would be the result. But here, as always, we have action and counteraction.[92] This counteracting influence is chiefly exerted by analogy, as we explained in Chapter V. It is, however, not always analogy which brings about the readjustment or unification.
We have already had occasion to point out that our word-division, though undoubtedly based on real and sufficient grounds, is not consistently or even commonly observed in SPEAKING. Our thoughts are, indeed, expressed not in words but in word-groups; and letters, even though they stand at the end or at the beginning of words, have often had a special phonetic development, in cases where these words occurred in very frequent or in very intimate connection with other words. The differences so created have very commonly, though not by any means universally, found expression in writing. As an instance of a differentiation of which the written language takes no cognisance, we may take the French indefinite article. Few are unaware that when un stands before a consonant the n is not pronounced, leaving in the spoken word only a trace of its existence in the fact that the vowel is nasalised. When un comes before a vowel, on the other hand, the vowel is much less strongly, if at all, nasalised, and the n is clearly pronounced. Thus (using the circumflex to indicate the nasal quality of the vowel and ö for the sound of u in un), un père = ö̂ père, but un ami = ön ami or ö̂n ami. The corresponding difference which exists in English is expressed in writing: a father, an aunt.
Just as the article is closely connected with the noun, so preposition and noun, or preposition and verb, are very intimately connected in pronunciation. Hence—though many, who have never carefully observed either their own pronunciation or that of others, may dispute or deny the assertion—in ORDINARY conversation, in the phrases, in town, in doors, we employ the n sound; but when the word in stands before Paris and Berlin, we use an m sound, just as we say impossible by the side of interest. Similarly, we pronounce generally ‘in coming’ with ng for n, just as we speak of a man’s ingcome. This differentiation of the pronunciation of the preposition in into three forms—in, im, ing—is not, however, consistently expressed by us in writing. The Greeks, on the other hand, who similarly differentiated the terminal consonants of the prepositions in their spoken language, but on a much larger scale (accustomed as they were to a far closer correspondence between their spoken and their written language than the Englishman observes), did actually, in many cases, write as they spoke: κάδ δὲ,—κὰκ κεφαλὴν, κὰγ γόνυ—κὰπ πεδιόν, etc., instead of employing the normal form of the preposition, κατά. So we find in inscriptions τὴμ πόλιν, τὴγ γυναῖκα, τὸλ λογόν, ἐμ πόλει, etc.
The first step on the road towards unification is frequently that the external reason which caused the difference in form, disappears or loses force, and one form is found in connections where, historically or phonetically speaking, the other is correct. We may instance this by the common mistake of children when they say, e.g., a apple instead of an apple. In this case, however, the correct form is so very frequently heard that the encroachment of a on the domains of an is not likely to lead to permanent confusion. Where, however, circumstances are less favourable to the preservation of the historically correct usage, it happens that either form encroaches on the domain of the other, or else it may result that the encroachment is reciprocal,—when, after a period of confusion in which both forms are used indifferently, one becomes obsolete and falls into oblivion, not without often leaving some striking form or phrase to testify to what once existed. Thus, for instance, our word here, Old High German hier, or hêr, was, in the period of transition from Old to Middle High German, differentiated in accordance with a phonetic law of that time, viz. that final r was dropped after a long vowel. If not final however, r remained untouched, and this whether it stood in the body of a word or within a group of intimately connected words. Of the two forms hie and hier, the former, as the form employed when the word was used independently, was in Middle High German often set before words beginning with a vowel; and we find hie inne (= here-in) or even, by contraction, hinne, for hier-inne. On the other hand, it is probably owing to the frequency of combinations similar and equivalent to our here-in, here-upon, etc., that the form hier encroached successfully upon the domain of hie, and finally supplanted it. Hie, however, remained, singularly enough, in the one expression hie und da (here and there), where the form without r is not and has never been, phonetically speaking, correct. An excellent example of this differentiation is furnished by one, an.
The best example of the process is furnished by the history of the working of Verner’s law, and of the gradual disappearance of its effects. We have before (pp. 172, 173) explained this law and quoted instances of forms created in agreement with it, which have now been replaced by others. To repeat this explanation here with other examples would be superfluous; to give a full history, even confining ourselves to an enumeration of all the various ways in which it has been operative and the areas of its influence, would transcend the scope of this work. To carefully note all instances of its occurrence and its neglect, and to closely investigate the possible courses of the latter, is a task which may most usefully challenge the attention of philologists. We will illustrate the truth of this by a single example: (though even this we cannot discuss exhaustively). The forms which we employ at present as the past tense of the verb to be—sing. was and plur. (with grammatical change according to the law) were, belong to a root which in old English and Anglo-Saxon furnished a complete verb: pres. wese, past. wæs, p. part. wesen. Now we should naturally expect that in a time when the grammatical change was still preserved in freóse, fréas, fruron, froren, (to freeze) etc. ceóse, céas, curon, coren, (to choose) seóðe, seáð, sudon, soden, (to seethe, to boil) we should also find that change here, and that accordingly the past participle should be *weren. That such a form once existed is proved by the past participle forweorone (cf. Sievers, Anglo-Saxon Grammar, § 391). Everywhere, however, in Anglo-Saxon, in the past participle of this verb and in that of all similarly conjugated, such as lesan, læs, lesen; genesan, genæs, genesen, etc., the s has once more been fully established. The fact that these past participles had already so far proceeded on the road to unification, while the others as yet remained isolated, may be explained in this way,—the latter, IN ADDITION to the differentiation in accordance with Verner’s law, showed a difference of vowel-sound, which in the case of others did not exist. Hence the forms differentiated in two distinct ways were able to resist the tendency towards unification long after those which differed only in one respect had succumbed. In fact, of the former we still have such remnants as forlorn, from to lose; sodden, from to seethe. We may formulate the result which we have illustrated, thus: The greater the phonetic distance of two differentiated forms, the greater is the power of resistance against unification and equalisation.
But the ORDER in which we see the traces of the working of Verner’s law disappear one after another, and the study of such few remnants as still exist, brings out two other general truths concerning unification. We may without hesitation affirm that, close as is the etymological connection between the various tenses of the same verb, or, to speak perhaps more correctly, that clearly as that connection is felt by the speech-making community, it is still more strongly felt as between the various forms of the same tense, or the various cases of the same noun. Now, it is against the differentiation between the members of these most intimate groups that unification first takes place. In the declension of the noun, where nothing but the operation of Verner’s law had separated the various cases, the re-assimilation first took place, and though we can prove that, in this case also, the differences actually once existed—in the historic periods of the Teutonic dialects almost all traces thereof have been obliterated. In the past tenses of the verbs they are still at first found, supported as the differentiation had been by that other force—the gradation of vowels (the ‘ablaut’).[93] But again: unification between the singular and plural of the past tense took place first in cases where the vowels were alike in both, and next in those where the vowels differed—and again, this occurred before the unification of the past participle with the whole group. In agreement with this same rule, that very difference of vowel-sound has completely disappeared in all past singulars and plurals, even where—as, e.g., in German generally—the past participle still preserves the ‘ablaut.’
We can then lay it down as a second rule, that the closer the etymological connection is between differentiated forms, the sooner will unification be effected; whilst a consideration of such rare instances as the preservation of the interchange of s and r in I was, we were, which is clearly due to the very exceptional frequency with which these forms must always have been used, and the consequent firmness with which they are impressed on every speaker’s memory, exhibits a third law, viz. that the greater the intensity with which differentiated forms are impressed upon the minds of the community, the greater will prove their power of resistance against unification.