Language: Its Nature, Development and Origin

by Otto Jespersen

Here, too, there are many points to which I must take exception. It is not quite certain that the usual interpretation of Etruscan letters is correct; in fact, much may be said in favour of the hypothesis that the letters rendered p, t, k stand really for the sounds of b, d, g, and that those transcribed ph, th, kh (or Greek φ, θ, χ) represent ordinary p, t, k. However this may be, Feist seems to be speaking here almost in the same breath of the first (or common Gothonic) shift and of the second (or specially High German) shift, although they are separated from each other by several centuries and neither cover the same geographical ground nor lead to the same phonetic result. Neither Armenian nor primitive Gothonic can be said to be averse to voiced stops, for in both we find voiced b, d, g for the old ‘mediæ aspiratæ.’ And in both languages the old voiceless stops became at first probably not aspirates, but simply voiceless spirants, as in English father, thing, and Scotch loch. Further, it should be noted that we do not find the tendency to unvoice stops and to pronounce affricates either in Rhæto-Romanic (Ladin) or in Tuscan Italian; both languages have unaspirated p, t, k and voiced b, d, g, and the Tuscan pronunciation of c between two vowels as [x], thus in la casa [la xa·sa], but not in a casa = [akka·sa], could not be termed ‘aspiration’ except by a non-phonetician; this pronunciation can hardly have anything to do with the old Etruscan language.

According to a theory which is very widely accepted, the Dravidian languages exerted a different influence on the Aryan languages when the Aryans first set foot on Indian soil, in making them adopt the ‘cacuminal’ (or ‘inverted’) sounds ḍ, ṭ, ṇ with ḍh and ṭh, which were not found in primitive Aryan. But even this theory does not seem to be quite proof against objections. It is easy to admit that natives accustomed to one place of articulation of their d, t, n will unconsciously produce the d, t, n of a new language they are learning in the same place; but then they will do it everywhere. Here, however, both Dravidian and Sanskrit possess pure dental d, t, n, pronounced with the tip of the tongue touching the upper teeth, besides cacuminal ḍ, ṭ, ṇ, in which it touches the gum or front part of the hard palate. In Sanskrit we find that the cacuminal articulation occurs only under very definite conditions, chiefly under the influence of r. Now, a trilled tongue-point r in most languages, for purely physiological reasons which are easily accounted for, tends to be pronounced further back than ordinary dentals; and it is therefore quite natural that it should spontaneously exercise an influence on neighbouring dentals by drawing them back to its own point of articulation. This may have happened in India quite independently of the occurrence of the same sounds in other vernaculars, just as we find the same influence very pronouncedly in Swedish and in East Norwegian, where d, t, n, s are cacuminal (supradental) in such words as bord, kort, barn, först, etc. According to Grandgent (Neuere Sprachen, 2. 447), d in his own American English is pronounced further back than elsewhere before and after r, as in dry, hard; but in none of these cases need we conjure up an extinct native population to account for a perfectly natural development.

XI.—§ 5. Gothonic Sound-shift.

Since the time of Grimm the Gothonic consonant changes have harassed the minds of linguists; they became the sound-shift and were considered as something sui generis, something out of the common, which required a different explanation from all other sound-shifts. Several explanations have been offered, to some of which we shall have to revert later; none, however, has been so popular as that which attributes the shift to an ethnic substratum. This explanation is accepted by Hirt, Feist, Meillet and others, though their agreement ceases when the question is asked: What nationality and what language can have been the cause of the change? While some cautiously content themselves with saying that there must have been an original population, others guess at Kelts, Finns, Rhætians or Etrurians—all fascinating names to minds of a speculative turn.

The latest treatment of the question that I have seen is by K. Wessely (in Anthropos, XII-XIII 540 ff., 1917). He assumes the following different substrata, beginning with the most recent: a Rhæto-Romanic for the Upper-German shift, a Keltic for the common High-German shift, and a Finnic for the first Germanic shift with the Vernerian law. This certainly has the merit of neatly separating sound-shifts that are chronologically apart, except with regard to the last-mentioned shift, for here the Finns are made responsible for two changes that were probably separated by centuries and had really no traits in common. It is curious to see the transition from p to f and from t to þ—both important elements of the first shift—here ascribed to Finnic, for as a matter of fact the two sounds f and þ are not found in present-day Finnish, and were not found in primitive Ugro-Finnic.[44]

When Wessely thinks that the change discovered by Verner is also due to Finnic influence, his reasons are two: an alleged parallelism with the Finnic consonant change which he terms ‘Setälä’s law,’ and then the assumption that such a shift, conditioned by the place of the accent, is foreign to the Aryan race (p. 543). When, however, we find a closely analogous case only four hundred years ago in English, where a number of consonants were voiced according to the place of the stress,[45] are we also to say that it is foreign to the Anglo-Saxon race and therefore presupposes some non-Aryan substratum? As a matter of fact, the parallelism between the English and the old Gothonic shift is much closer than that between the latter and the Finnic consonant-gradation: in English and in old Gothonic the stress place is decisive, while in the Finnic shift it is very doubtful whether stress goes for anything; in both English and old Gothonic the same consonants are affected (spirants, in English also the combinations [tʃ, ks], but otherwise no stops), while in Finnic it is the stops that are primarily affected. In old Gothonic, as in English, the change is simply voicing, and we have nothing corresponding to the reduction of double consonants and of consonant groups in Finnic pappi / papin, otta / otat, kukka / kukan, parempi / paremman, jalka / jalan, etc. On the whole, Wessely’s paper shows how much easier it is to advance hypotheses than to find truths.

XI.—§ 6. Natural and Specific Changes.

Meillet (MSL 19. 164 and 172; cf. Bulletin 19. 50 and Germ. 18) thinks that we must distinguish between such phonetic changes as are natural, i.e. due to universal tendencies, and such as are peculiar to certain languages. In the former class he includes the opening and the voicing of intervocalic consonants; there is also a natural and universal tendency to shorten long words and to slur the pronunciation towards the end of a word. In the latter class (changes which are peculiar to and characteristic of a particular language) he reckons the consonant shifts in Gothonic and Armenian, the weakening of consonants in Greek and in Iranian, the tendency to unround back vowels in English and Slav. Such changes can only be accounted for on the supposition of a change of language: they must be due to people whose own language had habits foreign to Aryan. Unfortunately, Meillet cannot tell us how to measure the difference between natural and peculiar shifts; he admits that they cannot always be clearly separated; and when he says that there are some extreme cases ‘relativement nets,’ such as those named above, I must confess that I do not see why the change from the sharp tenuis, as in Fr. p, t, k, to a slightly aspirated sound, as in English (Bulletin 19. 50),[46] or the relaxing of the closure which finally led to the sounds of [f, þ, x], should be less ‘natural’ than a hundred other changes and should require the calling in of a deus ex machina in the shape of an aboriginal population. The unrounding of E. u in hut, etc., to which he alludes, began about 1600—what ethnic substratum does that postulate, and is any such required, more than for, say, the diphthongizing of long a and o?