Yet there is no doubt whatever that in a number of instances the new pronunciation is easier than its predecessor: we now say last instead of latst, examples of which earlier form may be found in the Ormulum, for instance. Similarly, best is easier than betst, impossible than inpossible; and we may refer also to the numerous words still written with a gh which is no longer pronounced. In the word knight, the k was formerly sounded before the n, and the gh represents a sound which may still be heard in the German word knecht; and, in fact, all spellings like know, gnat, night, though, etc., with their numerous mute letters, represent older and undoubtedly more laborious pronunciations. That all these sounds have been dropped has unquestionably facilitated the utterance of the words, and there is a similar gain of ease in all the well-known instances of complete or partial assimilation in all languages. So in Italian otto for Latin octo, Latin accendo for adcendo, etc. When, however, we come to estimate the comparative facility of separate single sounds, or even many combinations, we find ourselves as yet without any certainty of result or fixed standard. Much that has been advanced is individual and subjective: all depends on practice; and this practice we acquire at an age when we are as yet wholly unable to form or pronounce an opinion on any question. In fact, most of our facility of speech comes to us in infancy.
But whatever the cause, we now understand that the memory-picture of movement and position is shifting and unstable in its very nature. Unless the majority of pronunciations around us all alter in the same direction, the sound-picture does not alter, and it exerts a retarding control upon the rapidity with which our position-picture, and therewith our own pronunciation, might otherwise do so. Here, however, we must draw attention to the fact that we spoke of the majority of pronunciations around us and not of speakers. For our sound-picture the number of persons from whom we hear a word is immaterial; it is the number of times we hear it pronounced that is alone of importance.
All that we have hitherto said has had reference to changes of pronunciation in the same speaker, and in this case alone can we speak of alteration or change in the strict sense of the word. But when we say that ‘a language has altered,’ we use the term in a wider sense, and include the case when one generation is found to use a new pronunciation in place of one current at a former time; when, in fact, it would be strictly correct to say that an old pronunciation has died out, and that the new one—created instead—differs more or less from that which was its model.
A child, in learning to speak, attempts to imitate the sound it hears; and, as long as the resulting imitation sounds sufficiently correct, any small peculiarity of pronunciation is generally overlooked. In such a case, therefore, the child acquires a movement or position-picture which at once materially differs from that of the former generation. We all know by experience that sounds are difficult to ‘catch,’ and we must remember that the vocal organs may undergo certain variations in position without producing a correspondingly large difference in acoustic effect;[4] and further, that any sound produced by a particular position of the vocal organs has a tendency to change in a different direction and at a different rate from the course which would seem natural to the same sound if it had been produced by a different position of the vocal organs.
If, then, we speak a word to a child, and if the child utters it (a) with a slightly altered pronunciation, and (b) with an articulation which differs from that which WE should naturally employ to produce the pronunciation which the child gives to the word, then two comparatively important steps upon the path of change have already been taken. And thus it is clear that, though changes in language are constantly and imperceptibly occurring throughout the whole life of the individual speaker, yet their rise is most likely and their progress is most rapid at the time when language is transferred from one generation to another.
The above, however, will not explain all the changes which words have undergone. There are some which have hitherto resisted any other explanation than this: they appear as the results of repeated errors of utterance, which errors, owing to particular circumstances attending each case, must have been committed by several or by most of the speakers of the same linguistic community. Such are—(1) Metathesis, i.e. where two sounds in the same word reciprocally change their positions, whether they are (a) contiguous or (b) separated by other sounds. Of the first kind we have instances in the Anglo-Saxon forms ascian and axian, both of which occur in extant documents, and also survive in the verb ask and the provincial equivalent aks. Cf. also the form brid, found in Chaucer, for bird (e.g. ‘Ne sey I neuer er now no brid ne best.’—Squire’s Tale, 460), and, vice versâ, birde for bride (e.g. Piers Plowman, 3, 14: ‘ðe Justices somme Busked hem to ðe boure ðere ðe birde dwelled’). Again, we may compare the English bourn, Scotch burn, with Dutch bron, German brunnen; A.S. irnan and rinnan, both meaning to run, and irn, as pronounced by a west-countryman, with run.[5]
Of the second kind of Metathesis (b) we find traces in O.H.G. erila, by the side of elira = N.H.G. erle and eller; A.S. weleras, the lips, as against Gothic wairilos; O.H.G. ezzih, which must have had the sound of etik before the sound-shifting process began, = Lat. acetum; the Italian word, as dialectically pronounced, grolioso = glorioso; and, again, crompare = comprare; M.H.G. kokodrille = Lat. crocodilus. We may also refer to such cases of mispronunciation as indefakitable for indefatigable. These are evanescent, because they meet with speedy correction.
Besides Metathesis, we must class here (2) the assimilation of two sounds not standing contiguous in the word (as Lat. quinque from *pinque; original German finfi (five) = *finhwi, etc.), and (3) dissimilations, as in O.H.G. turtiltûba, from the Lat. turtur; Eng. marble, from Fr. marbre, Lat. marmor; M.H.G. martel with marter, from martyrium; prîol with prîor; and conversely, M.H.G. pheller with phellel, from Lat. palliolum; O.H.G. fluobra, ‘consolation,’ as against O.S. frôfra and A.S. frôfor; M.H.G. kaladrius with karadrius; Middle Lat. pelegrinus, from peregrinus.
We must now conclude this chapter with a few words on the question, Are the laws of sound-change, like physical laws, absolute and unchanging? do they admit of no exceptions? In thus stating the question, we challenge a comparison between physical laws and the laws of sound-change, but we must never forget the essential difference existing between them. Physical laws lay down what must invariably and always happen under certain given conditions; the laws of sound-change state the regularity observed in any particular group of historic phenomena.
We must, in dealing with this question, further distinguish between two closely allied but not identical kinds of phenomena, i.e. between those which come under the law of sound-change in the strict sense of the word, and those which are rather to be considered as instances of sound-correspondence or sound-interchange. When, for instance, some sound happened to be, at any particular stage of some language, identical in the various forms of the same word; and if this sound, owing to difference in its position, or of its accent, or from some other cause, has changed into a different sound in some forms of the word, while in other forms of the same word it has remained unchanged; and if many similar cases are remarked in the same language,—we summarise them in our grammars in a form which, though convenient, is not strictly correct. There are in French, for instance, many adjectives which form their masculine termination in f and their feminine in ve. It is scarcely necessary to point out that in these words the feminine form, derived as it is from the Latin feminine, cannot correctly be described as derived from the masculine in its contemporaneous form: nor yet does the individual speaker, in using the two genders, derive the one from the other; he reproduces both from memory, or, possibly by a process to be discussed in Chapter V., he produces one by analogy with other similar forms.