ART. XI.—The duty of English-speaking Orientalists in regard to united action in adhering generally to Sir William Jones's Principles of Transliteration, especially in the case of Indian Languages; with a Proposal for Promoting a Uniform International Method of Transliteration so far at least as may be applicable to Proper Names. By Sir MONIER MONIER-WILLIAMS, K.C.I.E., D.C.L.

BEFORE entering on the subject of Transliteration, I propose to submit a few preliminary remarks on the formation of graphic systems generally, and the rules which ought to regulate their application.

The following rules or principles will, I think, be accepted by most scholars in the present day :

General Alphabetical Rules.

1. In each graphic system each vowel-sound should invariably be represented by the same vowel-symbol and by no other; e.g. the sound of a in 'zebra,' 'organ,' 'cedar,' should always be represented by the symbol a and not sometimes by u as in 'gun,' or by o as in 'son,' or in any other way (see artificial sentences given at p. 612).

2. Each consonantal sound should invariably be represented by the same consonantal symbol and by no other; e.g. the consonantal sound of j in our word 'jib' should always be represented by the symbol j and not sometimes by g as in 'gibbet'; and again, the sound of s should always be represented by s and not sometimes by c; and the sound of k by k, and not sometimes by c, as in the words 'cinder,' 'kind,' 'cicatrize.'

8

3. Each simple sound should be represented by a single symbol. This rule should apply to the sound represented by c in the Italian 'civita,'' dolce,' but represented by ch in our VOL. XXII-[NEW SERIES.]

40

'chisel,' 'church,' and by tsch or tch in German and French transliteration of Indian words. (Note; it is difficult to get rid of the prejudice that our sound of ch is an aspiration of c, but that it is not so is proved by the fact that it has an aspirated form; see Nāgarī alphabet, next page).

4. Aspirated consonantal sounds should in strictness be represented by single symbols or by some mark of emphatic breathing or by some modification of the unaspirated consonant, and not by adding another distinct letter such as h; e.g. the aspiration of t should be represented by a single symbol, as in the Greek ; or by adding a mark as in for t or t'; or as in Anglo-Saxon d for dh. (Note, however, that the Latin method of adding h, as in theologus for Deoλóyos, philologus for piλoλóyos, is now, in my opinion, too universally practised to be set aside.')

I next pass to certain actual alphabets, and begin with the Nāgari or Deva-Nāgarī alphabet of India, as more complete, and as conforming more nearly than any other to the above rules. It is provided with 47 simple symbols (14 vowels and 33 consonants), and, to make a fount of types complete, there should be provision for about 500 compound consonants.

This alphabet, thought to have borrowed some of its forms from a Phoenician source,2 was developed through the gradual expansion of the symbols used in the Southern Asoka inscriptions, written in a kind of Prākṛit or Pāli, but was probably not thoroughly fixed in its present shape till about the twelfth century of our era. The equivalent letters, here added, are those of the second edition of my SanskritEnglish Dictionary, published by the University of Oxford. It is, however, a mistake to suppose that the Nāgarī is peculiarly a Sanskrit alphabet; for the Sanskrit language, though it has adopted the Nagarī, may also be written in Telugu, Kanarese, Malayalam or Grantham letters.

1 I have arrived at this conclusion since writing my papers on transliteration for the Berlin and Leiden Congresses in 1881 and 1883.

2 Of the Indian Inscription characters those called the Southern Aṣoka-from which the Nagari was derived-are not, we know, so clearly traceable to a Phoenician source as the Northern, and some hold that the Nagari had its origin in Southern India, or at any rate that some of its symbols are original.

THE NAGARI ALPHABET.

Vowels.

अ a, आ, Tā; (▼ lṛi, a lṛī), T,

इ, fi, ई, i; उ, u, 5,

ऋri, ऋ ri

9

a

au.

e; ऐ, ^ ai; ओो, े ०; औ,

Anusvāra n or m; Visarga: h; Virāma`.

Secondly, I give the Hellenic or Greek alphabet-known to be from a Phoenician source -arranging it (as far as is possible without palatals and cerebrals) in the same order. It has only twenty-four simple symbols and two compound; but has both capital and small letters (reckoned as one) thus:

THE HELLENIC, OR GREEK ALPHABET.

Vowels.

A, a; E, e; H, n; I, ; T, v; O, o; N, w.

Consonants.

Gutturals, K, к; X, x; г, y.

Dentals,

Labials,

T, T; 0, 0; 4, 8; N, v.
II, π;, ; B, B; M, μ.

Semirowels, P, p; 4, λ.

Sibilants, , o, s; Z, L.

1 The primary source, as all scholars now agree, was the Hieratic Egyptian.

Thirdly, I give the Latin or Roman alphabet, which has modified the Greek, and now also has twenty-four simple symbols and one compound. It has both capital and small letters, and eight vowel-symbols, thus:

THE LATIN OR ROMAN ALPHABET.

Vowels.

A, a; E, e; I, i (J, j); U, u (V, v); 0, o; Y, y.

Consonants.

Gutturals K, k; C, c; G, g; Q, q.

Palatal

(J, j; properly a vowel-symbol).

Dentals T, t; D, d; N, n.

Labials P, p; F, f1; B, b; M, m.

Semivowels R, r; L, 1; (Y, y; V, v; properly vowels).

Sibilants S, s; Z, z.

Aspirate h.

Compound X, x.

Fourthly ought to come our own English alphabet.

This need not be given as it is the same as the Latin alphabet, with the addition of the symbol W, w, and with the understanding that we also use J, j; V, v; Y, y as consonants; making twenty-six symbols in all.

All the Indian languages of the Aryan family have adopted the Nagarī, though with various modifications.

On the other hand, the South Indian or Dravidian family (e.g. Telugu, Kanarese and Malayalam) have not adopted the Nagari. They have an equally complete equipment of symbols similarly classified. The Tamil alone is defective, and makes use of Grantham characters in writing Sanskrit.

As a matter of fact, however, even the Nagari is by no means a perfect alphabetical system. It has no symbols for the sounds of short ă, ě and Ŏ (as in our 'hat,' 'let,' 'on'), nor for some of the vowel sounds common in German and French, nor for our sounds of ≈ and f, nor for the French sound of j, and it cannot adequately represent 1 Partly dental.

some of the sounds common in Semitic and Turanian 1 languages. Nor has it any capital letters. It has doubtless a great advantage over the Greek and Latin alphabets in its ampler supply of vowel-symbols. Yet from its making short ǎ inherent in every consonant (which is not marked by virama), it is encumbered by far too great a burden of intricate compound letters.

All that I am at present concerned in asserting is that the Nagari, is the best and only standard to be taken as a guide in amplifying the Latin graphic system and adapting it to the transliteration of Indian languages.

The inadequacy of the Greek, Latin, and English equipment of letters, when unamplified by supplementary symbols, to accomplish all the purposes which every scientifically formed graphic system ought to serve-that is, to represent its own language in such a way that each sound shall invariably be represented by the same symbol-is apparent when we arrange these three alphabetical systems (as I have attempted to do) in the same order with the more perfect Nāgari.

This is more especially brought home to us when we realize the fact that in our own language, which, like Sanskrit, Greek and Latin, belongs to the Aryan family, we have only 26 letters to represent about 40 sounds, while, with curious perversity, instead of adding 14 symbols or distinctive marks to our 26 letters, we employ about 400 combinations of these letters to represent our 40 sounds.

The need of a fuller number of symbols is especially noticeable in respect of the vowel-sounds. To remedy this deficiency, and avoid violating the rules laid down at the beginning of this paper, it is usual in German and French to combine simple vowel-symbols together (thus, ai, au, eu, ou, etc.), or to make use of distinctive dots, marks and accents. (as in German and French, thus ö, ü, è, é, é, etc.), according to one invariable method and with some regard for scientific accuracy.

I use the name Turanian-though unsatisfactory-as the best yet invented to comprehend all non-Aryan and non-Semitic languages.