ture, and has introduced, as he says, into the northern languages some affixes for those in former use. In admitting this, however, he admits more than he seems aware of. For we very seldom, or rather never find that a people, though receiving a great number of foreign words into the dictionary of its language, have adopted at the same time a foreign grammatical system, so that the mere fact, that one part of the grammar of the northern languages is evidently of Sanscrit origin, would seem to speak by itself very much in favour of admitting the same for the other part.

But then the chief point is to consider, whether the instances brought for ward as unexplainable by the rules of Sanscrit grammar and by the principles of the general structure of the Indo-Germanic languages, may not be found to be formed by grammatical elements which have been similarly used by the Arian languages, particularly by their modern representatives. And this I think I am able to do for every grammatical form which Dr. Stevenson has pointed out as non-Sanscritical or Turkish, because he found it not exactly the same as in the old and classical Sanscrit, while a comparison of the modern development, which other Indo-Germanic languages have taken, will clearly show the analogies existing between the changes which the Indian language has experienced in the course of two thousand years and those known in other branches of the Indo-Germanic family.

It may be remembered that at present I meditate only a vindication of the Bengali language, which, if successful, will perhaps throw some light also upon the other northern dialects. As far as the southern languages are concerned, I abstain from giving any decided opinion, and shall content myself with noticing some coincidences between them and the dialects of the north.

Beginning with the declensions, Dr. Stevenson remarks, that these eight languages (viz. Bengali, Hindi, Gujerathi, Marathi, Telugu, Carnatica, Tamil and Singhalese) are all deficient in the number of cases required to mark the different relations of nouns, and supply the deficiency by particles placed after the root or some of the cases.

This is a fact, which, far from being surprising, would have been anticipated by every one acquainted with the relation in which modern languages stand to their parent tongues. The original and expressive forms by which the old language of India formed its admirable system of declension, have in the course of centuries, and particularly during the lapse of an illiterate middle age, lost their pure form and their distinctive power. But the Indian language found in itself the principle and elements of a new life, and we find it again at the period of its regeneration in the possession of richer and more powerful means than many of the modern languages of Europe can boast of. It is true that the dual of the Sanscrit language has entirely disappeared in Bengali, and that the nominative is the only case of the plural which has preserved an original form; but all the elements which have been substituted in order to form the number and cases of words are undoubtedly of Sanscrit origin; and we find sometimes the first traces of their grammatical employment in ancient works, and much more in the modern books, and especially in those of the Buddhistic collection.

The same is to be said of the gender of substantives and adjectives. The substantives, adjectives, pronouns and verbs have no different forms in Bengali for the masculine, feminine and neuter; and we meet only with some feminine terminations in certain cases, where it was necessary to distinguish the two genders, as vágh, a tiger; vághí, a tigress; khudá, uncle; khudí, aunt. The changes of the final letter which take place in these words are entirely founded on the rules of Sanscrit grammar, only that they have been sub

jected to the influence of the historical progress of the Indian language. Sometimes the feminine is also expressed by composition, just as we say, a she-goat, a French-woman, while in French and German the feminine is expressed by the mere change of the final letter. Thus the Bengali say, sasaru, a hare; strísasaru, a female hare; inréj, an Englishman; inréjér méyé, an English woman; méyé being the Sanscrit word maya, which means illusion, deception, or according to the notions of India, woman, and in a philosophical sense the female magic power, or the whole apparent world, which exists as long as the eternal soul looks upon it as existent, but vanishes as soon as the great Self returns to itself and gets free from the passion of worldly existence.

As to the single cases of declension, Dr. Stevenson further remarks, that there are several striking analogies running through most of these languages in the letters that characterise the principal cases. Thus the letter n is a very general characteristic of the genitive singular. It enters into the Gujerathi common genitive no, ni, num; the ancient Marathi genitive chéni, now usually contracted into chi and into the Tamil in; in all of which it runs through all the declensions. It is found also in the ni of the first of the three declensions in Telugu, and in the ana and ina of the first and fourth of the four Canarese declensions. It is singular, Dr. Stevenson remarks, that in the Turkish the termination of the genitive ung should afford so near a parallel to the above, and that we should have the remains of such a genitive in mine and thine, and the Germans in mein, dein, sein.

Although upon this point the Bengali is left quite unmentioned, because its genitive in r is of too clear a Sanscrit origin, yet I must say a few

words upon the n, as the sign of the genitive case in the languages quoted by Dr. Stevenson. Gothic forms like meina, theina, seina, are certainly puzzling at first sight, not however so much as for it to be necessary to assign a Turkish origin to them. It can easily be seen that the genitive has often, as far as the sense is concerned, the function of an adjective, so that phrases like "the work of the day," "the tribes of the mountains," may be expressed by "the daily work," "the mountainous tribes." It is also generally admitted that some genitive formations in the Indo-Germanic languages have preserved a close affinity to the formations of adjectives, with the only difference that the latter have different terminations for gender, number and cases, and could therefore be declined again like substantives. In some Indian dialects, as for instance the Hindi, we find even genitives with different terminations for the different genders. I do not say however that either the adjective has been derived from the genitive, or the genitive from the adjective, but I only maintain that the principle of their formation has been the same. Now it is known that the suffix na is of very frequent occurrence for the derivation of adjectives, and I have therefore little doubt that forms like the Gothic meina (bearing some analogy to the Zend mana) ought to be considered as adjectival formations; just as in Greek and Latin, éμós, oós for èμov, σov, meus and tuus for mei and tui. We may observe in Sanscrit also how the nasal sound n extends its influence in forming new bases to which the regular terminations of the cases are added, a fact which, particularly in reference to verbal formations, has been profoundly illustrated by Professor Lepsius. I feel therefore inclined to consider the nasal sound in all the instances quoted by Dr. Stevenson as an augment of the inflectional base, while the final vowel in some of his quotations may have the power of the genitive termination.

For the termination ke, used for the dative and accusative in Bengali, Dr. Stevenson brings forward the following analogies: ko in Hindi, ku in Telugu, gai in Singhalese, and gya or ge in Tibetan. He supposes this termination to take its origin from the Marathi dative, lági, derived from the verb lagane, to come in contact with, by changing the vowel, sharpening the consonant, and omitting the first syllable la, which syllable serves again as the sign of the dative in other Indian languages. Even if we admitted this derivation to be true, the word lagane is a well-known Sanscrit root, which Dr. Stevenson is very well aware of. But how fond he is of the autochthonical origin of these forms, we may see from the following remark of his: that this word lági itself, he says, may be derived from the Sanscrit is no objection whatever; for it may have been derived from a root common to many languages, and be just as independent of the Brahminical tongue as our own word lug.

There are two methods of accounting for grammatical elements which occur in modern languages. The one may be called the linguistical, the other the historical. The former consists in pointing out analogies between the form and meaning of inflectional elements in different languages of the same family. This method has generally been adopted and carried out successfully by Prof. Bopp and his school. It is indeed the only possible method in comparing the grammatical forms of languages which historically and geographically stand so far the one from the other, as for instance, the German from the Sanscrit. In comparative researches of this kind it is only required to trace analogies in the form and character of the elements, which constitute the grammar of a language, and to show etymologically the origin and the development of these grammatical forms. Whether the one language be anterior in its formation, and whether there existed a historical connection between them, is a question which originally has nothing to do with these linguistical inquiries.

The case however becomes different when we compare languages, the historical progress of which we can follow through certain periods. Here it becomes necessary to give to comparative inquiries as much as possible a historical character, by trying to explain modern grammatical forms by elements, which were used, though in a different way, by the same language in its anterior state, and to show if possible the period of transition froin the one to the other. Thus in a comparative analysis of the modern Persian grammar it would be necessary first of all to have recourse to the previous forms under which the Persian language appears to us at certain historical periods, and only in the case that neither the grammar of the Pazend and the Pehlevi, nor that of the Achaemenidian or Zend language furnishes the key for the grammatical forms of the modern Persian, it would be of interest to look for analogies in other kindred languages. For it is certainly true, though difficult to account for, that in several cases, where a historical connection exists between two languages, it is notwithstanding impossible to explain the grammatical forms of the one by those of the other, while languages, distant in time and place, afford the most unmistakeable analogies. Although then we prefer, when an opportunity is given, the historical method, yet we must admit even for languages, which have a historical growth, like the Bengali, the New Persian, the New German, &c., the right of the merely linguistical method, and I choose the present case, the question about the origin of the syllable ke, as sign of the accusative, as an opportunity for contrasting the relative merit of these two methods.

Even from a linguistic point of view it is difficult to find an analogy be

tween the formation of the accusative in the Indo-Germanic languages and in Bengali. An accusative formed by means of a syllable like ke, seems to be quite foreign to the genius of these languages. The resemblance which is found in the Gothic and Anglo-Saxon declension, in which there is a k for the accusative termination of personal pronouns, is not admissible, because the origin of this termination is founded on linguistical rules, so essentially Germanic, that it is not possible to apply the same rules to an Indian dialect. There is however some utility to be drawn from this comparison, for we see that Gothic accusatives (mi-k=me, thu-k=te, si-k=se) are derived from their pronominal roots by means of the same particle as the datives (mi-s= mihi, thu-s=tibi); and we find in the syllable sma, from which the k of the accusative and the s of the dative are both derived, a termination not restricted, like other terminations, to the accusative only, but serving, by means of its general and extended signification, to express, like the rá in Pehlevi and New Persian, at the same time both the accusative and the dative. We believe therefore that the Bengali ke is not to be considered as implying the relation usually represented by the accusative or dative, but as a particle of purely demonstrative power. As to analogy in other languages, Latin forms, as hi-c, hui-c, hun-c, tun-c, &c., may be quoted where the final c is the same as that we see repeated in forms like hi-c-ce, hun-c-ce, &c., serving to enhance the demonstrative signification. According to the theory of Prof. Bopp (Comparat. Grammar, § 305), this ce must be considered as an indefinite ticle, or rather as a particle, which, compounded with an interrogative pronoun, takes away its interrogative power and changes the interrogative into an indefinite pronoun. The same savant traces this particle through different languages, such as Sanser. ci-t (kaçcit, some one), ca-na (kaçcana, any one), Dor. Ká (TOKά, once), Ion. Té (Toré), Lat. que (quisque), quam (quisquam), ce (hic and hicce), pe (quippe), piam (quispiam), Goth. uh (hvazuh).

par

But it seems necessary that a distinction should be made between two particles, which under a similar form have played very different parts in the progress of languages. The one, almost the same as the copulative particle (ca, Té, que), serves, in generalizing, and to form from the root of the interrogative, an indefinite pronoun, when the other gives a determinate form. We recognise the former in words like kaçcit, kaçcana, Toká, noré, quisque, quisquam (always in negative phrases like Sanscrit kaçcana), quispiam, hvazuh, (Modern German wasauch). Adopting the system, which as far as I know has first been introduced by the ingenious G. Curtius of Berlin, this change of letters in Sanscrit, Greek and Latin may be represented in the following equation: pañca, névтe, quinque=ca, re, que. Sanscrit cit and Latin quam are enlargements of the same particle, as quispiam, which corresponds to Sanscrit kasapi, regularly changed into ko'pi. But we cannot believe with Prof. Bopp that the same particle, which by its peculiar power gives to interrogative pronouns an indefinite signification, has given rise to the demonstrative pronoun hic, by being compounded with the Sanscrit interrogative pronoun ka and ki. Besides, as Professor Bopp acknowledges that in hic-ce, hunc-ce, &c. the latter ce is the repetition of the same element, we find already combined with the pronoun (hi-c), and as the genius of the Latin language does not permit a doubt on the purely demonstrative meaning of this particle, we do not think that it is altogether contrary to the system of this learned grammarian, to consider ce as a determinative particle, different from the other and identical with the Greek ye and the Bengali ke.

But though from a linguistical point of view we might admit the Bengali termination ke to have its origin in this demonstrative particle, it is still the

question whether a comparison of languages, historically connected with the Bengali, might not furnish a more satisfactory solution. In the first case, we must remember that in Bengali itself ke is very often omitted, and that the accusative is represented by the same form as the nominative, when the whole structure of the sentence shows that the substantive, dependent upon a verb, is to be taken as its object and therefore as an accusative. Besides, ke is not so much to be considered as the termination of an accusative, but rather as that of an objective case, because it is frequently used to represent the dative also, as Hari bahudhan Haridáske dilen (Hari gave much money to Haridâs). Nor is it, like the other terminations r, te, rá, added to the secondary form of a substantive (manushye-r, manushye-te, manushye-rá), but to the absolute form (manushya-ke, purush-ke). Now if we go back to Sanscrit, particularly in its more modern form, and to the Pracrit dialects, we may observe a great tendency of the language to put the suffix ka at the end of many words without changing considerably their meaning. It is true that in some cases the affix ka serves to express contempt, pity, &c., but generally the meaning of the word remains the same, only assuming a more concrete, objective or neuter character. Thus lohita means read, lohitaka, a ruby, vác is speech, vácikam, a delivered speech or discourse. There exists a close relation, logically as well as grammatically, between the neuter in its nominative and accusative and the accusative of the masculine. The accusative represents the substantive, which is active and independent, if expressed in the nominative, as a passive object, and we may account thereby why in many cases the same grammatical element, which serves to express the neuter gender, has been employed for expressing the objective case of the masculine, as am in Sanscrit, um in Latin, ov in Greek. If then the affix ka has already in Sanscrit the signification which we have just explained, it might seem well-adapted for words which by their relation to other words convey the meaning of passive objectivity. Although therefore this particle may not have become, neither in Sanscrit nor in Pracrit, the mere conventional sign of the accusative case, yet its analogous use gets so extensive in Pracrit, and particularly in the Çâkkarí dialect, that we have sufficient reason for tracing the Bengali ke historically back to the Sanscrit and Pracrit ka.

The termination of the ablative also, which is te in Bengali and Marathi, ta in Pushtoo and in Singhalese, and which Dr. Stevenson considers therefore as a remnant of the language of the aboriginal Hindús, is certainly of Sanscrit origin. In Bengali te is at the same time the termination of the ablative and the locative. Besides, there is still another more Sanscritic termination in e, for the locative of words ending in a consonant or the vowel a. The same form is, by a false analogy, employed also for words which end in other vowels, such as râtre (at night), instead of rátrite or rátrikále. The termination te, if employed as the sign of the dative, corresponds to the Sanscrit termination tra. The change of tra into te is justified by the aversion for all harsh sounds and double consonants which we frequently find in modern languages. Thus Professor Bopp derives ingeniously the Greek σe in alooé, &c., from Sanscrit tra, supposing the suppression of r and the usual transition of t into s. The same suppression of the r takes place in Bengali, where the short a, as usually at the end of words derived from Sanscrit, is changed into e.

The termination te, if used as the sign of the ablative, represents the Sanscrit suffix tas; and the change of tas into te appears even more regular, when we remember that the Pali and Pracrit languages suppress equally the Instead of changing the a into o, as in the case of these dialects and

final s.