Analogy, then, in most cases acts as a conservative agent in language by securing that its propagation and its continuity shall be subject to some degree of regularity. On the other hand, this very tendency to promote regularity and uniformity often makes itself felt by the destruction of existing words or flections which deviate from a given goal; and it is mainly when its destructive powers are manifest that its effects are deserving of separate discussion.
So long as a speaker employs or a nation continues to use the ‘correct’ form,—gradually, regularly, and naturally developing it according to the regular laws of phonetic change and growth to which it is subject for the time being,—it is immaterial for the student of language whether, in any particular case of the employment of a word, this regularity is due to memory or to analogy. It is when analogy produces forms phonetically irregular that its operation becomes of importance; and it is from the study of such ‘novelties’ amongst its productions, that we can alone derive full information about its nature. As long as we find that the A.S. stánas remained stánas, or even that this form was gradually changed into stones, we are not tempted to call in the aid of Analogy, nor are we challenged to prove its operation. Similarly, as long as the plural of eáge remains eágan, or eáge changes into eye, and forms its plural eyen, no temptation presents itself to inquire into Analogy or its operation. Even in this case, however, we cannot help remarking that Chaucer might conceivably have formed his plural eyen by analogy with other plurals in en. But it is when the form eyen is replaced by eyes, that we naturally inquire whence comes the s? And since no phonetic development can change n into s, we know that analogy with other substantive plurals is and must be the reason of the appearance of this otherwise inexplicable form. Thus the French mesure could and did become the English measure; but the French plaisir could not, according to the laws of phonetics, develop into pleasure. We can only explain the latter form by assuming that it is founded on the analogy of the older forms measure, picture, etc.[23]
We ascribe to Analogy those cases of change in form of words, in syntactical arrangement, or in any other phenomenon of language, such as gender, etc., where the existing condition has been replaced by something new modelled upon some pattern furnished by other more numerous groups. Thus, for instance, we find that the Latin feminine nouns in -tas, -tatis, have developed French derivatives in -té, all of the feminine gender. Why, then, is été masculine, though equally derived from a feminine Latin æstatem? The answer lies in the fact that printemps, automne, and hiver, being all masculine, the feeling set in that the ‘names of the seasons’ should be masculine: just as names of trees are feminine in Latin, and this possibly under the influence of arbor. Thus été followed the example of the others, and was classed with them. The affinity in signification here caused the difference in gender to be felt as an incongruity, and the less strong came to be assimilated to the stronger and more universal type. Similarly, such words as valeur seem to have become feminine after the analogy of Latin abstracts in -ura, -tas, etc. In the former of these particular instances we had to deal with a ‘MATTER-GROUP’ of four cognate ideas, viz. ‘the seasons;’ in which group, as three of the terms agreed in another accidental peculiarity, viz. that of gender, this peculiarity was imposed likewise upon the fourth member, so as to produce a more complete uniformity in every respect.
In other cases we find, perhaps indeed more frequently, MODAL groups thus extending their domain. Thus the comparative forms, which nearly all end in er, create the feeling that if a word expresses a comparative degree it may be naturally expected to end in er; and more from mo, lesser instead of less—nay, even worser for worse is the result. In the case of more, its very form led to the supposition that mo was a positive form.
Similarly, the existence of the plurals in s in Anglo-Saxon, aided no doubt by the frequency of s plurals in French, has caused this way of expressing the plural to embrace almost all English nouns; or, at all events, to embrace their formation to such an extent that the older methods (such as vowel modification, e.g. mouse, mice; foot, feet; formations in en—ox, oxen, etc.) now appear as exceptions, themselves needing explanation; and, again, as in the case of more, when once the rule was formulated which laid down that if a word expresses the plural it must end in s, the conclusion was drawn that, if a word ending in s be used as a plural, this s is the termination, and must be omitted in the singular. It thus happens that to the analogy of fathers as against father, trees as against tree, etc., we owe the sets Chinese used as a plural noun with its newly coined singular Chinee; Portuguese with its singular Portuguee; cherries (Fr. cérise), cherry; pease (Lat. pisum), pea. Nay, it is not even always necessary that the s form be used in a plural signification to cause the s to be ‘removed’ in order to express the singular; a raedels was perfectly good Old English, but as two riddles was right, the conclusion was natural that one riddles was wrong. Two chaise would not give offence, but it seemed natural to write and say one shay.
The modal group, again, consisting of such formations as despotism, nepotism, patriotism, etc., created the feeling that tism was the correct ending instead of ism, and so has manifested a tendency to supplant it. Thus the correcter form egoism has made way for egotism. Thus it is to the pianist, machinist, violinist, that the tobacconist owes his n, to which he has no right; he ought, properly speaking, to appear as tobaccoist.
The most widely reaching result of the operations of analogy is where modal and matter groups, in their cross classifications, unite to cancel irregularities created in the first instance by phonetic development. Thus the Anglo-Saxon form scæd (neuter) exists side by side with another form, sceadu (feminine). The Gothic form skadus proves the latter to belong to the u declension. But even in Anglo-Saxon this declension was but sparingly represented, most words originally belonging to it being declined according to the far more common scheme of words, like stán, stone; dóm, doom, etc.; others varying in their declensions between the feminines whose stem ended in wâ, or like those in â. In both these declensions the nominative ended in u; an example of the wâ declension being— Nom. beadu, Gen. beadwe,
and of the â declension— Nom. giefu, Gen. giefe.
Our word sceadu long oscillated between these two paradigms, and we consequently meet with a Gen. sing. sceade, as well as an Acc. plur. sceadwa. This termination, where w was maintained, developed into our present termination ow, seen in shadow; whilst the form shade is, properly speaking, a nominative form. Analogy, however, depending upon other nouns in which all cases in the singular had become identical in form, caused the form shadow to be used in the nominative as well as in other cases, and extended the use of shade over those cases which were declined. Similarly, the two forms mead and meadow are due, the one to a nominative, the other to the inflected cases of the same word, the A.S. mǽd. In these cases both forms survived, and the meanings became slightly differentiated; it more frequently happens that one succumbs. Thus the A.S. Nom. plur. of the pronoun for the second person gé developed into ye, the inflected case éow into you. The latter has now almost completely ousted the once correct nominative ye, which survives only in dialects or in elevated language, where, in its turn, it frequently supplants the accusative and dative you.
The regular development of preterite and past participle in many verbs, together with the dropping of the prefix ge, which in several Teutonic languages has become specialised as a mark of that participle, caused both these forms to converge into one. This has in its turn been the cause why, in the case of many verbs, where regular phonetic development kept preterite and participle asunder, one of these forms was made to serve for both.
The A.S. verb berstan was, in its preterite, conjugated thus:— Indic. Bærst Subj. burste ” burste ” burste
” bærst ” burste ” burston ” bursten ” burston ” bursten ” burston ” bursten and its past participle was borsten. Thus the u was present in four of the six forms in the indicative, and in six subjunctive forms. The first effect of the operation of Analogy was to abolish this useless and cumbersome irregularity, and the u supplanted the æ, not long after this æ had become a (barst). Then the process set in which we explained above, and the past part. borst (en) was replaced by burst.
It would be easy to multiply these instances ad infinitum. Enough has, however, been said to explain the working of Analogy and to show how wide its application is. The student who has mastered this sketch, should proceed to study carefully the corresponding chapter in Paul’s ‘Principles of Language,’ and the pamphlet, cited above, by Professor Wheeler, where many illustrations will be found taken from English and many other languages. One of the main points which are clearly brought out in the latter work is that the phenomena of folk-etymology show that these groupings are effectual in modifying form only in so far as a supposed likeness of contents or idea is associated (erroneously) with the resemblance of form.