into the windpipe. Where the withdrawal of the velum is rendered impossible by disease, such a case came under Czermak's 2 observation, nasals cannot be produced.8

pure

The so-called mouillé or softened nasal, and all other mouillé consonants, are produced by the addition of a final y, and need not be classified as simple letters.

Aspirated Checks.

For most languages the letters hitherto described would be amply sufficient; but in the more highly organized forms of speech new distinctions were introduced and graphically expressed which deserve some explanation. Instead of pronouncing a tenuis as it ought to be pronounced, by cutting sharp through the stream of breath or tone which proceeds from the larynx, it is possible to gather the breath and to let it explode audibly as soon as the consonantal contact is withdrawn. In this manner we form the hard or surd aspirates which occur in Sanskrit and in Greek, kh, th, ph.

If, on the contrary, we pronounce g, d, b, and allow the soft breathing to be heard as soon as the contact is removed, we have the soft aspirates, which are of frequent occurrence in Sanskrit, gh, dh, bh.

Much discussion has been raised on these hard and soft aspirates, the question being whether their

1 Czermak, Wiener Akademie, xxiv. p. 9.

2 Funke, p. 681. Czermak, Wiener Akademie, xxix. p. 173.

8 Professor Helmholtz has the following remarks on M and N: "M and N resemble the vowels in their formation, because they cause no noise in the buccal tube. The buccal tube is shut, and the voice escapes through the nose. The mouth only forms a resounding cavity, modifying the sound. If we watch from below people walking up-hill and speaking together, the nasals m and n are heard longest."

first element was really a complete consonantal contact, or whether the contact was incomplete, and the letters intended were hard and soft breathings. As we have no means of hearing either the old Brahmans or the ancient Greeks pronounce their hard aspirates, and as it is certain that pronunciation is constantly changing, we cannot hope to derive much aid either from modern Pandits or from modern Greeks. The Brahmans of the present day are said to pronounce their kh, th, and ph like a complete tenuis, followed by the spiritus asper. The nearest approach to kh is said to be the English kh in inkhorn, though this can hardly be a good illustration, as here the tenuis ends and the aspirate begins a syllable. The Irish pronunciation of kind, town, pig, has likewise been quoted as in some degree similar to the Sanskrit hard aspirates. In the modern languages of India where the Sanskrit letters are transcribed by Persian letters, we actually find kh represented by two letters, k and h, joined together. The modern Greeks, on the contrary, pronounce their three aspirates as breathings, like h, th, f. It seems to me that the only two points of importance are, first, whether these aspirates in Greek or Sanskrit were formed with or without complete contact, and, secondly, whether they were classed as surd or as sonant. Sanskrit grammarians allow, as far as I can judge, of no doubt on either of these points. The hard aspirates are formed by complete contact (sprishta), and they belong to that class of letters for which the glottis must be completely open, i. e. to the surd or hard consonants. These two points once established put an end to all speculations on the

subject. What the exact sound of these letters was is difficult to determine, because the ancient authorities vary in their descriptions, but there is no uncertainty as to their physiological character. They are said to be uttered with a strong out-breathing (mahâprâṇaḥ), but this, as it is shared by them in common with the soft aspirates and the hard breaths, cannot constitute their distinctive feature. Their tech

correct

nical name "soshman," i. e. " with wind," would admit of two explanations. "Wind" might be taken in the general sense of breath, or- and this is more in the sense of the eight letters called “the winds" in Sanskrit, h, ś, sh, s, tongue-root breath (Jihvâmûlîya), labial breath (Upadhmânîya), neutral breath (Visarga), and neutral nasal (Anusvâra). Thus it is maintained by some ancient grammarians1 that the hard aspirates are the hard letters k, t, p, together with the corresponding winds or homorganic winds; that is to say, kh is = k + tongueroot breath, th=t+s, ph = p + labial breath. The soft aspirates, on the contrary, of which more hereafter, are said to be produced by the union of the soft g, d, b, with the soft 'h. It is quite clear that the Sanskrit 'h, which is not the spiritus asper (though it has constantly been mistaken for that), but a sonant letter, could not possibly form the second element in the hard aspirates. They were formed, as here described, by means of complete hard contact, followed by the hard breaths of each organ. The objections which other grammarians raise against this view do not affect the facts, but only their explanation. As they look upon all letters

1 Survey of Languages, p. xxxii. Śâkala-Prâtiśâkhya, xiii. 18.

as eternal, they cannot admit their composite character, and they therefore represent the aspiration, not as an additional element, but as an external quality, and prescribe for them a quicker pronunciation in order to prevent any difference between them and other consonants. In other letters the place, the contact, and the opening or shutting of the glottis form the three constituent elements; in the aspirates a fourth, the breath, is added. The Sanskrit hard aspirates can only be considered as k, t, p, modified by the spiritus asper, which immediately follows them, and which assumes, according to some, the character of the guttural, dental, or labial breaths.

As to the Greek aspirates, we know that they belonged to the aphōna, i. e. that they were formed by complete contact. They were not originally hemiphona or breaths, though they became so afterwards. That they were hard, or pronounced with open glottis, we must gather from their original signs, such as ПIH, and from their reduplicated forms, tí-thēmi, kéchyka, pé-phyka.1

It is more difficult to determine the real nature of the Sanskrit soft aspirates, gh, dh, bh. According to some grammarians they are produced by the union of g, d, b, with 'h, which in Sanskrit is a sonant letter, a spiritus lenis, but slightly modified.2 The same grammarians, however, maintain that they are not formed entirely with the glottis closed, or as sonant letters, but' that they and the h require the glottis "both to be opened and to be closed." What

1 Raumer, Aspiration, 96. Curtius, Gr. Etymologie, ii. p. 11.

2 If Sanskrit writing were not of so late a date, the fact that the Vedic dh or lh is actually represented by a combination of and h might be

A letter may be

this means is somewhat obscure. A letter

either surd or sonant, but it can hardly be both; and the fact that not. only the four soft aspirates but the simple 'h1 also were considered as surd-sonant, would seem to show that an intermediate rather than a compound utterance is intended. One thing is cer tain, namely, that neither the hard nor the soft aspirates were originally mere breaths. They are both based on complete contact, and thus differ from the hard and soft breaths which sometimes take their places in cognate tongues.

We have thus finished our survey, which I have kept as general as possible, without dwelling on any of the less normal letters peculiar to every language, every dialect- nay, to the pronunciation of every individual. It is the excessive attention paid to these more or less peculiar letters that has rendered most works on Phonetics so complicated and unintelligible. If we have clearly impressed on our mind the normal conditions of the organs of speech in the production of vowels and consonants, it will be easy to arrange the sounds of every new language under the categories once established on a broad and firm basis. To do this, to arrange the alphabet of any given language according to the compartments planned by physiological research, is the office of the grammarian, not of physiologist. But even here, too much nicety is dangerous. It is easy to perceive some little difference between k, t, p, as pronounced by an Englishman and by a German, yet each has only one set of tenues, and to class

1 Śâkala-Prátiśákhya, xiii. 1. The expression "the breath becomes both sonant and surd between the two," i. e. between the complete opening and shutting, shows that an intermediate sound is meant.