Our knowledge of most of the modern vernaculars of India has been much extended during the interval which has elapsed since the last edition of this Gazetteer was published. Not only have the highways of inquiry been widened and more clearly defined, but pioneers have ventured into the little-touched jungle of uncultivated dialects. There they have opened out paths which have sometimes led to unexpected results, and have disclosed secrets little suspected by those whose feet were necessarily confined to the main track that had previously been laid down with so much skill and energy. The progress has been most conspicuous in regard to the Aryan languages. The late Mr. Beames’s Comparative Grammar, a book to the learning and lucidity of which worthy tribute was paid in 1886, was quickly succeeded by the similar work of Dr. Hoernle. The Grammar of Eastern Hindi, written by that eminent scholar, occupied much the same ground as the volumes of Mr. Beames, but carried the inquiries farther, and cast the main results into a form which has ever since been almost universally accepted. What has subsequently been done has principally dealt with matters of detail, or with the investigation of new languages of which satisfactory grammars did not previously exist.

Our knowledge of the Indo-Chinese languages has also made considerable progress. The Assam Government has liberally encouraged the production of textbooks of the forms of speech current in that polyglot territory; and, in Europe, scholars like Professor E. Kuhn, of Munich, Professor Conrady, formerly of Leipzig, and Pater W. Schmidt, of Vienna, have succeeded in reducing to something like order the amazing confusion which hitherto existed in this department of philology. The Mundā languages, too, have received considerable attention. New grammars and dictionaries have seen the light, and, in Europe, Scandinavian scholars have made a special study of this family of tongues. Theories of the most wide-reaching significance have been put forth concerning them, but these have not yet all earned general acceptance.

In regard to the Dravidian languages, on the other hand, our knowledge has been almost stationary. Bishop Caldwell’s monumental Comparative Grammar of the Dravidian Languages, which was fully utilized in the last edition of this Gazetteer, still remains our one authority. Only a few grammars of unimportant tribal dialects, such as Gond, Kurukh, and Kandh, together with one or two more grammars and dictionaries of the well-known classical languages of Southern India, have appeared during the past two decades.

The final word has not, however, been said regarding any of the Indian vernaculars, not even the Aryan ones. While we know a good deal about some of the languages, our information as to the dialects is, with one or two exceptions, most incomplete. Even in respect to the forms of speech with which we are familiar, and whose habitats are matters of commonplace, we often do not know where these habitats begin or end. There are many languages, too, spoken by wild tribes of the Hindu Kush, or of Further India, of which we know little or nothing except the names. A consideration of these facts has led the Government of India to commence a systematic survey of all the forms of speech employed in Northern and Eastern India, and in the Presidency of Bombay. This is rapidly approaching completion, and we may hope that its results when published will materially increase the world’s information regarding one of its most interesting language-fields. So far as these results are available, they have been incorporated in the present chapter.

All this is a subject about which natives of India, a land whose literary glory may almost be said to be founded on the labours of its indigenous grammarians, are curiously incurious. Few natives at the present day are able to comprehend the idea connoted by the words ‘a language.’ Dialects they know and understand. They separate them and distinguish them with a meticulous, hair-splitting subtlety, which to us seems unnecessary and absurd; but their minds are not trained to grasp the conception, so familiar to us, of a general term embracing a number of interconnected dialects. It is as if we, in England, spoke of ‘Somersetshire’ and ‘Yorkshire’ dialects, but never used the term ‘English language.’ It thus follows that, while the dialect-names in the following pages have been taken from the indigenous nomenclature, nearly all the language-names have had to be invented by Europeans. Some of them, such as ‘Bengali,’ ‘Assamese,’ and the like, are founded on words which have received English citizenship, and are not real Indian words at all; while others, like ‘Hindostānī,’ ‘Bihārī,’ and so forth, are based on already existing Indian names of countries or nationalities.

Five great families of human speech have their homes, as vernaculars, in India. These are the Aryan, the Dravidian, the Mundā, the Mon-Khmer, and the Tibeto-Chinese. If, under the name of ‘India,’ we include the territories subject to Aden, we have to add at least two more, the Semitic and the Hamitic. These families will now be described in the above order. The oldest languages of India are probably those which we class as Mundā, and if we arranged our subjects according to priority of occupation, we should have to commence with them. But practical reasons compel us to begin with the Aryan forms of speech, for, whether we consider the influence which they have exercised upon the development of Indian civilization, or the total number of their speakers, they are by far the most important.

The modern Aryan vernaculars, although derived from languages which were highly synthetical in structure, with grammars as complicated as those of Latin or Greek, are now essentially analytical. As was said in the last edition of this work, the terminals of their nouns and verbs have given place to postpositions, and to disjointed modern particles to indicate time, place, and relation. The process was spontaneous, and it represents the natural course of the human mind. ‘The flower of synthesis,’ to use the words at once eloquent and accurate of Mr. Beames, ‘budded and opened; and when full-blown began, like all other flowers, to fade. Its petals, that is, its inflexions, dropped off one by one; and in due course the fruit of analytical structure sprung up beneath it, and grew up and ripened in its stead.’

Originally the patois of pastoral tribes who found their way across the Hindu Kush, these tongues have spread over the whole of Northern India as far as Dibrugarh in the extreme east of Assam, and reaching south to Kanara in Bombay. While the speakers have in most instances succumbed to the influences of climate, and have lost their ethnical type by intermixture with the numerically superior aborigines, the languages have preserved their identity, and have superseded, and are still superseding, the indigenous forms of speech. When an Aryan tongue comes into contact with an uncivilized aboriginal one, it is invariably the latter which goes to the wall. The Aryan does not attempt to speak it, and the necessities of intercourse compel the aborigine to use a broken ‘pigeon’ form of the language of a superior civilization. As generations pass this mixed jargon more and more approximates to its model, and in process of time the old aboriginal language is forgotten and dies a natural death. At the present day, in ethnic borderlands, we see this transformation still going on, and can watch it in all stages of its progress. It is only in the south of India, where aboriginal languages are associated with a high degree of culture, that they have held their own. The reverse process, of an Aryan tongue being superseded by an aboriginal one, never occurs.

The Aryan languages form one branch of the great Indo-European family of speeches. The original home from which the populations whom we now group together under the title of ‘Indo-European’¹ spread over Europe and parts of Western and Southern Asia has been a subject of long discussion, extending over many years. It has been located on the Caucasus and on the Hindu Kush. Other scholars maintained that it was in North-western Europe. Others have claimed Armenia and the country round the Oxus and Jaxartes as the centre of dispersion. The latest researches tend to show that the oldest domicile of the Indo-Europeans is probably to be sought for on the common borderland of Europe and Asia—in the steppe country of Southern Russia. Here they were a pastoral people; here some of their number took to agricultural pursuits; and from here they wandered off to the east and to the west.

From the point of view of language, the first great division of the Indo-Europeans was into the so-called centum-speakers and satem-speakers. The former, who originally began the word for ‘hundred’ with the letter k, travelled westwards and do not concern us. The latter, who expressed the same idea with some word beginning with a sibilant, mostly wandered to the east, and from their language have descended the speech-families which we call Aryan, Armenian, Phrygian, Thracian, Illyrio-Albanian, and Balto-Sclavonic. We have only to do with the first of these six.

One of the clans of these satem-speakers, who called themselves Aryans, migrated eastwards, probably by a route north of the Caspian Sea. They settled in the country lying on the banks of the Jaxartes and the Oxus, and we may, with some certainty, name the oasis of Khiva as one of their most ancient seats. Thence, still a united people, they worked their way up the courses of these rivers into the highlands of Khokand and Badakhshan, where they split up into two sections, one portion marching south, over the Hindu Kush, into the valley of the Kabul, and thence into the plains of India, and the other eastwards and westwards, towards the Pāmirs and towards what is now Merv and Eastern Persia. After the separation, the common Aryan speech developed on two different lines, and became, on the one hand, the parent of the Indo-Aryan, and, on the other hand, the parent of the Eranian (often spelt ‘Iranian’) family of languages.

The Eranians who journeyed eastwards penetrated even as far as Yarkand, but their language, as a national speech, has survived only in the Pāmirs, and its eastern limit may be taken as Sariqol. Those who travelled to the west ultimately occupied not only Merv, but the whole of Persia and Baluchistan, and nearly the whole of Afghanistan. At the earliest period of which we have documentary evidence, we find Eranian divided into two not very different dialects, commonly called Persic and Medic. Persic was the official language of the Court of the Achaemenides, and was employed by Darius I (B.C. 522–486), in the celebrated Behistun inscription. It developed into the Middle Persian or Pahlavi of the Sassanids (third to seventh centuries A.D.), and finally became modern Persian. Persian is not a vernacular of India; but under Musalman dominion it became one of the great vehicles of Indian literature, and some of the most famous Persian books, including the great lexicographical works, have been composed in Hindustan. Medic, on the other hand, was the language of the Avesta. It was spoken not only in Media (North-western Persia), but all over East Eran. From it are descended the two great Eranian languages belonging to India—Pashto and Baloch; and also, besides others, the so-called Ghalchah languages of the Pāmirs and Sariqol.

Commencing from the south, the first of these is Baloch. It is in its outward shape the most archaic of all the Eranian tongues, still possessing forms which fifteen hundred years ago had already begun to decay in the cognate Persian. As its name implies, it is the principal language of Baluchistan, and is geographically split up by the Dravidian-speaking Brāhūis of the central hills into two dialects—that of the north, and that of Makrān in the south and west. Its southern boundary is the Arabian Sea, from near the Indus to about the fifty-eighth degree of east longitude. Northwards it extends to near Quetta, and as we go westwards it is found even farther than this, up to the valley of the Helmand. The Indus valley itself is occupied by speakers of Indo-Aryan languages, but the eastern boundary of Baloch follows the course of that river at a short distance to the west up to about Dera Ghāzī Khān. The northern dialect is much more rich in Indian loan-words than is Makrānī, and both dialects borrow freely from Arabic and Persian, words from the former often appearing in curiously distorted forms. Baloch can hardly be called a written language, although both the Persian and the Roman alphabets have been employed for transcribing it. The number of speakers of Baloch returned at the Census of 1901 was 152,188.

To the north of Baloch lies Pashto, the main language of British and independent Afghanistan. In the latter it is not the vernacular of the Hazara country or of the tract lying to the north of the Kābul river, including Laghmān and Kāfīristān, but elsewhere it is in general use. It is the principal language of Swāt and Buner, and of the country to the west of the Indus as far south as Dera Ismail Khan. The Indus is almost, but not quite, the eastern boundary; for, while the valley itself in its lower course is peopled by speakers of Indo-Aryan dialects, in the north Pashto has crossed the river and occupied parts of the British Districts of Hazara and Rāwalpindi. As a lingua franca it is in common use still farther up the Indus, at least as far as the junction with the river Kāndia where the Indus turns to the south. It was returned as spoken by 1,224,807 people in British India at the Census of 1901, the area in which it is employed being bilingual. Pashto is spoken by Pathāns, while the Hindus employ an Indo-Aryan dialect locally known as Hindko.

Unlike Baloch, Pashto is a written language possessing an alphabet of its own based on that employed for Persian, and has a fairly copious literature. It has been the subject of considerable study, not only by English scholars, but also by Russians, French, and Germans. The rugged character of its sounds suits the nature of its speakers and of the mountains which form their home, but they are most in-harmonious to the fastidious Oriental ear. Although harsh-sounding, it is a strong, virile language, which is capable of expressing any idea with neatness and accuracy. It is less archaic in its general characteristics than Baloch, and has borrowed not only much of its vocabulary, but even part of its grammar, from Indian sources. It has two recognized dialects, a north-eastern, or Pakhto, and a south-western, or Pashto, which differ little except in pronunciation, the two names being typical examples of the respective ways of uttering the same word. Each has many tribal sub-dialects, which again differ merely in the pronunciation of the vowels. There is, for instance, the Afridi sub-dialect, noted for the broad sound of its $ā$; while the Wazīris change every $ā$ to $ō$, and every $ū$ to $ī$.

The Pathāns have been identified with the Pakthas, a tribe mentioned in the Rig-veda, and with the Pāktuves of Herodotus; while the ‘Ἅπάνται’ of the Father of History are probably the same as the Afrīdīs, or, as they call themselves, Aprīdis.

Allied to Pashto, although quite a distinct language, is Ormuri, spoken by a small tribe settled round Kānigoram in Wazīristān. It is employed by members of the Bargistā tribe, who claim to be descendants of the Bārakīs that accompanied Mahmūd of Ghazni in his invasions of India. These Bārakīs are said to have taken a prominent part in the capture of the famous gates of Somnāth, and, pleased at the service rendered by them, the Sultan gave them a perpetual grant of the country round Kānigoram. The language, like Pashto, belongs to the Medic branch of Eranian speech. It is even more inharmonious than Pashto, and possesses one consonant, imperfectly represented in English letters by kshr, which even Pathān mouths find difficult to pronounce.

The only other Eranian languages with which we are called upon to deal are the Ghalchah languages of the Pāmirs. The home of these tongues, Wakhi, Shighni, Sariqoli, Ishkashami, and Munjān, is beyond the British frontier; but the last-named has crossed the Hindu Kush by the Dora pass, and is also spoken in the Leotkuh valley of Chitrāl, where it is known as Yüdghā. This differs considerably from the standard language of Munjān, and has developed into an independent dialect. The spill of an Eranian language over the great watershed of the Hindu Kush is but a repetition of what occurred centuries ago when the Aryans first settled in the Pāmirs. At that early time, if linguistic evidence may be accepted, some of these Aryans crossed the passes and settled in what is now Laghmān, Kāfristān, Chitrāl, Gilgit, and Kashmīr. They migrated at a period when all the typical characteristics of Eranian languages had not yet become fixed, and in their new home their tongue developed on its own lines, partly Eranian and partly Indo-Aryan. The Aryans of India proper, who had entered the Punjab by the valley of the Kābul, had little intercourse or sympathy with these tribes, and nicknamed them Pisāchas, or flesh-eaters, ὁμοφάγοι, and in later years gruesome traditions attached to the name.

These Piśācha tribes must at one time have extended to some distance beyond their present seats. Sanskrit writers mention colonies of them in the Western Punjab and in Sind, and examples of the dialects spoken by them are found in the words which the Greeks employed to record names heard by them in North-western India, and in the versions of the inscriptions of Asoka found in the same locality. Indeed, there are traces of their influence still existing in the modern vernaculars of the Lower Indus valley. At the present day the languages are found only in the country between the Punjab and the Hindu Kush. They possess an extraordinarily archaic character. Words are still in everyday use which are almost identical with the forms they assumed in Vedic hymns, and which now survive only in a much corrupted state in the plains of India.

In their essence these languages are neither Eranian nor Indo-Aryan, but are something between both. In the southern portion of the area in which they are spoken they are much mixed with Indian idioms; and this is specially the case with Kāshmīrī, which has only a Piśācha substratum, overlaid by another language of Indian origin, which so effectually conceals the original basis, that Kāshmīrī must now be considered as Indo-Aryan, and not as belonging to the Piśācha group.

The true Piśācha languages of the present day are Pashai, spoken in Laghmān of Afghanistan; a number of Kāfir dialects, of which the principal are Bashgali, Wai, and Kalāshā; Khowār, the language of Chitrāl; and Shīnā, that of Gilgit and the neighbourhood. It is Shīnā which is the basis of Kāshmīrī, and it is also the foundation of several mixed dialects, spoken in the Indus and Swāt Kohistān, which are now being superseded by Pashto. Khowār occupies a somewhat independent position in regard to the others, while the Kāfir dialects, of which there are at least five, differ considerably among themselves. Wai-Veri, the most western of them, in some phonetic peculiarities shows points of agreement with the purely Eranian Munjānī. All the Piśācha languages are without literatures, and have been reduced to writing only in the past few years by European scholars. At the same time it may be remarked that the great collection of Indian folk-lore entitled the Brihat Kathā, of which no copy is known to exist at the present day, is said by tradition to have been composed in the tongue of the Piśāchas.

Returning to the immigration of the Indo-Aryans through the Kabul valley from the west, it is not suggested that this took place all at once. On the contrary, it was a gradual affair extending over centuries. The latest comers would not necessarily be on good terms with their predecessors, who quite possibly opposed them as intruders, nor did they speak the same language. At the earliest period of which we have any cognizance, we see the Punjab peopled by various Indo-Aryan tribes, one at enmity with another, and sometimes alluding to its opponent as a set of unintelligible barbarians.

In Sanskrit geography India is divided into the Madhya-deśa, or ‘Midland,’ and the rest. The Midland is constantly referred to as the true pure home of the Indo-Aryan people, the rest being, from the point of view of Sanskrit writers, more or less barbarous. The Midland extended from the Himālayas on the north to the Vindhya Hills on the south, and from Sahrind (vulgo Sirhind) in the Eastern Punjab on the west to the confluence of the Ganges and the Jumna on the east. It thus consisted of the Gangetic Doab, and of the country immediately to its north and south. The population of this tract had expanded from its original seat near the Upper Doab and the sacred river, the Saraswati. The particular Indo-Aryan dialect of these people developed into the modern language of the Midland. It also received literary culture from the most ancient times, and became fixed, in the form of Sanskrit (literally the ‘purified’ language), by the labours of grammarians, which may be said to have culminated in the work of Pānini about the year 300 B.C. Sanskrit thus represents a polished form of an archaic tongue, which by Pānini’s time was no longer a vernacular, but which, owing to political reasons and to the fact that it was the vehicle of literature, became a second language understood and used by the educated in addition to their mother tongue, and has so continued with a fluctuating popularity down to the present day. We may take the language of the Rig-veda as representing the archaic dialect of the Upper Doab, of which Sanskrit became the polished form. It was a vernacular, and, besides receiving this literary cultivation, underwent the fate of all vernaculars. Just as the spoken dialects of Italy existed side by side with Latin, and, while the evolution of Latin was arrested by its great writers, ultimately developed into the modern Romance languages, so the ancient Vedic form of speech developed first into that stage of language known as Prākrit, and then into one or more modern Indo-Aryan vernaculars. It is thus a mistake to say that any modern Indian language is derived from Sanskrit. The most that can be said is that it and Sanskrit have a common origin.

So far for the language of the ‘Midland.’ Round it, on three sides—west, south, and east—lay a country inhabited, even in Vedic times, by other Indo-Aryan tribes. This tract included the modern Punjab, Sind, Gujarat, Rajputana and the country to its east, Oudh, and Bihar. Rajputana belongs geographically to the Midland, but it was a late conquest, and for our present purposes may be considered as belonging to the Outer Band. Over this band were scattered different tribes, each with its own dialect; but it is important to note that a comparison of the modern vernaculars shows that these outer dialects were all more closely related to each other than any of them was to the language of the Midland. In fact, at an early period of the linguistic history of India there must have been two sets of Indo-Aryan dialects—one the language of the Midland, and the other the group of dialects forming the Outer Band. From this it has been argued, and the contention is entirely borne out by the results of ethnological inquiries, that the inhabitants of the Midland represent the latest stage of Indo-Aryan immigration. The earliest arrivals spoke one dialect, and the new-comers another. According to Dr. Hoernle, who first suggested the theory, the latest invaders probably entered the Punjab like a wedge, into the heart of the country already occupied by the first immigrants, forcing the latter outwards in three directions, to the east, to the south, and to the west.

The next process which we observe in the geographical distribution of the Indo-Aryan languages is one of expansion. The population of the Midland increased, and history shows that it exercised an important influence over the rest of India. The imperial cities of Delhi and Kanauj, and the holy city of Mathurā (Muttra), the Μόδουρα of Ptolemy, lay within its territory. With increased population and increased power it expanded and conquered the Eastern Punjab, Rājputāna and Gujarat (where it reached the sea, and gained access to maritime commerce), and Oudh. With its armies and with its settlers it carried its language, and hence in all these territories we now find mixed forms of speech. The basis of each is that of the Outer Band, but its body is that of the Midland. Almost everywhere the nature of the phenomena is the same. In the country near the borders of the Midland, the Midland language has overwhelmed the ancient language, and few traces of the latter can be recognized. As we go farther from the centre, the influence of the Midland weakens and that of the Outer Band becomes stronger and stronger, till the traces of the Midland speech disappear altogether. The present language of the Eastern Punjab is closely allied to that of the Upper Doab, but it gradually becomes the Lahndā of the Western Punjab, which has nothing to do with the Midland. So the language of North-eastern Rājputāna is very similar to that of Agra, but as we go south and west we see more and more of the original language of the Outer Band, until it is quite prominent in Gujarat. Again, in Oudh, which was a country with a literature and history of its own, there is a mixture of the same nature, although here the Midland language has not established itself so firmly as it has in the west and south.

Finally, where possible, the inhabitants of the Outer Band also expanded to the south and east. In this way we find Marāthī in the Central Provinces, Berār, and Bombay; and, to the east, Orīyā, Bengali, and Assamese, all of them true Outer languages unaffected in their essence by the speech of the Midland.

The state of affairs at the present day is therefore as follows:—There is a Midland Indo-Aryan language, occupying the Gangetic Doāb and the country immediately to its north and south. Round it on three sides is a band of Mixed languages, occupying the Eastern Punjab, Gujarat, Rājputāna, and Oudh, with extensions to the south in Baghelkhand and Chhattisgarh. Again, beyond these, there is a band of Outer languages, occupying Kashmir, the Western Punjab, Sind (here it is broken by Gujarāt), the Marāthā country, Orissa, Bihār, Bengal, and Assam. To these should be added the Indo-Aryan languages of the Himālayas north of the Midland, which also belong to the Intermediate Band, being recent importations from Rājputāna. The Midland language is therefore now enclosed in a ring fence of intermediate forms of speech.

We have seen that the word ‘Sanskrit’ means ‘purified.’ Opposed to this is the word ‘Prākrit,’ or ’natural, unartificial.’ ‘Prākrit’ thus connotes the vernacular dialects of India as distinguished from the principal literary form of speech. The earliest Prakrit of which we have any cognizance is the Midland vernacular current during the Vedic period. We have no record of the contemporary Prakrits of the Outer Band. We may call all these vernaculars (including the tongue of the Midland) the Primary Prakrits of India. These Primary Prakrits were in a linguistic stage closely corresponding to that of Latin as we know it. They were synthetic languages, with fairly complicated grammars, and with no objection to harsh combinations of consonants. In the course of centuries they decayed into what are called Secondary Prakrits. Here we find the languages still synthetic, but diphthongs and harsh combinations are eschewed, till in the latest developments we find a condition of almost absolute fluidity, each language becoming an emasculated collection of vowels hanging for support on an occasional consonant. This weakness brought its own nemesis and from, say, 1000 A.D. we find in existence the series of modern Indo-Aryan vernaculars, or, as they may be called, Tertiary Prakrits. Here we find the hiatus of contiguous vowels abolished by the creation of new diphthongs, declensional and conjugational terminations consisting merely of vowels worn away, and new languages appearing, no longer synthetic, but analytic, and again reverting to combinations of consonants under new forms, which had existed three thousand years ago, but which two thousand years of attrition had caused to disappear.

Returning to the Secondary Prākrits, they existed from, at least, the time of the Buddha (550 B.C.) down to about 1000 A.D. During these fifteen hundred years they passed through several stages. The earliest was that now known as Pali. Two hundred and fifty years before Christ, we find the edicts of Asoka written in a form of this language, and it then had at least two dialects, an eastern and a western. In this particular stage of Pali one of the Secondary Prākrits was crystallized by the influence of Buddhism, which employed it for its sacred books. As vernaculars, the Secondary Prākrits continued the course of their development, and in a still more decayed form reached the stage of what, in various dialects, is known as The Prākrit par excellence. When we talk of Prākrits, we usually mean this later stage of the Secondary Prākrits, when they had developed beyond the stage of Pali, and before they had reached the analytic stage of the modern Indo-Aryan vernaculars.

At this stage, so far as materials are available, we notice the same grouping of the Prākrit dialects as exists among the vernaculars of the present day. We have no definite information what was the language of the Punjab; but as for the rest of India, there was a Prakrit of the Midland, the so-called Sauraseni, called after the Sanskrit name, Sūrasena, of the country round Mathurā (Muttra). It was close to the great kingdom of Kanauj, the centre of Indo-Aryan power at this time. To its south and east was a band of dialects agreeing in many points among themselves, and also in common points of difference when compared with Sauraseni. These were—in the east, in the country now called Bihār, Māgadhī; in Oudh and Baghelkhand, Ardhamāgadhī; and, south of Ardhamāgadhī and Sauraseni, Māhārāshṭrī with its head quarters in Berār. Ardhamāgadhī, as might be expected, was partly a mixed language, showing signs of the influence of Sauraseni, but, in all its essential points, its relationship with Māgadhī is undoubtedly clear. Māhārāshṭrī was closely connected with Ardhamāgadhī, which formed the connecting link between it and Māgadhī, but in its rather isolated position it struck out on somewhat independent lines. It is important to remember that it (under the name of Sauraseni) was once the language of Gujarat, before that country was overwhelmed by the invasion from the Midland.

Vidarbha, or Berār, the home of Māhārāṣṭrī, was the seat of a powerful kingdom, whose rulers encouraged literature, not only in Sanskrit but also in the vernacular. Māhārāṣṭrī received culture at an early period. In its native land it became the vehicle of some of the most charming lyrics ever composed in an Indian tongue; and its popularity carried it over the whole of Hindustān, where it was employed both for epic poetry and also by the later Jain religious writers. But it is best known from the Indian dramas, in which, while most of the vernacular prose was written in Sauraseni, the language of the Midland, the songs are usually in Māhārāṣṭrī¹.

The next and last stage of the Secondary Prākrits was that known as ‘Literary Apabhrāṃśa.’ ‘Apabhrāṃśa,’ meaning ‘corrupt’ or ‘decayed,’ was the title given by Indian grammarians, after the Prākrits had begun to receive literary culture, to the true vernaculars on which these polished literary dialects were founded. Ultimately, these Apabhrāṃśas became themselves employed in literature, and were even studied by native grammarians, successors of those who in previous generations had despised them. This was a mere repetition of history. Sanskrit became fixed, and in time ceased to be generally intelligible. Then the vernacular Pali was used for popular literature. When literary Pali became generally unintelligible, the vernacular Prakrit was employed for the same purpose. Prakrit itself became crystallized, and in the course of generations had to yield to Apabhrāṃśa. While the earlier Prakrits had been manipulated for literary purposes by the omission of what was considered vulgar and by the reduction of wild luxuriance to classical uniformity, so that the result was more or less artificial, the Apabhrāṃśas were not nearly so severely edited, and the sparse literature which has survived affords valuable evidence as to the actual spoken language at the time of its committal to writing. The modern vernaculars are the direct children of these Apabhrāṃśas. The Śaurasena Apabhrāṃśa was the parent of Western Hindi and Panjābī. Closely connected with it were Avanti, whose head quarters were round what is now Ujjain, the parent of Rajasthānī; and Gaurjari, the parent of Gujarāti. The remaining intermediate language, Eastern Hindi, is sprung from Ardhamāgadha Apabhrāṃśa.

Turning to the Outer Band, an unnamed Apabhramśa was the parent of Lahndā and Kāshmiri, the latter, as has been said, having as its base some Piśācha language akin to Shīnā, over which the modern language lies as a second layer. Sindhī is derived from a Vrāchāda Apabhramśa spoken in the country of the lower Indus, and Marāṭhī is the child of the Apabhramśa of Mahārāṣṭrā. In the east, the great Magadha Apabhramśa is not only the parent of Bihārī in its proper home, but has also branched out in three directions. To the south it became Oriyā; to the south-east it developed into the Bengali of Central Bengal; while to the east, keeping north of the Ganges, its children are Northern Bengali, and, farther on, Assamese. These three branches can be distinctly traced. In some respects Oriyā and Northern Bengali preserve common features which have disappeared in Central Bengal.

Concurrently with the development of the Indo-Aryan vernaculars, we have Sanskrit, the literary language of the Brāhmanical schools, endowed with all the prestige which religion and learning could give it. In earlier times its influence was strongest in its proper home, the Midland. Allowing for phonetic corruption, the vocabulary of Sauraseni Prākrit is practically the same as that of Sanskrit. The farther we go from the Midland the more strange words we meet, words which are technically known as deśya or ‘country-born.’ These, though Indo-Aryan, are not descended from the particular archaic dialect from which Sanskrit sprang, but belong to the vocabularies of the dialects of distant parts of India which were contemporary with it. On the other hand, the prestige of the literary Sanskrit has exercised a constant influence over all the Aryan vernaculars of India. Universally, but wrongly, believed to be the parent of all of them, the would-be children have freely borrowed words from the vocabulary of their adoptive parent, and this tendency received an additional impetus with the revival of learning which dates from the early part of the last century. In some of the modern languages it then became the fashion to eschew as much as possible all honest vernacular words derived from the Prākrits, and to substitute borrowed Sanskrit words, much as if a Frenchman were to substitute the Latin siccus for his own sec, or as if an Englishman were to use the Anglo-Saxon hliford instead of ’lord.’ Native grammarians call these borrowed words tatsamas, or ’the same as “that” (sc. Sanskrit),’ while the true vernacular words derived from Prākrit are tadbhavas, or ‘having “that” (sc. Sanskrit) for its origin.’ We thus see that the Aryan portion of the vocabulary of a modern Indo-Aryan vernacular is composed of three elements: tatsamas, tadbhavas, and deśyas. The distinction is of some importance, for the literary language of some of them, such as Bengali, is so overloaded with the fashionable tatsamas that it may almost be called a national misfortune. For the sake of a spurious dignity the written word has been rendered unintelligible to the vast multitudes who have not received the education imparted by the higher schools.

Other languages have contributed their quotas to the Indo-Aryan vernaculars. Many words have been borrowed from Dravidian languages, generally in a contemptuous sense. Thus the common word pillā, ‘a cub,’ is really a Dravidian word meaning ‘son.’ The most important additions have come from Persian, and through Persian from Arabic. These are due to the influence of Mughal domination, and their use is universal. Every peasant of Northern India employs a few, while the literary Urdu of Lucknow is so full of them, that little of the true vernacular remains except an occasional postposition or auxiliary verb. A few words also have been borrowed from Portuguese, Dutch, and English, often in quaintly distorted forms. Few Englishmen would recognize the railway term ‘signal’ in sikandar, which also, as a true Hindostānī word, means ‘Alexander the Great.’

We thus arrive at the following list of the modern Indo-Aryan vernaculars:

Of these, the Pahāri languages are offshoots of Rājasthānī spoken in the Himālayas. Kohistānī includes the mixed dialects of the Swāt and Indus Kohistānī. The Census of 1901 did not extend to these tracts, and hence few speakers were recorded. We now proceed to consider each of these forms of speech in the order of the above list.

The word ‘Hindi’ is very laxly employed by English writers. It properly means ‘Indian,’ and can be used to signify any Indian language. By Europeans it is sometimes reserved for a particular form of Hindostānī which will be described below, but is more often employed as a vague term to denote all the rural dialects of the three languages—Bihārī, Eastern Hindi, and Western Hindi—spoken between Bengal proper and the Punjab. In the present pages it is used only in the former of these two senses; that is to say, as meaning that form of Hindostānī which is the prose literary language of those Hindus who do not employ Urdu. In English ‘Hindi’ is specially applied to the languages of Oudh and of the Midland, and, to avoid the introduction of a strange terminology, these are here called ‘Eastern Hindi’ and ‘Western Hindi,’ respectively. They are two quite distinct languages.

Western Hindi is, therefore, the modern Indo-Aryan vernacular of the old Midland, i.e. of the Gangetic Doab and the country to its north; and, as in ancient times, it is by far the most important of all the languages of India. It is true that speakers of Bengali exceed in number those whose vernacular is Western Hindi, but the forty millions shown above by no means exhaust the number of speakers of the latter. Bengali is confined to its own Provinces; but Hindostānī, the principal dialect of Western Hindi, is not only a local vernacular, but is also spoken over the whole of the north and west of continental India as a second language, a lingua franca employed alike in the court and in the market-place by every one with any claim to education. Hindostānī is that dialect of Western Hindi whose home is the Upper Gangetic Doab, in the country round Meerut. The city of Delhi lies close to the southern border of this tract. Here the dialect was in general use, and from here it was carried everywhere in India by the lieutenants of the Mughal empire. It has received considerable literary cultivation at the hands of both Musalmāns and Hindus. The former employed the Persian character for recording it, and enriched its vocabulary with a large stock of Persian and Arabic words. When this infusion of borrowed words is carried to an extreme, as is the fashion, for instance, in Lucknow, the language is intelligible only to educated Musalmāns and to those Hindus who have been educated on Musalmān lines. This Persianized form of Hindostānī is known as Urdū, a name derived from the Urdū-e mu’alla, or royal military bazar outside Delhi Palace, where it took its rise. When employed for poetry, Urdū is called Rekhta (‘scattered’ or ‘crumbled’), from the manner in which Persian words are scattered through it. The extreme Persianization of Urdū is due to Hindu rather than to Musalmān influence. Although Urdū literature is Musalmān in its origin, the Persian element was first introduced in excess by the pliant Hindu Kāyasths and Khatts employed in the Mughal administration and acquainted with Persian, rather than by Persians and Persianized Mughals, who for many centuries used only their own language for literary purposes. In the Deccan, even where Dravidian languages are the principal vernaculars, Urdū is very generally employed by Musalmāns,

and here Urdû literature took its rise. ‘Dakhînî Hindostânî,’ as it is called, differs somewhat from the modern standard of Delhi and Lucknow, and retains several archaic features which have disappeared in the north. During the first centuries of its existence Urdû literature was entirely poetical. Prose Urdû owes its origin to the English occupation of India, and to the need of textbooks for the College of Fort William. The Hindi form of Hindostânî was invented at the same time by the teachers at that college. It was intended to be a Hindostânî for the use of Hindus, and was derived from Urdû by ejecting all words of Arabic and Persian birth, and substituting in their place words borrowed or derived from the indigenous Sanskrit. Owing to the popularity of the first book written in it, and to its supplying the need for a lingua franca which could be used by the strictest Hindus without their religious prejudices being offended, it became widely adopted and is now the recognized vehicle for writing prose by those inhabitants of Upper India who do not employ Urdû. Although originally differing from that language merely in vocabulary, it has in the course of a century developed some idioms of its own, so that it is not often that one finds a native who can write both forms of Hindostânî with equal correctness. Indeed, there is one well-known book, written by a Muhammadan, which does not contain a single Arabic or Persian word from cover to cover, and which is nevertheless considered by Hindu purists to be written in Urdû, because idioms are found in it belonging to that form of the dialect, and not to Hindi.

Urdû, as becomes its origin, is usually written in a modified form of the Persian character, while Hindi is generally written, like Sanskrit, in the Devanāgari character. While the former is enlisted into the service of both prose and poetry, the latter is employed only for prose. When a Hindu writes poetry he betakes himself to one of the naturally-born dialects of Eastern or Western Hindi, usually Awadhi or Braj Bhāshā. The name ‘Hindostānī,’ when connoting any particular form of speech, is properly reserved for a language whose vocabulary is neither excessively Persianized nor excessively Sanskritized.

The other dialects of Western Hindi are Bāngarū, Braj Bhāshā, Kanaujī, and Bundeli. The first is the language of the Bāngar, or highland of the South-eastern Punjab, immediately to the west of the Ganges. It is sometimes called Hariānī, and is much mixed with Panjābī and Rājasthānī. Of all the dialects, Braj Bhāshā is the nearest relative to Saurāsenī. It is spoken round Mathurā (Muttra) and in the Central Gangetic Doāb. It has a copious literature, mainly poetical, and was the principal literary form of Western Hindi employed by Hindus before the invention of Hindi. Kanaujī is almost the same as Braj Bhāshā. It is spoken in the lower part of the Central Doāb as far down as, say, Cawnpore, and in the country to its north. Bundelī is the dialect of the greater part of Bundelkhand, and also of a good portion of the Narbadā valley in the Central Provinces. It has a respectable literature.

As languages, Western Hindi, and its neighbour Eastern Hindi, rival English in their flexibility and copiousness. When not spoiled, as Western Hindi too often is, by an excessive display of Arabic and Persian or of Sanskrit words, they are two beautiful, vigorous forms of speech, not overburdened by complicated grammars, and capable of expressing any idea which the mind of man can conceive with ease, elegance, and crystal clearness. They both have enormous native vocabularies, and each has a complete apparatus for the expression of abstract terms. Their old literatures contain some of the highest flights of poetry and some of the most eloquent utterances of religious devotion which have found their birth in Asia.

Turning to the Intermediate languages, we first deal with those in which the language of the Midland is the predominant feature. Rājasthānī and Gujarāti may be considered together, as representing the flow of the inhabitants of the Midland to the south-west, to meet the sea. Rājputāna, in which Rājasthānī is spoken, is divided into many states and many tribes. Each claims to have a language of its own, but all these are really dialects of one and the same form of speech. They fall into four main groups—a northern, a southern, an eastern, and a western. The typical dialect of the north is Mewātī or Bighotā. Of all the dialects of Rājputāna it is, as might be expected, that which most nearly resembles Western Hindi. To the north-east it shades off into Braj Bhāshā, and to the north-west into Bāngarū. Mālī, the main dialect of Southern Rājputāna, is spoken in Mālīwā. Neither it nor Mewātī has any literature to speak of. In Eastern Rājputāna we have Jaipurī, with many sub-dialects, and many closely connected forms of speech with various names. The western dialect, Mārwārī, is by far the most important. It is the vernacular of Mārwārī, Mewār, Bikaner, and Jaisalmer, and its speakers, who are enterprising merchants and bankers, have carried it all over India. It is the most typical of the Rājasthānī dialects, and has a copious literature, written in a peculiar character, the aspect of which is familiar to every Indian official who has had occasion to inspect the accounts of native bankers.

Rajputana has sent out many colonies into Northern India. The most important are the inhabitants of the Himalayas from Chamba in the Punjab to Nepal. Some centuries ago bands of Rajputs at various times invaded and conquered these hills. They settled there and intermarried with the original inhabitants, on whom they imposed their language. The Rajasthani here transplanted has developed on independent lines, and was no doubt influenced by the form of speech which it superseded. What that form of speech was we do not know, except that we have some old plays in one of the original languages of Nepali. This was akin to what is now modern Bihari. The modern Rajasthani dialect now spoken in Nepali is called by Europeans ‘Naipali’—a wrong name, for it is not the main language of the country but is spoken only by the ruling classes. The other inhabitants employ various Tibeto-Burman dialects. Its speakers call it ‘Khas,’ from the name of one of the tribes which employ it. Farther west these dialects are simply called ‘Pahari,’ or ’the Language of the Hills.’ We have a Western Pahari spoken north of the Central and Eastern Punjab, and a Central Pahari north of the United Provinces. To these Khas may be added, under the name of ‘Eastern Pahari.’ Other offshoots of Rajasthani are Gujar, the language of the Gujars wandering with their herds over the mountains of Kashmir and the Swat valley; and Labhani, spoken by the Labhanias or Banjaras, the great carrying tribe of Central and Western India. There are numerous Gujars in the plains of the Punjab, where they have given their name to two Districts, but these nowadays speak ordinary Panjabi.

Mârwar is bounded on the west by the Indian Desert, beyond which we find Sindhi, one of the Outer languages, but to the south we enter easily into Gujarat. Gujarati, the language of this country, is the most western of those over which the language of the Midland exercises sway, and at its base we can see distinct traces of the old Saurāshṭrī Prākrit, which belonged to the Outer Band. Gujarati has a printed character of its own, modelled on the cursive form which Devanāgari takes all over Northern India, especially in Mārwār. Owing to the survival of a number of ancient grammars, we have a connected history of the language from the time when it first came into existence as a modern Indo-Aryan vernacular some nine hundred years ago. Literature has always flourished in Gujarat from very early times, and the modern vernacular presents no exception. The Bhils and the inhabitants of Khandesh speak mixed forms of speech which are dialects of Gujarati.

Of all the Intermediate languages, Panjābī is the one which most nearly agrees with the modern speech of the Midland. It is spoken in the Central Punjab, and is the vernacular of the Sikhs. Immediately to its west lies Lahndā, an Outer language, and the change from the one to the other is most gradual. It is quite impossible to fix a definite boundary between these two, but we may take the seventy-fourth degree of east longitude as an approximate conventional dividing line. Lahndā once extended far to the east, but, as has been explained, was there superseded by the language of the Midland, whose influence gradually diminished as it went westwards. It is this mixed language which became the modern Panjābī. Its proper written character is related to that employed in Mārwār. It is known as Lāndā, or ‘clipped’ (quite a distinct word from Lahndā, the name of the language of the Western Punjab), and is distinguished for its illegibility when once it is put upon paper. Only its writer, and not always he, can read Lāndā as commonly scrawled. An improved, and legible, form of Lāndā is known as Gurmukhī. This was invented about three hundred years ago for writing the Sikh scriptures, and is now the character in ordinary use for printing, although the Persian and the Devanāgarī are also employed. The standard Panjābī is that spoken in the neighbourhood of Amritsar; and the only real dialect is Sogri, the vernacular of the State of Jammu, and, with slightly varying inflexions, of a part of Kāngra. Of the languages connected with the Midland, Panjābī is the purest and most free from the burden of terms borrowed from either Persian or Sanskrit. While capable of expressing all ideas, it has a charming rustic flavour indicative of the national characteristics of the sturdy peasantry that use it.

The remaining Intermediate language is Eastern Hindi, which differs from the others in that it is based on the eastern languages of the Outer Band, and that the influence of the language of the Midland is not nearly so strong as in Rajputana and the Punjab. Here the two elements meet in nearly equal proportions. It is the language of Oudh, of Baghelkhand, and of Chhattisgarh in the Central Provinces, and has a long history behind it. It is the vernacular of the country in which the hero Rāmachandra was born; and the Jain apostle Mahāvira used an early form of it to convey his teaching to his disciples. The local Prakrit, Ardhamāgadhi, thus became the sacred language of the Jains. Its modern successor, Eastern Hindi, through the work of a great genius, became the medium for celebrating the Gestes of Rāma, and, in consequence, the dialect employed for nearly all the epic poetry of Hindustān. It is spoken nowadays not only in its own tract, but is also used by uneducated Musalmāns far to the east—right into the heart of Bihar; and Oudh men, who are accustomed to travel to distant parts in quest of service, have carried it far and wide over the whole of India. It is commonly heard even in the streets of Calcutta and Bombay.

Eastern Hindi has a great literature, probably larger than that of any other of the modern Indo-Aryan vernaculars; and this literature, being founded on the genuine tongue of the people, and acquiring no fictitious dignity by bastard additions of Sanskrit words, has reacted on the spoken language, so that the form of speech heard in the fields of Oudh possesses the characteristic beauties of poetry and clearness. Every Oudh rustic is soaked in his national literature, and quotations from his great writers fall more naturally from his lips than the words of Burns fall from those of a Scotsman. Overshadowed at the present day by the official Hindostānī, it has been studied by but few Europeans, but no one who has once wandered into its magic garden ever leaves it willingly.

In the Central Provinces, Eastern Hindi meets Marāṭhī and shades off into that language through a number of mixed dialects. It and Oriyā are the only forms of speech which are not separated from Marāṭhī by a distinct dividing line, and it thus still bears witness to the intimate relationship which existed between the Ardhamāgadhī and the Māhārāṣṭrī Prākrits two thousand years ago.

Eastern Hindi has three main dialects. Besides the standard Awadhi spoken in Oudh, there is the Baghelī of Baghelkhand, and the Chhattisgarhī of the eastern part of the Central Provinces.

It will have been noticed that the Outer languages have been divided in the table given on p. 364 into three sets, a north-western group, a southern language, and an eastern group. Owing to its somewhat isolated position, and to the influence of the Piśachha languages already referred to, the north-western group, although closely agreeing with the other two in its general structure, has struck out on independent lines. The most northern of the group is Kāshmīri, the language of the State of Kashmir. Tradition informs us that this country was originally inhabited by Piśachas, who must have spoken a tongue allied to Shīnā; but at an early period it suffered an invasion from the south, and was colonized by folk from the Punjab. The modern language fully bears this out. Although at the bottom we find a layer of Shinā words¹ and idioms, this is almost entirely hidden by an overlayer of a second language, closely allied to the Lahdā of the Western Punjab. Owing to the large number of broken vowels which it possesses, and to the changes which they undergo through the influence of others which follow them but are themselves silent, Kāshmīri is almost as difficult for a foreigner to pronounce as is English. It has an old literature of considerable extent, but the modern language has borrowed so freely from Persian and Arabic that the books written two or three centuries ago are hardly intelligible to natives at the present day. The bulk of the population is now Muhammadan, only a few Pandits preserving the memory of the ancient language. Kāshmīri has two or three dialects, of which the most important is Kishtwāri.

Kohistānī is the old language of the Indus and Swāt Kohistānī. It is now nearly superseded by Pashto, only a few tribes still employing it. Each of these has its own dialect. The country has not been thoroughly explored, and very little is known about these forms of speech. Like Kāshmīrī, they have a Shīnā basis, covered by an overlayer from the Western Punjab.

Lahndā or Western Panjābī is a language which appears under many names, such as Pokhwāri, Chibhālī, Jātkī, Multānī, or Hindko. None of these names is suitable, as each indicates only the dialect of some special tribe or of some special locality. ‘Lahndā,’ i.e. ‘Western,’ has been lately suggested, and has been tentatively adopted, although it, too, is far from satisfactory. The name ‘Western Panjābī’ suffers from the disadvantage of suggesting a connexion which does not exist with Panjābī proper. Lahndā is spoken in the Western Punjab as far east as, say, the seventy-fourth degree of east longitude. It once extended much farther to the east, but has there been superseded by the Midland language, from which the modern Panjābī has sprung. There is no definite boundary between these two languages. As explained under the head of the latter, they merge into each other very gradually. If we take the conventional boundary line just suggested, we shall find plenty of Lahndā characteristics to its east, gradually diminishing as we proceed, and at the same time many traces of Panjābī for a considerable distance to its west. The population is mixed, and has been mixed for centuries. The Sanskrit

APPENDIX I

writers had a very poor opinion of the Central and Western Punjab, although these tracts were not far from the holy Saraswati. The inhabitants are described as possessing no Brāhmans, living in petty villages, and governed by princes who supported themselves by internecine war. The population was castless, had no respect for the Vedas, and offered no sacrifices to the gods. They were flesh-eaters (a Piśācha characteristic) and hard drinkers, and their women were charged with polyandry like the Jats of the present day.

West of the Indus, up to the Afghan border, Lahndā under various names is spoken by Hindus, while the Pathān Musalmāns speak Pashto. Lahndā has two main dialects, one spoken north and the other south of the Salt Range. It has no literature. Its written character is, properly, the Lāndā also employed for Panjābī, but this has been nearly superseded by a modification of the Persian.

Sindhī is the language of Sind and the neighbourhood. It is closely connected with Lahndā, and, owing to its isolated position, it preserves many phonetic and flexional peculiarities which have disappeared elsewhere. There was, in former days, a Piśācha colony in Sind, and traces of their language are still to be found in Sindhī, which is, in other respects, a typical speech of the Outer Band of languages. It has no literature to speak of, and has received little cultivation of any kind. The population which employs it being largely Musalmān, its vocabulary borrows freely from Persian; and, since the country has come under British rule, an adaptation of the Persian character has been employed for writing it, although Lāndā is also used for personal memoranda and accounts. Sindhī has four main dialects—Siraikī, spoken in Upper Sind; Lāfī (the standard dialect) in Lārū or Lower Sind; Tharelī in the Thar or Desert; and Kachchhi in Cutch. The first approaches Lahndā, while Tharelī represents Sindhī merging into Mārwāri. Kachchhi is a mixture of Sindhī and Gujarāti, in which the former predominates.

South of Sindhī the Outer Band of Indo-Aryan vernaculars is interrupted by Gujarāṭī, the Intermediate language which has reached the seaboard. South of Gujarāṭī, extending from near Damān along the coast of the Arabian Sea to beyond Goa, we come to the great daughter of Māḥārāṣṭīrī Prākrit, the southern Indo-Aryan language, Marāṭhī. The Saurāṣṭīrī dialect of Māḥārāṣṭīrī once covered Gujarāt, but has been superseded by the Midland language. We find, however, traces of Saurāṣṭīrī not only in Gujarāṭī, but probably also right down the coast as

far as the modern Marāṭī extends. In the Bombay Presidency Marāṭī covers the north of the Deccan plateau and the strip of country between the Ghats and the Arabian Sea. It is also the language of Berār and of a good portion of the north-west of the Nizam’s Dominions. It stretches across the south of the Central Provinces (except a small portion of the extreme south, in which Telugu is the language), and, in a very corrupt form, occupies most of the State of Bastar. Here it merges into Oriyā through the Bhatri dialect of that language. It has to its north, in order from west to east, Gujarātī, Rājasthānī, Western Hindi, and Eastern Hindi. The first three are connected with the Midland, and Marāṭī does not merge into them. On the contrary, there is a sharp border-line between the two forms of speech. In the east it shows several points of agreement with the neighbouring Chhattisgarh dialect of Eastern Hindi, and it shades off gradually into Oriyā, both these languages being based on Prākrits of the Outer Band. Oriyā is its near neighbour to the east. On the south lie Dravidian languages, and it is bounded on the west by the Arabian Sea. In Marāṭī we first meet in general use a past participle, and its resulting past tense, of which the characteristic is the letter L. This extends through all the remaining languages of the Outer Band—Oriyā, Bengali, Bihārī, and Assamese. It is also found, in restricted use, in Gujarātī, alongside of the Midland form without the L, and is there one of the relics of the old Saurāṣṭrī Prākrit. This L-participle, therefore, not only covers the whole of Aryan East India, but reaches, through an almost unbroken chain of dialects all imperceptibly shading off into each other, to the Arabian Sea. This illustrates the intimate relationship which exists among all these forms of speech; and although Assamese is widely different from Marāṭī, and although a speaker of the one would be entirely unintelligible to a speaker of the other, a man could almost walk for 1,500 miles, from Dibrugarh to Goa, without being able to point (except, perhaps, in Bastar) to a single stage where he had passed from one language to another.

Marāthī has a copious literature of great popularity. The poets wrote in the true vernacular of the country, and employed a vocabulary mostly composed of honest tadbhavas. The result is that the language at the present day is rich in them; and though the scholars for whom the Marātha country is famous have in later times striven with some success to heighten the style of the language by the use of tatsamas, these parasites have not obtained the complete mastery over the

literary form of speech that they have in Bengal. The country was not invaded by the Musalmāns till a comparatively late period, and was ultimately successful in repelling the invasion, so that the number of words borrowed from and through Persian is small. As Mr. Beames says, Marāthi is one of those languages which may be called playful. It delights in all sorts of jingling formations, and has struck out a larger quantity of secondary and tertiary words, diminutives and the like, than any of the cognate languages.

Standard Marāthī is printed in the Deva-nāgarī character, but for purposes of writing a current hand, known as mādi or ’twisted,’ is in common use. It has three main dialects. The standard dialect, commonly called ‘Deśī Marāthī,’ is spoken in its greatest purity in the country round Poona. Sub-dialects of it are also found in the Northern and Central Konkan. In the Southern Konkan there is a distinct dialect known as ‘Konkāni.’ It differs so widely from standard Marāthī that some of its speakers claim for it the dignity of a separate language. To its south and west the Dravidian Kanarese is spoken, so that the Kanarese alphabet is generally employed for recording Konkani. Natives also employ the Deva-nāgarī character for the same purpose, while the Portuguese missionaries of Goa have introduced the use of the Roman character among their converts. The Marāthī of Berār and of the Central Provinces is the third dialect. It agrees more closely with the standard of Poona, the main differences being those of pronunciation. To these forms of speech may be added Halbī, which, however, can hardly be called a true dialect. It is spoken in the State of Bastar and the neighbourhood by Dravidian tribes who have attempted to abandon their aboriginal tongues. It is a mechanical mixture of bad Marāthī, bad Oriyā, and bad Chhattisgarhī, which varies in the proportions of its constituents from place to place. On the whole, Marāthī inflexions form its most prominent feature.

We now come to those languages of the Outer Band which are directly derived from the ancient Māgadhī Prākrit. They form the Eastern group of Indo-Aryan vernaculars, and are Bihārī, Orīyā, Bengali, and Assamese. Of these the first-named occupies the original home of the common parent, from which colonies have issued in three directions, to the south, the south-east, and the east, where each developed on its own lines into one of the other three.

Magadha, the land where the Buddha first preached, and in which the famous Asoka had his capital city, corresponds to

what we now call the Districts of Patna and Gayā. To its north, across the Ganges, lies the land of Tirhut, known in ancient times as Mithilā. To its west lies the Bhojpur country, comprising the west of modern Bihār and the east of the United Provinces. It may be taken as extending to the degree of longitude passing a few miles west of the city of Benares. To the south of Magadha lie the two plateaux of Chotā Nāgpur, the northern coinciding with the District of Hazāribāgh, and the southern with that of Rānchī. To its east lies Bengal proper. With the exception of Bengal, all these tracts together form the home of the present Bihārī language. It has three dialects, Maithīli, Magahī, and Bhojpuri, the last of which differs considerably from the two others. Maithīli, which is spoken in Tirhut, has a most complicated grammatical system, its verb changing its form, not only with regard to the subject, but also with regard to the object. It has a small literature dating from the fifteenth century, and, when written by Brāhmans, has a character of its own akin to that employed for Bengali. The people who speak it are among the most conservative in India, and rarely emigrate from their overcrowded fields to other parts of the country. Their character is reflected in their language, which abounds in archaic expressions. The original Aryan language of Nepāl before the Rājput invasion was an old form of Maithīli. Magahī, the language of the ancient Magadha, or South Bihār, is also spoken in the northern or Hazāribāgh plateau of Chotā Nāgpur, immediately to its south. It resembles Maithīli in the complexity of its verbal conjugation and in general character; but, owing to the long Musalmān domination of this part of India, it is as a rule more flexible and less conservative. The language of Magadha is looked upon by the inhabitants of other parts of India as typically boorish. Although directly descended from the language in which Buddhism was first preached, it has no literature and no traditions, and its speakers are as a whole poor and uneducated.

Far different is Bhojpurī. This dialect is spoken in the east of the United Provinces and in West Bihar. It has also spread to the southern, or Rānchī, plateau of Chotā Nāgpur, where, under a slightly altered form, it is called Nagpurīā. The Bhojpurī of the United Provinces differs somewhat from that of Bihar; but over the whole area the dialect has the same characteristics, being a flexible form of speech, adapted for current use, easy to learn, and not overencumbered by grammatical subtilties. Here again the language reflects the

national peculiarities. The Bhojpuris are as free from conservatism as the people of Tirhut are the reverse. They wander all over Northern India, and there is hardly a considerable town in which they do not possess a colony.

Apart from the peculiar character employed by the Tirhutiā Brāhmans, all the dialects of Bihārī are generally written in the current form of Deva-nāgarī known as ‘Kaithī.’

Oriyā is the language of Orissa and of the adjoining parts of Madras and the Central Provinces. It is spoken in an isolated part of India, has been but slightly affected by contact with other languages, and has changed little since the fourteenth century, at which period we find it in use in inscriptions. It has a considerable literature of some merit, and was formerly written by indenting marks with a stylus upon leaves of the talipot palm. On such a surface a straight indented line along the grain tends to cause a split; and this accounts for the characteristic of its peculiar alphabet, in which the long line familiar to readers of Deva-nāgarī is replaced by a series of curves.

Oriyā is a musical language, with a grammar which is simple but complete. It borrows very freely from Sanskrit, and the chief defect of its literary style is this overloading with tatsamas.

In its own home Bengali has a greater number of speakers than any other Indian language. In 1901, out of the forty-four and a half millions who returned this language as their vernacular, forty-four and a quarter millions inhabited the territories then subject to the Lieutenant-Governor of Bengal (with the connected States) and the Bengali Districts of Assam. The remaining quarter million were scattered throughout India, mainly finding employment as clerks or the like. Over the huge area in which it is a vernacular, Bengali is by no means uniform. Its main dialectal division is not, however, according to locality, but lies between the literary and the spoken language. If we except the language employed by the Musalman inhabitants of the eastern part of the Gangetic delta, the literary dialect is the same over the whole country. This is never used when speaking, except in formal addresses and the like. Even the most highly educated natives employ the colloquial dialect in their ordinary conversation. The literary form of the language differs from the colloquial not only in its highly Sanskritized vocabulary but in its grammar, in which the dead forms of three centuries ago are retained in a state of fictitious animation. This literary style dates from the revival of learning which took place in Calcutta, under English influences, at the commencement of the last century. Up to that time Bengal

had an indigenous poetical literature of its own, written in a purified form of the spoken vernacular. With the advent of the English there arose a demand for prose literature, and the task of supplying it fell into the hands of Sanskrit-ridden pandits. Anything more monstrous than this prose dialect, as it existed in the first half of the nineteenth century, it is difficult to conceive. Books were written, excellent in their subjects, eloquent in their thoughts, but in a language from which something like ninety per cent.¹ of the genuine Bengali vocabulary was excluded, and its place supplied by words borrowed from Sanskrit which the writers themselves could not pronounce. During the past fifty years there has been a movement, without much success, to reduce this absurd Sanskritization; but, still, at the present day many words current in literary Bengali are mere ideograms. The Bengali vocal organs are not adapted to the pronunciation of Sanskrit words, and so these words spell one thing, and, when read aloud, sound something quite different. Under such circumstances literary Bengali is divorced from the comprehension of every native to whom it has not been specially taught. It is this which is the official language of Government and of missionaries, and which (with few exceptions) is taught in the grammars written for European students. Bengalis themselves call their Sanskritized book-language ‘sādhu-bhāshā,’ i.e. the ’excellent speech’; but the adjective which they apply to anything approaching their true vernacular is the significant one of ‘sweet.’ It is this ‘sweet’ language which every one with a pen in his hand, be he European or native, endeavours to ignore. It is an instance of history repeating itself. In the old days the classical language was called sanskrit, ‘purified,’ but the epithet applied to the true vernacular Prākrit was amia, or ’nectar.’

The many dialects of spoken Bengali fall into three groups: the western or standard, the eastern, and the northern. Western Bengali is spoken in the country on both sides of the Hooghly and to the west. The centre of Eastern Bengali may be taken as the city of Dacca. It extends to the east into the Districts of Sylhet and Cāchār, and, southwards, to beyond Chittagong. The Bengali of Chittagong is very corrupt, and is quite unintelligible to an untravelled native of Calcutta. Farther inland, in the Chittagong Hill Tracts, there is a still more debased dialect called Chākmā, which is written in an alphabet akin to that of Burmese. Northern Bengali is

spoken north of the Ganges and at the lower end of the Assam valley. It is a dialect which, though closely connected with standard Bengali, really owes nothing to it. It is, by derivation, an intermediate speech between Bihārī and Assamese. In some respects it agrees with Oriyā rather than with the language of Calcutta.

Bengali and Assamese are written in very nearly the same alphabet, which is related to that employed by the Brāhmans of Tirhut. It is of the same stock as Deva-nāgarī, but has existed as an independent script since at least the eleventh century A.D.

The origin of Assamese has been described above. It is the language of the middle and upper parts of the Assam valley. It is more nearly related to colloquial than to literary Bengali; and its claim to be considered as an independent form of speech, and not as a dialect of that language, depends mainly upon the fact that it possesses an important literature. It has also several well-marked peculiarities of pronunciation. The literary style is happily free from the Sanskritisms which deface that of Bengali. The literature itself is of ancient date and is varied in its character, being particularly rich in historical works. Assamese has no real dialects, though it varies slightly from place to place. Māyāng, one of the languages spoken in the polyglot State of Manipur, may, however, be classed as a dialect of this language.

The Dravidian race is widely spread over India, but all the members of it do not speak Dravidian languages. In the north many of them have become completely Aryanized, and have adopted the language of their conquerors while they have retained their ethnic characteristics. Besides these, Dravidians are almost the only speakers of two other important families of speech, the Mundā and the Dravidian proper. Owing to the fact that these languages are nearly all spoken by persons possessing the same physical type, many scholars have suggested a connexion between the two groups of speech, but a detailed inquiry carried out by the Linguistic Survey of India has shown that there is no foundation for such a theory. Whether we consider the phonetic systems, the methods of inflexion, or the vocabularies, the Dravidian have no connexion with the Mundā languages. They differ in their pronunciation, in their modes of indicating gender, in their declensions of nouns, in their method of indicating the relationship of a verb to its objects, in their numeral systems, in their principles of

conjugation, in their methods of indicating the negative, and in their vocabularies. The few points in which they agree are points which are common to many languages scattered all over the world.

Leaving, therefore, the fact of the Dravidian race speaking two different families of languages to be discussed by ethnologists, we proceed to consider those forms of speech which are called ‘Dravidian’ by philologists. Most of these are spoken in Southern India or in the hills of Central India. Two of them have found their way into Chotā Nāgpur and the Santāl Parganas, where they exist side by side with Mundā dialects; and one, Brāhūī, has its home far to the north-west, in Baluchistān. The last was not known to Sanskrit writers, who were familiar with two great languages spoken in their time all over Southern India: namely, the Andhra-bhāṣā and the Drāviṇa-bhāṣā, the former corresponding to the modern Telugu, and the latter to the rest. This old division agrees with the classification of the modern vernaculars, which is as follows:—

| A. Dravidia group: |

The following general account of the main characteristics of the Dravidian forms of speech is taken, with one or two verbal alterations and omissions, from the Manual of the Administration of the Madras Presidency:

In the Dravidian languages all nouns denoting inanimate substances and irrational beings are of the neuter gender. The distinction of male and female appears only in the pronoun of the third person, in adjectives formed by suffixing the pronominal terminations, and in the third person of the verb. In all other cases the distinction of gender is marked by separate words signifying “male” and “female.” Dravidian nouns are inflected, not by means of case terminations, but by means of suffixed postpositions and separable particles. Dravidian neuter nouns are rarely pluralized. Dravidian languages use postpositions instead of prepositions. Dravidian adjectives are incapable of declension. It is characteristic of these languages, in contradistinction to Indo-European, that, wherever practicable, they use as adjectives the relative participles of verbs, in preference to nouns of quality or adjectives properly so called. A peculiarity of the Dravidian (and also of the Mundā) dialects is the existence of two pronouns of the first person plural, one inclusive of the person addressed, the other exclusive. The Dravidian languages have no passive voice, this being expressed by verbs signifying “to suffer,” &c. The Dravidian languages, unlike the Indo-European, prefer the use of continuative participles to conjunctions. The Dravidian verbal system possesses a negative as well as an affirmative voice. It is a marked peculiarity of the Dravidian languages that they make use of relative participial nouns instead of phrases introduced by relative pronouns. These participles are formed from the various participles of the verb by the addition of a formative suffix. Thus, “the person who came” is in Tamil literally “the who-came.”

Tamil, or Arava, covers the whole of Southern India up to Mysore and the Ghāts on the west, and reaches northwards as far as the town of Madras and beyond on the east. It is also the vernacular of the northern part of Ceylon², and has been widely spread over Further India by emigrant coolies. As domestic servants its speakers are found all over India. It is the oldest, richest, and most highly organized of the Dravidian languages: plentiful in vocabulary, and cultivated from a remote period. It has a copious literature, which is couched in a somewhat artificial dialect known as ‘Shen’ (i.e. ‘perfect’), in contrast with the colloquial form of speech, which is called ‘Kodum’ or ‘Codoon’ (i.e. ‘rude’). Only a few insignificant dialects of the spoken language have been recorded. The name ‘Tamil’ and the word ‘Draviṇa’ are both corruptions of the same original, ‘Drāmiṇa.’ The language has an alphabet of its own.

Malayalam is a modern offshoot from Tamil, dating from the ninth century A.D. It is the language of the Malabar coast, and has one dialect, Yerava, spoken in Coorg. Its most noteworthy features are that, except among certain tribes, it has dropped all the personal terminations of verbs, and that the words which it has borrowed from Sanskrit are particularly numerous. It has a large literature, and employs the old Grantha character used in Southern India for Sanskrit writings.

Kanarese is the language of Mysore and of the neighbouring portion of the Ghāt country, including the southern corner of the Bombay Presidency. It, also, has an ancient literature, written in an alphabet closely connected with that employed for Telugu. It has two petty dialects, Badaga and Kurumba, both of which are spoken in the Nilgiris. Kodagu, the language of Kodagu, Coorg, is also considered by some to be a dialect of Kanarese. It lies midway between it and Tulu, the language of a portion of the South Kanara District of Madras. Toda and Kota are petty forms of speech spoken by small tribes on the Nilgiris.

Kurukh, or Orāon, is the vernacular of a Dravidian tribe in Chotā Nāgpur and the adjoining portions of the Central Provinces. It is more closely connected with ancient Tamil and with ancient Kanarese than with any other of the great Dravidian languages. The people themselves say that they and the Maler actually did come to their present seats from the Kanara country. Malto is the language of these Maler, a tribe nearly related to the Orāons, and now settled, still farther north, near Rājmahāl on the bank of the Ganges. Neither of these two languages has any literature or any alphabet. The Roman alphabet is usually employed for recording them.

The Gond language is spoken in the hill country of Central India. Many of the Gonds have abandoned their own dialects and have taken to Aryan forms of speech. The true Gond is intermediate between the Dravīda and Andhra tongues, and has numerous dialects. It is unwritten, and has no literature.

Telugu is the only important Andhra language. It is the principal form of speech in the eastern part of the Indian Peninsula, from the town of Madras to near Orissa. It is also spoken in the east of the Nizam’s dominions and in the extreme south of the Central Provinces, extending into Berar. It has an extensive literature, written in a character of its own, akin to Deva-nāgarī, which, like Oriyā, owes its numerous curves to the fact that it has been written on palm-leaves.

Kandh, or Kuī, is spoken by the Khonds of the Orissa Hills. Kolāmī and other petty dialects of distant Berār are quite uncultivated.

Brāhūī, also an uncultivated language, is heard in the central highlands of Baluchistan. Owing to its isolated position, it has developed on lines of its own; but, although its speakers show none of the Dravidian ethnic characteristics, it is undoubtedly a Dravidian language. Ethnologists differ as to whether the speakers of Dravidian languages entered India from the north-west, or from the hypothetical Lemurian continent, now under the Indian Ocean, in the south. If they came from the north-west, we must look upon the Brāhūis as the rear-guard; but if from the south, they must be considered as the advance-guard of the Dravidian immigration. Under any circumstances it is possible that the Brāhūis alone retain the true Dravidian ethnic type, which has been lost in India proper by admixture with other aboriginal nationalities such as the Mundās. This is suggested by the linguistic circumstances, and is worthy of investigation.

The Mundā languages are often called ‘Kolarian,’ a name which is founded on a false theory, and which is, moreover, misleading. The name ‘Mundā’ was first given to this family of speech by the late Professor Max Müller long before ‘Kolarian’ was invented. These languages are among those which have been longest spoken in India, and may, with great probability, claim to be aboriginal. It is of importance to note that there exists a common element in them, on the one hand; and in the Mon-Khmer languages of Further India, in the dialects of certain wild tribes of Malacca and Australonesia, and in Nicobarese, on the other, although the two sets of speech are not otherwise connected. This is best explained by the supposition that a common language was once spoken over both Further India and a great part of India proper, and that in the latter it is represented at the present day by the Mundā languages, while in Further India, Malacca, Australonesia, and the Nicobars³ it was overwhelmed by an invasion of other languages (much as was the case with the original Piśacha language of Kashmir), and there now shows only sporadic, though convincing, traces of its former general use.

The Mundā languages are agglutinative, and preserve this characteristic in a very complete manner. Suffix is piled upon suffix, and helped out by infix, till we obtain words which have the meaning of a whole sentence. For instance, the word dal means ‘strike,’ and from it we form the word da-pa-l-ocho-akan-tahen-tae-tiñ-a-e, which signifies ‘he, who belongs to him who belongs to me, will continue letting himself be caused to fight.’ Not only may we, but we must employ this posy of speech, if, for instance, my slave’s son was too often getting himself entangled in affrays. As compared with Dravidian languages, Mundā languages have a series of semi-consonants which correspond to the so-called ‘abrupt’ tone of the languages of Further India. The distinction of gender is between animate and inanimate nouns, and not between rational and irrational ones. The noun has three numbers—a singular, a dual, and a plural; and the cases of the direct and indirect object are indicated by suffixes added to the verb, while the noun remains unchanged. The numerals are counted by twenties and not by tens. As in Dravidian, the pronoun of the first person plural has two forms, one including, and the other excluding, the person addressed, but in other respects the pronouns are altogether different. There is no agreement whatever between the conjugations of the Mundā and of the Dravidian verb. The latter is simple, while the former exhibits an almost bewildering maze of participial forms, which in every case are converted into tenses by the addition of the letter a. Finally, the Mundā languages do not possess anything corresponding to the Dravidian system of negative conjugation.

The principal home of the Mundā languages (the race is much more widely spread) is Chotā Nāgpur. Speakers are further found in the adjoining Districts of Bengal, Orissa, Madras, and the Central Provinces, with an outlying colony far to the west in the Mahādeo hills north of Berār. The following is a list of these forms of speech:—

Of these, Kherwāri is much the most important. It has several dialects, which are often wrongly considered to be distinct languages. They are Santālī or Hār, Mundāri, Bhumij, Birhār, Kōdā, Ho, Tūri, Asurī, Agariā, and Korwā. Of these, Santālī and Mundāri have received much attention from scholars, and we have excellent grammars of them, as well as a dictionary of the former. Ho is the dialect of the Larkā, or ‘fighting,’ Kols of Singhbhūm, while the others are spoken by petty forest tribes. The home of Santālī is the Santāl Parganas, but it is also found much farther south, down the western border of Bengal proper into Northern Orissa. The rest are all spoken in Chotā Nāgpur and in the neighbouring hill tracts of Orissa and the Central Provinces.

Kürkū is the Mundā language of the Mahādeo Hills. With Khaṭīa and Juāng it forms a linguistic sub-group, but is more nearly related to Kherwāri than are the other two. It, also, has received some study, and we have an excellent grammar of it. Khaṭīa is found in the south-west corner of Rānchī and in the adjoining States of Jashpur and Gāngpur. The tribe extends much farther south, but they have as a rule exchanged their own language either for the Dravidian Kurukh or for some broken Aryan patois. The language is dying out, and is nowhere spoken in its original purity. It has borrowed freely from neighbouring forms of speech, and has been compared to a palimpsest, the original writing of which can only be deciphered with some difficulty. Juāng resembles Khaṭīa. It is the language of a small wild tribe in the Orissa Hills. From the leaf-garments of its speakers it is sometimes called ‘Patuī.’ Savara and Gadaba are two languages spoken in Madras territory close to the Orissa border. Very little is known concerning them; but it is plain that they are much mixed with the Telugu spoken round about them, and they may probably be grouped as akin to Khaṭīa and Juāng. The Savaras are an ancient and widely spread tribe, who were known to the Indo-Aryans in Vedic times, and are mentioned by Pliny and Ptolemy. Only a few of them still adhere to their own language.

None of the Mundā languages have any proper written character or any literature. The Roman character is generally employed for recording them.

The languages of Further India, together with those spoken in Tibet, are usually grouped under the general name of ‘Indo-Chinese’⁴ which includes two distinct families, the Mon-Khmer and the Tibeto-Chinese. The original home of all these people seems to have been North-western China, between the upper courses of the Yang-tse-kiang and the Ho-ang-ho, and from here they spread out in all directions. So far as British India is concerned, they followed river valleys in their migrations, down the Chindwin, the Irrawaddy, and the Salween into Burma, down the Brahmaputra into Assam, and up the Brahmaputra into Tibet. From Tibet they occupied the Himalayas, and are now found in Nepal and in other mountainous tracts lying south of the main watershed. Three successive waves of completed migration can be traced. First, there was, in prehistoric times, a Mon-Khmer invasion into Further India and Assam. Secondly, there was the first Tibeto-Chinese invasion, that of the Tibeto-Burmans, into the same localities and into Tibet, the period of which is also unknown. Thirdly, there was the second Tibeto-Chinese invasion, that of the Tai branch of the Siamese-Chinese into Eastern Burma, which took place in force about the sixth century A.D. Finally, another Tibeto-Burman invasion, that of the Kachins, was actually in progress when it was stopped by the British conquest of Upper Burma. The later invaders drove the first to the sea-board or into the hills overlooking the river valleys; and thus we find the earliest immigrants to India, the Mon-Khmers, confined at the present day to the coast country of Pegu and a few mountain tracts in Assam and Burma, while the Tais, who found most room for expansion in the direction of Siam, have driven the Mon-Khmers of that country to the sea-coast also.

All the Indo-Chinese languages are monosyllabic. Each word consists of one syllable, and refuses to be classed under any of the well-known categories of noun, verb, and particle. It expresses an indefinite idea, which may be employed to connote any part of speech, according to its position in the sentence and its relation to its neighbours. The words being monosyllables, the necessary paucity of different sounds is eked out by tones, each sound being raised or lowered in pitch, shortened or prolonged, according to the idea which it is intended to convey. For instance, the Shan monosyllable kau means ‘I,’ ‘be old,’ ’nine,’ ‘a lock of hair,’ ‘indifference to an evil spirit,’ ‘an owl,’ ‘a butea-tree,’ ‘complaining of any thing,’ ’the shin,’ ’the balsam plant,’ or ‘a mill,’ according to the tone with which it is pronounced. The number of tones differs in various languages. Shan has fifteen, while Western Tibetan is said to have only one. The most characteristic of these languages, Chinese and Siamese, belong to what is known as the isolating class—i.e. every monosyllable has a distinct definite meaning of its own, and complex ideas are expressed by compounding two or more together. For instance, ‘he went’ would be indicated by three words, one meaning ‘he,’ another connoting the idea of ‘going,’ and a third connoting the idea of ‘completion.’ Others belong to what is known as the agglutinating class, in which certain words are now only used as suffixes to indicate relationship of time or space, and cannot be employed independently with meanings of their own. It is as if the word ‘completion’ in ‘he-going-completion’ had lost

its original meaning, and was now only used as a sign to indicate that the idea connoted by some other word performing the function of a verb was also the idea of a completed action.

We have already mentioned the fact that the Mon-Khmer languages agree with some Malacca dialects and with Nicobarese in having at their base another non-related language which is connected with Mundā forms of speech, and which must have been the aboriginal language of those tracts of Further India which were conquered by the Mon-Khmers. The Mon-Khmer languages are numerous in Indo-China. In British India they are only four in number. The most important is Khāsi, spoken in the hill country south of the Central Assam valley, where it has survived as an island amid a sea of Tibeto-Burman speech. It has been given a literature by the missionaries who work among its speakers; and this language, which a century ago was rude, uncultured, and unwritten, is now one of the Indian vernaculars recognized in the examination halls of the Calcutta University. It is written in the Roman character and has 177,827 speakers. The other important language is the Mon or Talaing of Pegu and the coast districts round the Gulf of Martaban (174,510 speakers). Palaung (67,756) and Wa (7,667) are two smaller dialects spoken in the eastern hills of Upper Burma.

The Tibeto-Burman branch of the Tibeto-Chinese languages is very widely spread. It includes two great languages, Tibetan and Burmese, each of which has an alphabet of its own akin to Deva-nāgarī, as well as an extensive literature. Tibetan is one of several dialects grouped under the general name of ‘Bhotīā,’ from Bhot, the Indian name of Tibet. Besides the Bhotīā of Tibet or Tibetan, there are the Bhotīā of Bāltīstān or Bātīti, that of Lādākh or Lādākhī, that of Sikkim or Denjongke, that of Bhutān or Lho-ke, and so on. Connected with Bhotīā, but not dialects of it, are a number of Himālayan languages of which the most noteworthy are Newāri (the main language of Newār, i.e. Nepāl), Rong or Lepcha (of Sikkim), Mangar, and Murmi. Most of these are really Nepāl languages, whose speakers (many of them soldiers in our Gurkha regiments) are temporary visitors to British India. This group is called the ‘Non-pronominalized Himālayan languages,’ to distinguish it from another, of which Kanāwari, Limbū, and the so-called Kirāntī forms of speech are the most important members, and which Hodgson classed as the ‘Pronominalized Himālayan languages.’ Although this latter group is in the main Tibeto-Burman in character, it also shows manifest traces of an older substratum having striking points of resemblance to the Mundā tongues. There are the same distinctions between things animate and inanimate, the same system of counting in twenties, the same occurrence of a dual number, and of a double set of plural forms for the first personal pronoun, and the same tendency to conjugate a verb by means of pronominal suffixes. All this cannot be mere coincidence. It inevitably leads to the conclusion that these Himālayan tracts were once inhabited by tribes speaking a language connected with those now in use among the Mundās, who have left their stamp on the dialects spoken at the present day. We have already seen how a Mundā basis also exists in the Mon-Khmer languages, which has been traced into Malacca, Australonesia, and even Australia; and this line of Himālayan dialects offers an important clue to ethnological inquirers. West of Bhutān we come across another Tibeto-Burman group, spoken by wild tribes of the hills to the north of the Assam valley. These are Aka, Daflā, Abor-Miri, and Mishmi. In the lower Assam valley itself and the country to its south (omitting the Khāsi Hills) we have the Bodo group, spoken by 596,411 people, of which the principal languages are Bāā or Mech, the tongue of scattered tribes in the valley, Gāo of the Gāo Hills, and Tipurā or Mrung of Hill Tippera. Then we have the Nāgā languages of Central and Eastern Assam. The most important of these is Mikir of the Mikir Hills in the valley itself. To the south and south-east there are the Nāgā Hills, inhabited by many fierce tribes whom we are slowly winning to civilization, and each possessing a language of its own. Such are Angāmi, Semā, Ao, Lhotā, and Namsangā, with fourteen or fifteen others. None of them, of course, has any literature, and of many of them little but the names and a few words are known. The Angāmi Nāgās are those with whom we have fought most, and with whom we are best acquainted. East of Assam, in the confused mountainous country which forms the north of Upper Burma, are a number of cognate dialects grouped together under the general name of Kachin or Singpho. These wild Kachins were migrating into Burma itself, and had already penetrated far into the Shan States, when we annexed that country.

South of the Nāgā Hills lies the State of Manipur, and here we first meet the group of languages known as Kuki-Chin. Meithei, the official language of the State, is the only one of them which possesses an alphabet and a literature. Owing to the existence of the latter its development has been retarded,

so that it is in an older stage than the rest. The others are scattered in colonies over Manipur and Cachar, and extend south, through the hill country, as far as the Sandoway District of Burma. Since they occupied this latter area, there has been a constant tendency to expand northwards. On the west they were barred by the sea, and on the south and east by the stable government of Burma. Thus wave after wave has been driven to the north by those who were behind. The Kuki-Chins of Manipur and Cachar once occupied the hills immediately to the south, and these are now held by the Lushais, who were originally pushed forward from the south-east and drove them on. This progress has been arrested by our conversion of Cachar into settled territory. There are more than thirty Kuki-Chin languages, some with several dialects. The most important, both politically and in the numbers that speak them, are Lai in the Chin Hills, and Lushei or Dulien in the Lushai Hills. The Kuki-Chin are the most typical of all the Tibeto-Burman languages. They do not possess a real verb, the conception being expressed with the aid of a verbal noun. When a speaker of Lushei, for instance, wishes to say ‘I go,’ he says ‘my going’; and for ‘I went,’ ‘my going-completion.’

Passing over a number of hybrid dialects we come to Burmese, which is the predominant language, even where others are spoken, all over Upper and Lower Burma, except in the Chin Hills, the Shan States, and the Kachin country north of Bhamo. It, and the related Mrü, are the vernaculars of 7,498,794 people. It has many local dialects, but, with one or two exceptions, these are little known. The most important dialect is Arakanese, which branched off from the main stem at an early date, and has developed on independent lines. Burmese has a considerable literature, of which the poetry is written in a special and difficult dialect; and a written character of its own, derived from the ancient square Pāli, but abounding in curved lines, and connected, through the Pāli, with Devanāgarī. The development of the spoken language has proceeded more rapidly than that of the written language, so that words are nowadays seldom pronounced as they are spelt.

The only important Tai language of British India is Shan, spoken in the south-east of Upper Burma, and closely allied to Siamese. A Tai tribe called the Ahoms made themselves masters of Assam in the year 1228 A.D. They were followed by other Shan colonies, which still survive and speak their own dialects. The most important is Khāmtī. Ahom has been dead for centuries, though its literature still survives and can be interpreted by a few priests of the old religion. The Ahoms were pagans, but the rest of the Shans, like the Burmese, are Buddhists. Shan has a voluminous literature, and a written character based on that of Burmese.

The Karen tribe is principally scattered over Lower Burma, though its members are also found in the Shan Hills. Their language likewise belongs to the Siamese-Chinese branch of the Tibeto-Chinese family. The generally accepted theory regarding this form of speech is that it is connected with Chinese though not descended from it, while the people are pre-Chinese.

The remaining vernaculars of India proper are unimportant. The Selungs, a tribe of sea-gipsies inhabiting the Mergui Archipelago, speak a language akin to Malay. Such, also, is Nicobarese, which has, however, like Mon-Khmer, a substratum of Mundā. Some scholars class this as a Mon-Khmer language with Malay corruptions. Two languages have not yet been classed by philologists. These are Andamanese and Burushaski. The former is really a group of languages which are agglutinating, make free use of prefix, infix, and suffix, and are adapted only to the expression of the more simple ideas. Burushaski is spoken in the extreme north-west of India on the borders of Turkistan, by the inhabitants of Hunza-Nagar. No one has hitherto succeeded in tracing a connexion between it and any other known form of speech. It has an elaborate grammar, and its most characteristic feature is the frequent use which is made of pronominal prefixes, so as sometimes to alter greatly the appearance of a word. The country in which it is spoken did not fall within the operations of the Census of 1901, and hence no speakers of it were recorded.

The so-called ‘Gipsy’ languages have nothing to do with European Romani. They are a number of dialects spoken by wandering tribes, often of very bad reputation. Some are mere thieves’ jargons, others are hybrids developed in journeys from place to place, and some are real dialects of well-known languages.

In Aden we find Arabic and Somali spoken. The former belongs to the Semitic and the latter to the Hamitic family. They hardly fall within the lines of the present inquiry.

G. A. GRIERSON.

I. LIST OF THE INDO-CHINESE LANGUAGES SPOKEN IN BRITISH INDIA AND NEPAL

II. List of Minor Languages spoken in British India

APPENDIX II

Selected Authorities

GENERAL. Nearly all the languages mentioned in this chapter are fully described in the volumes of the Linguistic Survey of India. For each dialect there is a grammar and a selection of specimens. A general account of the languages of India, more full than that contained in the foregoing pages, will be found in chapter vii of the Report on the Census of India, 1901.

ERANIAN LANGUAGES GENERALLY. DARMESTETER, Études Iraniennes, Paris, 1883. GEIGER and KUHN, Grundriss der Iranischen Philologie, Strassburg, 1895–1904.

BALOCH. DAMES, A Text Book of the Balochi Language, Lahore, 1891.

PASHTO. BELLEW, Dictionary, London, 1867; Grammar, London, 1867. RAVERTY, Grammar, London, 1855, 1860, 1867; Dictionary, London, 1860; Manual, London, 1880. TRUMPP, Grammar, London, 1873. DARMESTETER, Chants populaires des Afghans, Paris, 1888-90. ROOS KEPPEL, Grammar, London, 1901. GEIGER and KUHN, as above. LORIMER, Grammar and Vocabulary of Waziri Pashto, Calcutta, 1902.

ORMURĪ. GHULAM MUHAMMAD KHAN, Qawâid-e Bargistâ (Grammar in Urdû), Lahore, 1886.

GHALCHAH LANGUAGES. SHAW, Journal of the Asiatic Society of Bengal, Vol. XLV (1876), Part I, pp. 139 ff. (Wakhi and Sariqoli); Vol. XLVI (1877), Part I, pp. 97 ff. (Shighni). GEIGER and KUHN, as above.

PIŚACHA LANGUAGES. LEITNER, Dardistan, Lahore, 1877. BIDDULPH, Tribes of the Hindoo Koosh, Calcutta, 1880. O’BRIEN, Grammar and Vocabulary of the Khowar Dialect, Lahore, 1895. DAVIDSON, Notes on the Bashgali (Kafir) Language, Calcutta, 1901. GRIERSON, The Pisacha Languages of Northwestern India, London, 1906.

INDO-ARYAN LANGUAGES GENERALLY. BEAMES, Comparative Grammar of the Modern Aryan Languages of India, London, 1872-9. HOERNLE, A Grammar of the Eastern Hindi, compared with the other Gaussian Languages, London, 1880. BHANDARKAR, The Phonology of the Vernaculars of Northern India in Journal of the Royal Asiatic Society (Bombay Branch), Vol. XVII, ii. 99-182. GRIERSON, On the Phonology of the Modern Indo-Aryan Vernaculars in Zeitschrift der Deutschen Morgenländischen Gesellschaft, Vols. XLIX, L (1895-6), pp. 393 ff., 1 ff.

WESTERN HINDI (including HINDOSTANÍ). GARCIN DE TASSY, HÍSTOIRE de la Littérature Hindouie et Hindustanie, 2nd ed., Paris, 1870-1. PLATTS, Grammar of the Hindustani or Urdu Language, London, 1874; Dictionary of Urdu, Classical Hindi, and English, London, 1884. BATE, Dictionary of the Hindee Language, Benares and London, 1875. FALLON, New Hindustani-English Dictionary, Benares and London, 1879; New English-Hindustani Dictionary, Benares and London, 1883. LYALL (C. J.), Sketch of the Hindustani Language, Edinburgh, 1880; Hindustani Literature, in Article ‘Hindustani’ in Encyclopaedia Britannica, 9th ed. CROOKE, Rural and Agricultural Glossary for the N.W. Provinces and Oudh, Calcutta, 1888. GRIERSON, The Modern Vernacular Literature of Hindustan, Calcutta, 1889. PINCOTT, Hindi Manual, London, 1890. MUHAMMAD ASHRAF ‘ALÍ, Muñalíhat-e Urdu (On Urdu idioms. In Urdu), Lucknow, 1890. KELLOGG, Grammar of the Hindi Language, 2nd ed., London, 1893. GREAVES, Grammar of Modern Hindi, Benares, 1896. RANKING, English-Hindustani Dictionary, Calcutta and London, 1905.

RAJASTHĀNĪ. ROBSON, A Selection of Khyâls or Mârwâri Plays, with an Introduction and Glossary, Beawr, 1866. KELLOGG, as above. MACALISTER, Specimens of the Dialects spoken in the State of Jeypore, Allahabad, 1898. RAM KARN SARMÃ, Mârwâri Vyâkarana (Mârwâri Grammar in that language), Jodhpur, (?) 1901. GRIERSON, Note on the Principal Rajasthani Dialects in Journal of the Royal Asiatic Society, 1901.

KHAS or NAIPALI. TURNBULL, Grammar and Vocabulary, Darjeeling, 1887.

GUJARATI. TISDALL, Simplified Grammar, London, 1892. TAYLOR, Student’s Grammar, Surat, 1893; London, 1895. BELSARE, Dictionary, Ahmedabad, 1895. KARBHARI, Dictionary, Ahmedabad, 1899. THOMPSON, Rudiments of the Bhili Language, Ahmedabad, 1895.

PANJABI. JANVIER, Dictionary, Ludhiana, 1854. TISDALL, Simplified Grammar, London, 1889. MAYA SINGH, Dictionary, Lahore, 1895. NEWTON, Grammar, Ludhiana, 1898. GRAHAME BAILEY, Grammar of Wazirabad Dialect, Lahore, 1904. A Grammar of the Dogri dialect is given by DREW in Jummoo and Kashmir Territories, London, 1875.

EASTERN HINDĪ. KELLOGG, as above. GARCIN DE TASSY, as above. GRIERSON, Modern Vernacular Literature, &c., as above. GROWSE, The Râmâyana of Tulsi Dâs (Translation), 1st ed., Allahabad, 1880-1. GREAVES, Grammar of the Râmâyan, Benares, 1895. HIRALAL, Grammar of the Dialect of Chhattisgarh, Calcutta, 1890 (reprinted from Journal of the Asiatic Society of Bengal, Vol. LIX (1890), Part I).

KĀSHMĪRĪ. ELMSLIE, Vocabulary, K.-English and Eng.-K., London, 1872. WADE, Grammar, London, 1888. GRIERSON, Essays on K. Grammar, London and Calcutta, 1899.

KOHISTANI. BIDDULPH, as above, under Pisacha.

LAHNDĀ. TISDALL’S Simplified Panjab Grammar; also deals, in an Appendix, with Lahndā. O’BRIEN, Glossary of the Multani Language, 1st ed., Lahore, 1881; 2nd ed., revised by WILSON and HARI KISHEN KAUL, Lahore, 1903. JUKES, Dictionary of the Jatki or Western Panjab Language, Lahore and London, 1900. WILSON, Grammar and Dictionary of Western Panjab, as spoken in the Shahpur District, Lahore, 1899.

SINDHI. STACK, Grammar and Dictionary, both Bombay, 1849. TRUMPP, Grammar, London and Leipzig, 1872. SHIRT, UDHARAM THAVURDAS, and MIRZA, Dictionary, Karâchi, 1879.

MARÂTHI. NAVALKAR, Grammar, 3rd ed., Bombay, 1894. MOLESWORTH and G. and T. CANDY, Dictionary (M.-English), 2nd ed., Bombay, 1857. PADMANJI, Dictionary (English-M.), 3rd ed., Bombay, 1889. MAFFEI, Könkanji Grammar, Mangalore, 1882; English-K. Dictionary and K.-Eng. Dictionary, both Mangalore, 1883. DALGADO, Könkanji-Portuguese Dictionary, Bombay, 1893. DIAS, English-K. Dictionary, Bombay, 1899. MURRAY MITCHELL, The Chief Marathi Poets in Transactions of the Congress of Orientalists, London, 1892, Vol. I, pp. 282 ff.

BIHÂRĪ. GRIERSON, Introduction to the Maithili Language of North Bihâr, Calcutta, 1881-2; 2nd ed., in the Press; Seven Grammars of the Dialects and Sub-dialects of the Bihârî Language, Calcutta, 1883-7. Modern Vernacular Literature, &c., as under Western Hindi.

ORIYA. SUTTON, Dictionary, Cuttack, 1841. BROOKE, Dictionary, Cuttack, 1874. HALLAM, Grammar, Calcutta, 1874. MALTBY, Practical Handbook, Calcutta, 1874. BROWNE, Primer in Roman Character, London, 1882. MONMOHAN CHAKRAVARTI, Notes on the Language and Literature of Orissa in Journal of the Asiatic Society of Bengal, Vol. XLI (1872), Part I, p. 120; XLII (1873), Part I, p. 59; XLIII (1874), Part I, p. 332.

BENGALI. HAUGHTON, Dictionary, London, 1833. MENDIES, Dictionaries, English-B., 2nd ed., Calcutta, 1872; B.-Eng., 3rd ed., ib., 1876. (Numerous Dictionaries have since issued. Most are based on Mendies.) NAGENDRANATH BASU, Visva-kosha (a Bengali Encyclopaedia), Calcutta, in course of publication. BROWNE, Primer in Roman Character, London, 1881. NICHOLI, Grammar, London, 1885; enlarged edition, under name of Bengali Manual, London, 1894. BEAMES, Grammar, Oxford, 1894. DUTT, The Literature of Bengal, 1st ed. (Pseudonymous), by ‘Arey Dae,’ Calcutta, 1877; 2nd ed., Calcutta and London, 1895 (under author’s name). HARAPRASAD SÁSTRÍ, Vernacular Literature of Bengal before the Introduction of English Education, Calcutta, no date. DINEÓ CHANDRA SEN, Banga-bhāshā o Sāhitya (History of Bengali Literature in Bengali), 2nd ed., Calcutta, no date (preface dated 1901). GRIERSON, Grammar of Rangpur Dialect in Journal of the Asiatic Society of Bengal, Vol. XLVI (1877), Part I, pp. 186 ff. PARGITER, Notes on Chittagong Dialect (Grammar, &c.), in ib., Vol. LV (1886), Part I, pp. 66 ff.

ASSAMESE. BROWN, Grammar (1st ed., 1848), revised edition by MOORE, Nowgong, Assam, 1893. BRONSON, Dictionary, Sibsāgar, 1867. HEM CHANDRA BARUA, Dictionary (edited by GURDON and HEM CHANDRA GOSAIN), Shillong, 1900. ANANDA RAM DHEKIAL PHUKAN, A few Remarks on the Assamese Language, Sibsāgar, 1855 (contains an account of Assamese Literature. Partly reprinted in Indian Antiquary, Vol. XXV (1896), pp. 57 ff.).

DRAVIDIAN LANGUAGES GENERALLY. CALDWELL, A Comparative Grammar of the Dravidian and South-Indian Family of Languages, 2nd ed., London, 1875. HODGSON, Essays relating to Indian Subjects, London, 1880.

TAMIL. BESCHIUS, Grammatica Latino-Tamulica, translated by MAHON, Madras, 1848. A Translation of BESCHIUS’s Grammar of High Tamil, published at Tranquebar, 1876. POPE, First Lessons in Tamil, 7th ed., Oxford, 1904, and numerous other works. VINSON, Manuel de la Langue Tamoule, Paris, 1903. PERCIVAL, Tamil-English Dictionary, New ed., Madras, 1900.

MALAYALAM. GUNDERT, Grammar (in Malayalam), Mangalore, 1868. FROHNMEYER, Grammar, Mangalore, 1889. ANON, Malayalam-Latin-English Dictionary. By a Discalced Carmelite Missionary, Verapoly, 1891.

KANARESE. KITTEL, Dictionary, Mangalore, 1894; Grammar, Mangalore, 1903.

TULU. BRIGEL, Grammar, Mangalore, 1872. MÄNNER, Dictionaries, Mangalore, 1886–8.

KURUKH. HAHN, Grammar, Calcutta, 1900; Dictionary, Calcutta, 1903.

MALTO. DROESE, Introduction to the Malto Language, Agra, 1884.

GOND. WILLIAMSON, Grammar, London, 1890.

TELUGU. ARDEN, Grammar, Madras, 1873. MORRIS, Grammar, London, 1890. PERCIVAL, Dictionary, 2nd ed., Madras, 1867. BROWN, Dictionary, new ed., Madras, 1895.

KANDH. LINGUM LETCHMAJEE, Grammar, 2nd ed., Calcutta, 1902. SMITH, Practical Handbook, Cuttack, 1876.

BRAHUI. TRUMPP, Grammar (Munich, 1880), translated by DUKA in Journal of the Royal Asiatic Society, Vol. XIX (1887), pp. 59 ff.

MUNDA LANGUAGES GENERALLY. HODGSON, Essays relating to Indian Subjects, London, 1880. MÜLLER, Letter to Chevalier Bunsen on the Classification of the Turanian Languages, London, 1854. KUHN, Beiträge zur Sprachenkunde Hinterindiens, Munich, 1889. THOMSEN, Bemærkninger om de Khervariske (Kolariske) Sprogs Stilling, Copenhagen, 1892. KONOW, Mundas and Dravidas in Indian Antiquary, Vol. XXXIII (1904), pp. 121 ff.

SANTALI. SKREFSRUD, Grammar, Benares, 1873. E. HEUMAN, Grammatisk Studie över Santal-språket, Copenhagen, 1892. COLE, Santali Primer, Pokhuria, 1896. MARTIN, Eng.-S. Vocabulary, Benares, 1898. CAMPBELL, S.-Eng. Dictionary, Pokhuria, 1899.

MUNDARI. NOTTROTT, Grammar (in German), Gütersloh, 1882. DE SMET, Grammar, Calcutta, 1891. HOFFMANN, Grammar, Calcutta, 1903.

HO. BHIM RAM SULANKHI, Hokajī (Grammar in Hindi), Benares, 1886.

KÜRKÜ. DRAKE, Grammar, Calcutta, 1903.

KHARIA. GAGAN CHANDRA BANERJEE, Grammar, Calcutta, 1894.

MON-KHMER LANGUAGES GENERALLY. HODGSON, Essays, as above. LOGAN, Essays in Journal of the Indian Archipelago, passim. KUHN, as above under Mundā. SCHMIDT, Grundzüge einer Lautlehre der Mon-Khmer-Sprachen, Vienna, 1905.

MON. HASWELL, Grammatical Notes and Vocabulary of the Peguan Language, Rangoon, 1874.

KHASI. W. SCHMIDT, Grundzüge einer Lautlehre der Khasi-Sprache in ihren Beziehungen zu derjenigen der Mon-Khmer-Sprachen …, Munich, 1904. ROBERTS, Anglo-Khassi Dictionary, Calcutta, 1875; Grammar, London, 1891.

TIBETO-BURMAN LANGUAGES GENERALLY. HODGSON, Essays, as above; also Essays on the Language, Literature, and Religion of Nepal and Tibet, London, 1874. LOGAN, Essays in Journal of the Indian Archipelago, passim. (Hodgson’s works contain many vocabularies of the different languages. Similarly there are many Nāgā and Kuki-Chin vocabularies in McCulloch’s Account of the Valley of Munnipore, Calcutta, 1859.)

TIBETAN. JAESCHKE, Dictionary, London, 1881; Grammar, London, 1883. RAMSAY, Western Tibet; A Practical Dictionary, Lahore, 1890. SANDEERG, Handbook of Colloquial Tibetan, Calcutta, 1894. DESGODINS, Dictionnaire Thibétain-Latin-Français, Hongkong, 1899. SARAT CHANDRA DAS, Dictionary, Calcutta, 1902. HENDERSON, Tibetan Manual, revised by AMUNDSE, Calcutta, 1903. BELL, Manual of Colloquial Tibetan, Calcutta, 1905. SANDEERG, Manual of Sikkin Bhutia Language, 2nd ed., Westminster, 1895. FRANCKE, Sketch of Ladakhi Grammar, Calcutta, 1901. WALSH, Vocabulary of the Tromowa [i.e. Chumbi] Dialect of Tibetan, Calcutta, 1905.

NEWARI. CONRADY, Grammar, in Zeitschrift der Deutschen Morgenländischen Gesellschaft, Vol. XLV (1891), pp. 1 ff.; Vocabulary, ib., Vol. XLVII (1893), pp. 539 ff. There is also a Vocabulary in WRIGHT’s History of Nepál, Cambridge, 1877.

RONG or LEPCHA. MAINWARING, Grammar, Calcutta, 1876; Dictionary (revised by Grünwedel), Berlin, 1898.

DHIMAL. HODGSON, Grammar and Vocabulary, in Vol. I of his Essays Relating to Indian Subjects, pp. 1 ff.

BAHING. HODGSON, Grammar (pp. 353 ff.), Vocabulary (pp. 320 ff.) of Vol. I of Essays, &c., as above.

VAYU or HAYU. HODGSON, Grammar and Vocabulary, pp. 216 ff. of Vol. I of Essays, &c., as above.

AKA. ANDERSON, Vocabulary, Shillong, 1896.

BÂRÂ. ENDLE, Grammar, Shillong, 1884. SKREFSRUD, Grammar, Benagoria, 1889.

CHUTIA. BROWN, Grammar, Shillong, 1895.

GÃO. RAMKHE, Dictionary (Bengali-Gāo), Tura, 1887. PHILLIPS, Grammar, Shillong, 1904. MASON, Dictionary, Shillong, 1905.

TIPURĀ. ANDERSON, Vocabulary, Shillong, 1885. RADHA MOHAN DEV VARMAN THAKUR, Grammar (in Bengali), Comilla, 1900.

MIKIR. NEIGHBOR, Vocabulary (Eng.-Mikir), Calcutta, 1878. KAY, Vocabulary (Eng.-M.), Shillong, 1904.

ANGÂMI. McCABE, Grammar, Calcutta, 1887.

NAMSANGIĄ. ROBINSON, Grammar, in Journal of Asiatic Society of Bengal, Vol. XVIII (1849), Part I, pp. 324 ff.; Vocabulary, pp. 342 ff.

MANIPURI. PRIMROSE, Grammar, Shillong, 1888.

RANGKHOL. SOPPITT, Grammar, Shillong, 1887.

THĀO or JANGSHEN. STEWART, Grammar in Journal of the Asiatic Society of Bengal, Vol. XXIV (1855), pp. 582 ff.

SIYIN. RUNDALL, Manual, Rangoon, 1891.

LAI. MACNABB, Handbook of the Haka or Baungshe Dialect of the Chin Language, Rangoon, 1891. NEWLAND, Grammar and Dictionary, Rangoon, 1897.

LUSHEI or DULIEN. BROJO NATH SHAHA, Grammar, Calcutta, 1884. LORRAIN AND SAVIDGE, Grammar and Dictionary, Shillong, 1898.

KHYENG. FRYER, Grammar and Vocabularies in Journal of the Asiatic Society of Bengal, Vol. XLIV (1875), Part I, pp. 39 ff. HOUGHTON, Essay on the Language of the Southern Chins, and its Affinities, Rangoon, 1892; Vocabulary in Journal of the Royal Asiatic Society, Vol. XXVII (1895), pp. 727 ff.

KHÂMI or KHWEYMI. STILSON, Grammar in Journal of the American Oriental Society, Vol. VIII (1866), pp. 313 ff. HOUGHTON, Vocabularies in Journal of the Royal Asiatic Society, Vol. XXVII (1895), pp. 111 ff.

KACHIN or SINGPHO. NEEDHAM, Grammar, Shillong, 1889. HERTZ, Grammar, Rangoon, 1895. HANSON, Grammar, Rangoon, 1896. SYMINGTON, Vocabulary, Rangoon, 1901.

BURMESE. JUDSON, Grammar, Rangoon, 1888; Burmese-Eng. Dictionary (revised by Stevenson), Rangoon, 1893; Eng.-B. Dictionary, 4th ed., Rangoon, 1894. TAW SEIN KO, Elementary Handbook, Rangoon, 1898. ST. JOHN, Reader, Oxford, 1894.

SIAMESE-CHINESE LANGUAGES.

KAREN. CARPENTER, Anglo-Karen Handbook and Reader, Rangoon, 1875. WADE and BINNEY, Anglo-Karen Dictionary, Rangoon, 1883.

SHAN. CUSHING, Grammar, Rangoon, 1871; Dictionary, Rangoon, 1881; Handbook, Rangoon, 1888.

AHOM. GRIERSON, Grammar in Zeitschrift der Deutschen Morgenländischen Gesellschaft, Vol. LVI (1902), pp. 1 ff.; Vocabulary in Journal of the Royal Asiatic Society, 1904, pp. 181 ff.

KHÂMTI. NEEDHAM, Grammar, Rangoon, 1893.

SELUNG. O’RILEY, Vocabulary in Journal of the Indian Archipelago, Vol. IV, pp. 411 ff. (There is also a Vocabulary in the British-Burma Gazetteer.)

NICOBARESE. DE ROEPSTORFF, Dictionary, Calcutta, 1884. MAN, Dictionary, London, 1889. TEMPLE, see below.

ANDAMANESE. MAN and TEMPLE, Lord’s Prayer in the Bojigngijida (South Andaman) Language, Calcutta, 1877. PORTMAN, Notes on the Languages of the South Andaman Group of Tribes, Calcutta, 1899. TEMPLE, Census Report for 1901 of the Andaman and Nicobar Islands, Calcutta, 1903, Chapter iv, ‘The Languages,’ pp. 96 ff.

BURUSHASKI. Grammar and Vocabularies in BIDDULPH, Tribes of the Hindoo Koosh, Calcutta, 1880. LEITNER, Dardistan, Lahore, 1877; Hunsa and Nagyr Handbook, Part I, Calcutta, 1889.