In this and the preceding chapter I have tried to pass in review the various circumstances which make for changes in the phonetic structure of languages. My treatment is far from exhaustive and may have other defects; but I want to point out the fact that nowhere have I found any reason to accept the theory that sound changes always take place according to rigorous or ‘blind’ laws admitting no exceptions. On the contrary, I have found many indications that complete consistency is no more to be expected from human beings in pronunciation than in any other sphere.
It is very often said that if sound laws admitted of exceptions there would be no possibility of a science of etymology. Thus Curtius wrote as early as 1858 (as quoted by Oertel 259): “If the history of language really showed such sporadic aberrations, such pathological, wholly irrational phonetic malformations, we should have to give up all etymologizing. For only that which is governed by law and reducible to a coherent system can form the object of scientific investigation; whatever is due to chance may at best be guessed at, but will never yield to scientific inference.” In his practice, however, Curtius was not so strict as his followers. Leskien, one of the recognized leaders of the ‘young grammarians,’ says (Deklination, xxvii): “If exceptions are admitted at will (abweichungen), it amounts to declaring that the object of examination, language, is inaccessible to scientific comprehension.” Since then, it has been repeated over and over again that without strict adherence to phonetic laws etymological science is a sheer impossibility, and sometimes those who have doubted the existence of strict laws in phonology have been looked upon as obscurantists adverse to a scientific treatment of language in general, although, of course, they did not believe that everything is left to chance or that they were free to put forward purely arbitrary exceptions.
There are, however, many instances in which it is hardly possible to deny etymological connexion, though ‘the phonetic laws are not observed.’ Is not Gothic azgo with its voiced consonants evidently ‘the same word’ as E. ash, G. asche, Dan. aske, with their voiceless consonants? G. neffe with short vowel must nevertheless be identical with MHG. neve, OHG. nevo; E. pebble with OE. papol; rescue with ME. rescowe; flagon with Fr. flacon, though each of these words contains deviations from what we find in other cases. It is hard to keep apart two similar forms for ‘heart,’ one with initial gh in Skt. hrd and Av. zered-, and another with initial k in Gr. kardía, kēr, Lat. cor, Goth. haírto, etc. The Greek ordinals hébdomos, ógdoos have voiced consonants over against the voiceless combinations in heptá, oktṓ, and yet cannot be separated from them. All this goes to show (and many more cases might be instanced) that there are in every language words so similar in sound and signification that they cannot be separated, though they break the ‘sound laws’: in such cases, where etymologies are too palpable, even the strictest scholars momentarily forget their strictness, maybe with great reluctance and in the secret hope that some day the reason for the deviation may be discovered and the principle thus be maintained.
Instead of exacting strict adherence to sound laws everywhere as the basis of any etymologizing, it seems therefore to be in better agreement with common sense to say: whenever an etymology is not palpably evident, whenever there is some difficulty because the compared words are either too remote in sound or in sense or belong to distant periods of the same language or to remotely related languages, your etymology cannot be reckoned as proved unless you have shown by other strictly parallel cases that the sound in question has been treated in exactly the same way in the same language. This, of course, applies more to old than to modern periods, and we thus see that while in living languages accessible to direct observation we do not find sound laws observed without exceptions, and though we must suppose that, on account of the essential similarity of human psychology, conditions have been the same at all periods, it is not unreasonable, in giving etymologies for words from old periods, to act as if sound changes followed strict laws admitting no exceptions; this is simply a matter of proof, and really amounts to this: where the matter is doubtful, we must require a great degree of probability in that field which allows of the simplest and most easily controllable formulas, namely the phonetic field. For here we have comparatively definite phenomena and are consequently able with relative ease to compute the possibilities of change, while this is infinitely more difficult in the field of significations. The possibilities of semantic change are so manifold that the only thing generally required when the change is not obvious is to show some kind of parallel change, which need not even have taken place in the same language or group of languages, while with regard to sounds the corresponding changes must have occurred in the same language and at the same period in order for the evidence to be sufficient to establish the etymology in question.
It would perhaps be best if linguists entirely gave up the habit of speaking about phonetic ‘laws,’ and instead used some such expression as phonetic formulas or rules. But if we are to keep the word ‘law,’ we may with some justice think of the use of that word in juridical parlance. When we read such phrases as: this assumption is against phonetic laws, or, phonetic laws do not allow us this or that etymology, or, the writer of some book under review is guilty of many transgressions of established phonetic laws, etc., such expressions cannot help suggesting the idea that phonetic laws resemble paragraphs of some criminal law. We may formulate the principle in something like the following way: If in the etymologies you propose you do not observe these rules, if, for instance, you venture to make Gr. kaléo = E. call in spite of the fact that Gr. k in other words corresponds to E. h, then you incur the severest punishment of science, your etymology is rejected, and you yourself are put outside the pale of serious students.
In another respect phonetic laws may be compared with what we might call a Darwinian law in zoology, such as this: the fore-limbs of the common ancestor of mammals have developed into flippers in whales and into hands in apes and men. The similarity between both kinds of laws is not inconsiderable. A microscopic examination of whales, even an exact investigation by means of the eye alone, will reveal innumerable little deviations: no two flippers are exactly alike. And in the same way no two persons speak in exactly the same way. But the fact that we cannot in detail account for each of these nuances should not make us doubt that they are developed in a perfectly natural way, in accordance with the great law of causality, nor should we despair of the possibility of scientific treatment, even if some of the flippers and some of the sounds are not exactly what we should expect. A law of fore-limb development can only be deduced through such observation of many flippers as will single out what is typical of whales’ flippers, and then a comparison with the typical fore-limbs of their ancestors or of their congeners among existing mammals. And in the same way we do not find laws of phonetic development until, after leaving what can be examined as it were microscopically, we go on telescopically to examine languages which are far removed from each other in space or time: then small differences disappear, and we discover nothing but the great lines of a regular evolution which is the outcome of an infinite number of small movements in many different directions.
XV.—§ 14. Conclusion.
It has been one of the leading thoughts in the two chapters devoted to the causes of linguistic change that phonetic changes, to be fully understood, should not be isolated from other changes, for in actual linguistic life we witness a constant interplay of sound and sense. Not only should each sound change be always as far as possible seen in connexion with other sound changes going on in the same period in the same language (as in the great vowel-raising in English), but the effects on the speech material as a whole should in each case be investigated, so as to show what homophones (if any) were produced, and what danger they entailed to the understanding of natural sentences. Sounds should never be isolated from the words in which they occur, nor words from sentences. No hard-and-fast boundary can be drawn between phonetic and non-phonetic changes. The psychological motives for both kinds of changes are the same in many cases, and the way in which both kinds spread through imitation is absolutely identical: what was said on this subject above (§ [11]) applies without the least qualification to any linguistic change, whether in sounds, in grammatical forms, in syntax, in the signification of words, or in the adoption of new words and dropping of old ones.