pean language. Let us then make a our starting point, and proceed to investigate the relations of the other sounds to it.

If the tongue were the only agent in producing the various positions which give the different vowel-sounds, we should have been able to start from a, where the tongue is in the lowest position, and arrange all the vowels in one scale, ending with the highest position, i.e. at i that is (ee). But this is not so. It has been already pointed out that each vowel sound so obtained can be modified by that contraction of the vocal tube which finds its outward expression in the greater roundness of the lipaperture. This fact necessitates a second line, again commencing with a and running as far as u (oo): this at least is the extremity of the scale in England, but in pronouncing u the tongue is still at the back of the palate: if it be put through the same range of positions in this scale as in the other, the limit will be the German ü. It can hardly be doubted that these three English sounds, ah, ee, oo represent the three original vowels A, I, U. No other sounds, so far as I know, have ever usurped their place in any European language except our own.

CH. IV.

85

Two lines of vowels

necessary.

vowels.

Now in each of these scales from a to i, and from a Dubious to u, there is a very great number of possible vowel-positions. I will point out only the most important in each scale. The first variations from a are however by no means easy to classify. There is a debateable land, and some sounds lie close on the unknown border-line. Such are the (a) and (u) in "ask" and "up"." By the (a) of "ask" is meant a middle sound between “ah” and “ 'an," which is constantly heard in conversation, but which has never been definitely assigned by custom to any word; it is nearly always heard in "ask." There is a transition constantly going on in England from the full ah-sound,

1 I differ here from Mr Bell, who classes these as "mid-back" sounds, they seem to me "mixed." My arrangement is nearer to that of Prof. Lepsius.

CH. IV.

(i) The "front" vowels.

which survives now in so few words, to the short a in
"an," which perhaps belongs to no other language but the
Sanskrit: and it would be quite possible to make some
point in the intervening space, which should represent the
(a) of "ask;" but this has never been done. This (a) is
an open sound; the corresponding close sound is the (u)
of "up" which certainly, to my ear, seems very like the
"neutral" vowel of unaccented syllables. These two
sounds, then, I classify in neither scale: but between the
first variations of each. First, in the scale from A to I,
the "front" vowels, as they may be called, come the short
English (a) of "an," which is open, and the (ě) of “ell,”
which is close: for these the back of the tongue is low',
(though higher than for any vowel of the "back" class):
and the point of the tongue is much depressed. Next to
these, but produced by a considerable raising of both the
back and front of the tongue, come the Italian open e (è), and
close e (é). We have no exact equivalent in common English
for the open è, except in words ending with the glide r,
e.g.
e. g. "pair:" the close e is nearly our long English (1)
when pronounced, as it almost always is, with a slight i-
sound following it, in fact as a diphthong; whether the i
be written or not, the pronunciation is the same, e.g.
"pail" or "pale." The vowel sound in "pail," if pro-
nounced, as it is in some English dialects, e.g. in Cumber-
land, is more like the open e3.

The last pair of sounds in this scale are open (1) and

1 I do not understand Mr Bell's diagram of the position of the tongue for these two sounds. It seems to me, after repeated experiments, that he has placed the back of the tongue much higher than it really is.

2 Compare Prof. Munro's Remarks on the Pronunciation of Latin, p. 6. 3 Mr Roby (Grammar, p. 9) makes the Italian open e a "front," that is a primary or close, not an open sound; the class to which, according to Mr Bell, "ell" belongs. Mr Roby himself compares the Scotch sound of "ell," which is certainly much more open. According to his arrangement, Italian open and close e do not fit into my system of the open and close vowels. There is also some discrepancy between us with respect to the open and close o. I am not competent to decide the question, but give my own arrangement as in my view most probable. Mr Bell does not deal with the Italian vowels.

close (ee): they are heard in "pill" and "peel" respectively. For these the tongue is almost exactly in the same position as for the consonant y, the back and point being both raised as much as possible, but not of course so near the palate as to produce friction. We have thus marked all the principal points from the extremely open a (ah), to the extremely close i (ee). All of these come into the class of" Front Vowels."

CH. IV.

and rounded vowels.

Next, let us take the list of "Back" and of "Rounded" (2)The back vowels: in English they are identical: in French and German we have rounded front-vowels. First, come open (0) and close (aw): heard in "Poll" (the bird) and "Paul1." Next, come the Italian open ò: the corresponding close sound is our (5): therefore for classification it will probably be better to take our (ō) as the typical sound, and to regard the Italian ó as a deviation from it. If (ō) be heard in "pole" the open o will be pretty nearly heard in the Cumberland variation of the same word, or in the word "pour," where r follows. For each of these o-sounds the tongue is raised at the back from its position for (ŏ) and (aw); but not so much as for e in "pair" and "pale;" and the lip aperture is rounded. If now the mouth be placed in the position for "pail," and then rounded, the modified o of Germany (ö) or the French eu will be heard. Again, if the tongue be slightly raised from the position for "pole," and the lips very much contracted, we get the final English pair (ū) and (oo) in “pull" (open) and "pool" (close). The position of the tongue for this u (oo) bears the same relation to that for the consonant w, as that for i, mentioned above, bears to that for y. Lastly, if the lips be rounded while the tongue is in the position for the front vowel i, the result will be the German ü, and nearly the French u. It will be seen that the modified vowels ö and i thus combine the characteristics of the two classes: in their tongue-position they belong to the first; in their

1 The Italian close o seems to be nearer to u, or, as Mr Roby makes it, a “High Mixed Wide (i.e. open) Round."

CH. IV.

Vowel pyramid.

rounded character to the second. They may therefore be fairly placed between them in the following "pyramid," the place which they occupy in that of Prof. Lepsius, though he gives no reason for putting them there.

The entire arrangement of the sounds is nearly that of Prof. Lepsius: but with the further distinction of open and close sounds throughout, and of the round vowels, for which I am indebted to Mr Bell. In each couple of sounds the open vowel is placed above the close, so that the lines pass from the most open sound (ah) to the most close (ee), ü, and (oo). The symbols A., A, are employed to denote the sounds between a and e, and a and o, which occur at the beginning of these two lines. The marks of quantity above the English symbols must be understood as applying to the English only.

Terms

It will be observed that the terms "open," and "close," "open" and in the above classification, sometimes are equivalent to

"close:"

* I have marked this u long for want of a better symbol, in order to distinguish it from the (u), e.g. "put" from "shut". But it is not really long, and cannot be pronounced so except in singing: see next page.

66

'short," and "long." This is always permissible in English, but not always in other languages. Thus in English it is unimportant whether we call the i in "sin" short or open, the ee in seen" long or close. In the tone of each man's ordinary speech it is impossible to pronounce "sin" long, without its becoming "seen:" but in singing, it is quite possible to lengthen the i without letting it pass into ee. It is indeed possible, on the other hand, to pronounce in our natural tone "seen" short, that is, so that the vowel shall occupy no longer time than the i of "sin;" which is the only test of quantity. Practically however with us, "seen" is always longer than "sin:" therefore, as I said, the terms "open," and "shut," are convertible in English; and also "long," and "close." The same may be said of u, open and close. But it must not be supposed that this convertibility is universal. In modern Italian, and in modern Greek, the short i is not the vowel sound of "sin," but really the short of "seen," that is, a short close vowel. Therefore we have four variations of i; short open, the English; long open, in singing; short close, the Italian ; long close, the Italian &, and English ee, which are the same. We need not doubt that the Graeco-Italian long i also was the last sound: but what was the Graeco-Italian ? The only evidence that can be adduced, is the modern pronunciation of each language; in which there is no appearance of any variation.

Such arguments do not commonly carry much weight. When we find so much change taking place in the vocalism of a language in a few hundred years, how much greater changes may reasonably be expected to have occurred between the first centuries of the Christian era and our time? Constant change is the normal state of language: and absolute agreement of ancient and modern sounds is almost a suspicious circumstance. Yet here we may perhaps give more weight than usual to the evidence, for this reason: close i is the very thinnest sound in language: the resonance-cavity behind the organs which pro