to his satisfaction. An instance just parallel to this is crape from the French crêpe: the sound could not have been maintained without a change of the spelling.

The Welsh pronunciation of borrowed words affords no trustworthy evidence, the forms being so much altered. It would for instance be very hazardous to conclude from the Welsh forms Lundain, Ffrainc, and Tain, that the words then first borrowed into Welsh had such sounds as our neighbours give them now: that London was (lindain), France (fraink), and Thames (tain).

27 'OU' words.

I pass on to the consideration of another compound sound, as to which again our Southern dialects. maintain a nearly uniform tradition, namely, the diphthong (so-called) heard in house, mouse, ground, &c. Our dialects do not all give quite the same sound, but in all it is a compound, and made up of nearly the same elements. It may not be amiss to investigate those elements; for though the nature of diphthongs and other compound vowel sounds has been abundantly discussed, the subject is perhaps not yet quite exhausted.

28 The vowels in their natural se

quence.

It has been proved by Willis's experiments* that the vowels in their natural sequence are

(ii) (ee) (aa) (AA) (00) (uu);

and this is the truth apparently, but not the whole truth. There are in fact-as is nowhere perhaps more fully shown than in Mr. Ellis's Key to Palæotype (Early English Pronunciation, pp. 3-10)—numerous, or rather innumerable, intermediate sounds, all delicately shading off into those next to them, that occupy the intervals between these sounds, or extend beyond the series at either extremity. For sounds not used by one nation or in one dialect are familiar in another, not to mention that probably no two individuals who speak any language utter vowels abso

* It is rather surprising that Mr. Melville Bell, when propounding his own ingenious observations and complicated vowel-system, has not shown the relation of his system to Willis's. The facts which the latter observed and described are still facts, and should not have been ignored. Lepsius also has overlooked them.

lutely identical, even when these are intended and supposed to be so. By way of illustration, here are a few of these additions to the vowel-scale. At one extremity of the

series we have the French (ii) somewhat thinner than our English (ii), and at the other the French u and German ü (yy) considerably thinner than our (uu). Then between (ii) and (ee) we have (ii), or more commonly (i) short-which in some dialects, especially in the West of England and north of the Tweed, is apt to approach very near (e)—and (ee). Between (ee) and (aa) we have (ææ), or more commonly (a) short, as well as (a) and (ǝ)—these two almost identical. Between (aa) and (AA), with various nuances of sound, we have the Prussian (aa). Between (AA) and (oo) there is perhaps no recognized sound intervening. Between (oo) and (uu) we have (oo), and the German ö, French eu, (œ). This somewhat more complete series may now be seen in the vowels of the following words:

il (Fr.), eel, ill, male, mare, man, vœu (Fr.), bun, path, mann (Pruss.), lawn, robe (Fr.), robe, jeune (Fr.), pool, flûte (Fr.) 29 Nature of diph- And now I have to observe that in the sothongs, which are called diphthongs we do not merely sound cernot merely two vowels combined. tain two vowels of this series in immediate juxtaposition, but we glide from one to the other, thus of necessity passing with extreme rapidity through all the intervening sounds. A diphthong therefore is not merely two vowels compressed, but a whole series compressed; and it is the length of the series compressed which marks out the diphthongs, and compels us to recognize them as such. When for instance Mr. Melville Bell says, "The diphthongal quality of the English a will not at first be admitted by every reader"-and a similar remark might be made about our ō*-wherein consists the difficulty of

"It is well known that nearly all the English long vowels, so called, are composed of two distinct elements; in other words, they are diphthongs. For the gratification of those who may wish to know how the Greeks express them in Romaic letters, we give the following table.

recognizing the diphthongal sound? Simply it escapes observation through the shortness of the series compressed, (ee) being united with its near neighbour (ii), and (00) with (uu). But in each of the three diphthongs i (əi), ou (ǝu), and oi (i)-as sounded in by, bough, boy-a long series is compressed into the resultant compound vowel. The long English u, which is commonly spoken of as a diphthong -and in this essay I shall for convenience' sake still so call it is not strictly a diphthong in the same sense; for it does not really consist of (ii) and (uu), but of the semi-vowel*y (J) and (uu), and has no more claim to the name of diphthong than has the precisely analogous compound of the semi-vowel (w) and (ii) which we have in the pronoun we and the French oui.

30 Definition of A diphthong then is a series of vowel sounds

diphthong.

taken in their natural order and uttered with extreme brevity and compression; and the differences that exist between the modes in which they are pronounced in different dialects or by different individuals depend on the exact points at which the speaker takes up, and drops, the series. Thus to sound the English long i we pronounce one of the (a) sounds and rapidly glide from it into one of the (i) sounds, in the inverse order of the above series; but one speaker will begin with the Prussian (a), another with (a), another—and this I think the most usual—with (æ), another nearly with (e); and one will finish with (i), which is most common, another with (i), a third-a foreigner probably-with the thin French (i).

31 Definition ap

ou, German au,

And what of our ow or ou of how and house?

plied to English It clearly begins early in the vowel series, and Dutch ui, &c. ends with one of the (o) or (u) sounds: in fact

ō is sounded nearly like óov, óa ,, note, vóovτ; store, σróap

й

ίου, ίονα ,, mute, μíovr; pure, πíovuρ

It will be observed that i as in fire, and u as in pure, are each composed of three distinct elements." E. A. Sophocles, Romaic Grammar, p. 3.

* I have discussed elsewhere the nature of the semi-vowels (of which I contend that we have three in English-y, the non-vibrant r, and w), as well as the true definitions of the terms vowel and consonant. Letter R" in the Transactions of the Philological Society for 1862-3, pp. 265

See paper

"On the

to 267.

32

33

almost all our vowels are compressed into it. An Essex man speaking of his house or cow begins his diphthong with (e); the Londoner commonly starts from (a); while the customary pronunciation begins with (or about) the (∞). The terminus ad quem is in Essex and commonly (0); but the Devonshire dialect prefers to terminate with (œ): you cannot without an effort advance as far as (u).

In German there is a similar diphthong, differently pronounced indeed—as might be expected-in different parts of Germany. It begins however with (a) or (a), and ends with (u) rather than (o). Yet it requires close observation to distinguish the German Haus from our house. (In Icelandic there is a like diphthong, written á, which Mr. Hjaltalín told me was pronounced exactly like the customary English ow of how and now; but see Ellis, p. 540.) And in Dutch we have the well-known ui or uy, which Mr. Ellis writes in palæotype (ǝy). He adds in a note (p. 235): "In the actual Dutch pronunciation of huis, muis, it is very difficult to distinguish the sound from (ǝu), and the difference seems mainly produced by altering the form of the lips into that for (yy), which is slightly flatter than for (uu), rather than by bringing the tongue into the (i) position. Still (ǝy) was the best analysis I was able to make on hearing the sound." This my own very careful observation corroborates.

dence as to 'OU'

words.

To return now to my argument: just as I have above Traditional evi- insisted on the traditional evidence concerning (əi), so I argue as to (ǝu). In a certain small class of words a sound almost identical is given in all our southern dialects, having been handed down from generation to generation; and this uniform tradition furnishes evidence of the greatest possible weight, and, unless there be strong opposing evidence, it fixes approximately the ancient sound, whether the symbol be ou or ow, or, as in A.S., ú.

Moreover to confirm this evidence, just as in the case of the (ǝi) words, we have in German and Dutch almost the same sound in many of these words; and these too are all

monosyllables, in which therefore the stress of the voice This evidence rests on this sound so as to render any change confirmed by

that about (əi)

German and of vowel-especially in so many languages and Dutch, just as dialects simultaneously-all the more difficult. words: list. The natural conclusion is that these words had the (ou) sound long prior to historical times, and when the great divisions of the Teutonic race had not yet split asunder.

It is needless to give all the dialectic forms, but here is a short list of words in which English, German, and Dutch; all give very nearly the same sound of (ǝu). Of course, if these are fixed, many others that habitually rhyme with them are fixed also, as well as many of the derivatives—

34 Roun, dowve, And the German raumen is very like the old doust, úp, ús. English roun=whisper, Anglo-Saxon rún. To these must be added three others which are now pronounced with the й of but, namely, dove, dust, and up. The provincial and Early English dowve or dowf is well known, and in Devonshire doust is commonly used in the sense of chaff: the former of these is Dutch duif, German Taube,