\ V* ^ ^U- \? %%' c** % ** otf ^ >iia' > : **4 & %. ^ v % \V tP- r"^ \ • V , COMPARATIVE GRAMMAR LONDON PEINTED BY SPOTTISWOODE AND CO. NEW-STEEET SQUAEE THE STUDENTS HANDBOOK OF COMPAEATIVE GEAMMAE APPLIED TO THE SANSKRIT, ZEND, GREEK, LATIN, GOTHIC, ANGLO-SAXON, AND ENGLISH LANGUAGES. BY EEV. THOMAS CLAEK, M.A. LATE HEAD 1IASTEE OE THE PEOPEIETAET SCHOOL, TAUNTON. S LONDON: LONGMAN, GREEN, LONGMAN, ROBERTS, & GREEN. 1862. By transfer from Pat. Offl** L.*>« • Ayr* 1*14. PREFACE. Comparative Grammar treats of several languages in conjunction. It explains what has become obscure in one by that which remains intelligible in others. It is thus enabled to trace the origin of grammatical forms, and to illustrate the nature of language itself more fully and satisfactorily than could be done by the in- vestigation of any one language separately. Lord Bacon judged that to be the noblest form of grammar* which should compare the properties of many tongues, ' both learned and vulgar,' and so attain to a perfect system — as Apelles shaped his Venus, not ac- cording to one model, but from the separate beauties of many individuals. No linguist, however, appeared for centuries to carry out this idea. Little was done until the discovery and study of Sanskrit literature gave the impulse and sup- plied the materials for those works upon the subject which have appeared in Germany during the last thirty years. * ' Nobilissima grarninaticse species.' VI PKEFACE The results of this study are already considerable. The resources of language have been applied to the elucidation of Eoman history,* and have established some important facts which escaped the penetration even of Mebuhr. The same means may reasonably be expected to place in a much clearer light the early social condition of many of the nations of antiquity* It is also obvious that what makes language itself more intelligible will render important service in philo- sophical and ethical enquiries. But it is in the acquisition and teaching of lan- guages that Comparative Grammar will be found most extensively useful. It has been already applied to the Greek and Latin grammars ; and it will not long be possible for anyone to teach them satisfactorily who has not at least made himself familiar with its leading principles. An acquaintance with Comparative Grammar will be equally serviceable to the learner. Hitherto he has had to learn by rote what was never explained. He was told, for instance, that habes and habetis are the sin- gular and plural of the same word, without being able to see how the one was derived from the other. He had to learn by heart several hundred equally unintel- ligible symbols, as mysterious to him as the Egyptian hieroglyphics. No doubt a great part of the reproach which has fallen upon the study of languages, as being a mere exercise of memory, is the consequence of so * See Mommsen's Rbmische Geschichte. PKEFACE VU many dead forms having to be learnt. As soon as life is imparted to them by proper explanation, the study of languages will be found to promote a more healthy development of the entire mental constitution than any other study. Many a youth who, under the old system, was glad to give up his Latin in exchange for the phy- sical sciences, or was content to be behind others in a matter which he deemed to depend only upon memory, will then see that his judgment is called into exercise, and will feel as much pleasure in the study of languages as in that of geology or chemistry. It is, therefore, not only important that the teacher should master this subject, but desirable also that its leading features should be made known to boys in the early part of their studies. The works hitherto published are too extensive for general use, and one in a smaller compass appeared to be wanting. In the following pages I have attempted little more than to put into a popular form what has been already established, and thus to meet the wants of those to whom the profounder and more voluminous works upon the subject are inaccessible. The materials have been in a great measure derived from the masterly Ver- gleichende Grammatik of Prof. Franz Bopp, to whom I desire to make the fullest acknowledgment here, as it would have too much broken the continuity of the work to state in every case how far his views have or have not been adopted. I have never differed from him without hesitation ; and when his reasons have not appeared to Vlll PREFACE me to be conclusive, if no better solution offered itself, I have stated his views, and have added his name as an authority. In the words of Monsieur Kegnier,* 'I shall be glad if my book helps to increase the number of the readers of his great work.' A larger number of languages could not well have been included in a work of this compass, and fewer would not have sufficed adequately to illustrate the principles of Comparative Grammar, and to give the subject a practical bearing for the English student. I should have been glad to adopt Dr. Lepsius' ad- mirable alphabetical system, but it would have required too great a departure from English associations for so elementary a treatise. The employment of Greek characters seemed un- avoidable. In other respects I have endeavoured to make the work available for the merely English student. * De la Formation des Mots Grecs. CONTENTS. Sect. I. Introduction 1-3 The Indo-European Family of Languages 4,5 The Indie Branch 6 The Iranic Branch . 7 The Letto-Slavic Branch . 8,9 The Grasco-Italic Branch . 10-12 The Celtic Branch . 13 The Teutonic Branch 14, 15 The High German . 16 II. Alphabets : 1. Sanskrit . 17-19 2. Zend . . . . 20 3. Greek . . . 21 4, Latin . 22, 23 5. Gothic .... 24,25 6. Anglo-Saxon . 26, 27 I. Sounds 28-31 1. Sanskrit Vowels . 32-35 „ Consonants . 36-45 2. Zend . .... 46-61 3. Greek ..... 62,63 4. Latin ... 64-66 5. Gothic . 67-74 6. Anglo-Saxon 75-83 General Remarks 84-92 The Transmutation of Consona nts . 93-104 CONTENTS IV. Roots Verbal Roots .... Guna and Vriddhi . First Conjugation Second Conjugation Table of Roots V. Stems Endings of Nominal Stems . Strong and Weak Stems . VI. Formation of the Cases of Nouns In the Singular (The Pronominal sma) . (Latin Names of Places in the Locative Singular) .... In the Plural .... In the Dual .... VII. Adjectives, Comparison of . VIII. Numerals, Cardinal . „ Ordinal Numeral Adverbs IX. Pronouns of the First and Second Persons „ of the Third Person . Derivative Adjective Pronouns Pronominal Adverbs X. The Verb Formation of the Middle Voice Persons : First Second Third The "Weight of Personal Endings Conjugations .... The Formation of Tenses Present .... Imperfect • . . . Aorist Perfect .... Pluperfect .... Future .... Sect. 105-113 114,115 116 117-122 123-129 130 CONTENTS XI Sect. Moods : Potential, Optative, Conjunctive . 241, 242 Imperative 243 Conditional 244,245 Passive, Causal, Desiderative, and Intensive Verbs 246-249 Denominatives 250 251 251a 252 253 254-257 258 259, 260 261 262 263 264 265, 266 267-270 271 272 273 274 275 276 277 278 279 280 281 282 282a 283 284 285 286,287 288-296 ". Derivation and Composition. a) Derivatives : Forms in NT or NTA 55 WANS . 55 MANA . ?> NDO ?> TAR. 55 TA . 55 NA . 55 TI, NI . 55 TU, NU . 55 SA, SE . Germanic Infinitive . Greek Infinitives Forms inYA . Bare Roots . Forms in A 5> I 55 U . 55 AN . 55 ANA. 55 AS . 35 LA, RA . 55 WA (VA) . 55 WAN (VAN) 55 NU . 55 MI . 55 KA . 55 TU . 55 TANA . 55 SYA. b) Compounds : Verbal . Nominal Xll CONTENTS XII. Indeclinable Wokds 1. Adverbs 2. Conjunctions 3. Prepositions Sect. 297 298 299 300 The following abbreviations are employed a. s. for Anglo- Saxon. e. „ English. ger. „ German. g°- „ Gothic. g r - „ Greek. 1. „ Latin. Du. for Dual. F. „ Feminine. M. „ Masculine. lith. for Lithuanian. o. g. „ Old German 0. s. „ Old Saxon. s. „ Sanskrit. V. „ Vedas. z. „ Zend. ST. for Neuter. PL „ Plural. S. „ Singular. THE STUDENT'S HANDBOOK OF COMPARATIVE GRAMMAR I. INTKODUCTIOK 1. Language is that which principally distinguishes man from the lower orders of creation. It is insepa- rable from his mental existence. Thought is internal language, and language is external thought. So dis- tinguishing a quality of human nature could not fail to arrest attention in a reflecting age. Locke and Leibnitz recognised its importance in the philosophy of the human mind. But it was necessary to em- ploy a mode of investigation which was then but little understood, before the essential connection and true relations of mind and speech could be discovered. As the animal economy needed the help of Comparative Anatomy for its elucidation, so the Comparison of Languages alone can explain some of the operations of the human mind, In other respects, also, Comparative Philology was long regarded as of great importance. It was looked upon as serving not only to promote a more exact ac- quaintance with particular languages, but also as deve- loping the nature of language itself, and thus aiding 2 INTRODUCTION in the solution of the difficult problem of the origin of language. These and other considerations led to many isolated efforts in this direction at an early period, but it is only in the present century that the subject can be said to have assumed the features, and acquired the propor- tions of a science. Since then it has had to encounter the fate of every new science. It has been magnified by extravagant pretensions, and assailed by equally extravagant depreciation. Half knowledge on the one hand, and utter ignorance on the other, are equally prejudicial. But meanwhile a large number of works upon the subject have appeared from men of literary eminence, composed in a spirit of moderation, and bearing evidence of great power and deep research. They have placed the science of language upon a secure basis, and furnished the materials for its wide and rapid extension. 2» Though an acquaintance with Comparative Gram- mar will not do away with all the labour and difficulty of acquiring the knowledge of languages, it will, never- theless, facilitate their acquisition. Much of the diffi- culty to a beginner lies in the strangeness of the forms which he meets with in a new language. Whatever diminishes this strangeness will proportionately diminish the difficulty of learning the language. He who is aware, for instance, that in certain cases, German words have s where the corresponding words in English have t, will more readily acquire a familiarity with the German words was, das, iveiss, from his previous acquaintance with the English words, ivhat, that, ivhite, than another will who begins his study in ignorance of this fact. The relation of Comparative Philology to history GENERAL REMARKS 3 admits of similar remarks. It has already thrown much light upon historical points which were obscure, and which, but for the scientific study of language, must have remained obscure. Some valuable illustra- tions of this may be found in the earlier part of Mommsen's ' Komische Greschichte.' But it is specially in regard to the ante-historical period of human exist- ence that the Science of Language promises important results. A nation naturally desires to discover its origin, but history can trace its course only from the time when it had already reached a mature age. Its infancy, boyhood, and youth are hidden in a mysterious obscurity, or coloured by legendary tales. The Divine Eecord offers but few hints that could serve to connect modern nations with the earliest period ; nor were they probably intended to prevent the inquiries of science, any more than the narrative of the fourth day's crea- tion was designed to supersede the investigations of Astronomy. It is not, however, exclusively nor chiefly on account of its practical utility that Comparative Philology deserves attention. It is worthy of being pursued for its own sake. Even in its present stage it shows that human language deserves no mean place among the objects of nature. It exhibits a growth as wonderful as that of the plants which furnish the materials for the science of botany, and develops laws as subtle as those by which astronomy explains the motions of the planets. 3 B The great variety of languages is perhaps referrible to the intimate connection between spirit and speech. The characteristics of the one are expressed in the other. The human mind is exactly the same, probably, in no two individuals. Some peculiarity distinguishes the spiritual nature of every member of the human family. This B 2 4 INTRODUCTION variety is reflected in the outward expressions of mind. In proportion as the intellect is cultivated, the coun- tenance assumes a more distinct individuality. In the higher stages of civilisation no two faces contain exactly the same features, whilst a want of culture tends to leave a dead uniformity of expression. Upon language, also, the mind impresses its own individuality, and but for artificial restraints against multiplication there would be almost as many languages as individuals. A thousand distinct languages are said to be spoken upon the earth. The number of dialects is immensely greater. There are places even in Europe where the inhabitants of each hamlet or small district speak such different dialects of one language, as to be almost unintelligible to each other. The uneducated inhabitants of one county in England, in some cases, deem the language of the next county strange and almost barbarous. The people of Lancashire and those of Hampshire, both speaking genuine English, would be almost unintelligible to each other. This natural tendency to diversity is checked by artificial means. The use of written and printed characters, the influence of education and social inter- course, impart a certain degree of uniformity to the speech of the same society, or town, or nation. It is evident that the attempt to form a Comparative Grammar of the entire languages of the human race would be futile. The materials of those languages, even, which have been stereotyped in a classical character are not yet in a sufficiently forward state to be all embraced in the same work. 4- The Indo-European branch of the great family of languages contains the most important literature, is the most easily accessible, and has, to a great extent, been SEVEN CLASSES OF LANGUAGES 5 already examined and classified by eminent men, both of the present and of past generations. This division includes nearly all the languages spoken in Europe, and a large proportion of those spoken in Asia, west of the Ganges. Its range has also been extended in modern times by migration. The English, French, and Ger- man languages are spoken in the principal portion of North America, and in isolated parts throughout the rest of the world. Various terms have been adopted as a collective designation of these languages. Amongst them Indo- European seems the best adapted for our present pur- pose. It is at once intelligible to English readers, and sufficiently comprehensive; for until these languages were carried abroad by emigration, they were little, if at all, spoken either eastward of India or westward of Europe, whilst at a very early period they stretched in an almost uninterrupted chain from the Ganges to the Atlantic. This division of languages, however, does not include those commonly called Semitic, and a few other strangers, of minor importance as far as concerns the literature which they possess, or the numbers by whom they are spoken, are mixed among the Indo-European family. Thus the Georgian and Turkish in Asia ; the Turkish, Hungarian, Finnish, Lappish, and Basque in Europe, must be left out of account. Probably further investigation will show that some of these are more or less intimately related in origin and development to their neighbours. 5. The languages thus remaining under the desig- nation Indo-European, are numerous and important. They may be conveniently arranged in seven classes, of which two belong to Asia and five to Europe. Another b INTRODUCTION arrangement of them might be made into three divi- sions, the first including five of the above classes, distinguished by the oldest grammatical forms, the second that which has intermediate forms, and the third that which has the newest forms. The reason for this will appear more fully hereafter. It will be suffi- cient at present to mention that in the nouns, 1. jugum, e. yoke, ger. joch, the letters g 3 &, ch (for Mi), distin- guish these divisions, and that the same distinction is indicated in the verbs, 1. docet, e. teaches, ger. zeigt, by the letters, d, t } z (for ts). 1. INDIC or Sanskrit. 6. In considering the seven classes, we begin with the most easterly, and that which also has the most ancient literature, i, e. the Sanskrit. It is a language which, though possessing voluminous and valuable works in prose and verse, has but recently become known to Europe. The Scienoe of Language, as it is now pursued, may, indeed, be looked upon as one of the results of the establishment of British dominion in India. For British residents, Sir William Jones amongst the first, collected and brought over the stores of this ancient literature, which German philologists, with profound research and indomitable perseverance, have made subservient to the elucidation of all the sister languages. The modern dialects of Northern India, though of the same stock, interest us less in the following inquiry. The elucidation which we seek will be sufficiently supplied by the Sanskrit, under which name, however, we include the Vedas, as well as the more recent literature specially called Sanskrit. 1. INDIC LANGUAGES 7 The Vedas are not only the most ancient literature of India, but are also the oldest expression of thought in the Indo-European languages altogether. The late Professor H. H. Wilson, reasoning from the later to the earlier compositions, supposed the heroic poems in Sanskrit to have been written about 300 years B.C., the laws of Manu three centuries earlier, the prose Brahmanas, which elucidate the Vedas, 800 B.C., and the Vedas themselves about' 1300, B.C. This, however, can only be looked upon as an approximation to the chronology of the whole, for the several parts are evidently the work of different authors, and some were manifestly composed at a much earlier period than the time when the whole were collected and arranged together as we now have them. The Vedas thus contain a literature older than the Homeric poems, and, what is more important in a gram- matical point of view, it is the literature of a people who had migrated a much shorter distance from the primitive abode of the race, and undergone much fewer political and social changes than the Greeks, the forms of whose language, therefore, for this very reason, must be much nearer to the primitive type than either the Greek or any other branch of the same great family. The Vedas consist of four collections, the last being more recently made than the other three. Their names are: 1, Big-Veda; 2, Taj uv- Veda ; 3, Sama-Veda ; and 4, Atharva- Veda. Each consists of two divisions, i. e. the text, or Mantras, and the commentary, or Brahmanas. The former comprises principally sacred hymns addressed to various deities. Many of these are representatives of natural objects, e.g. Agni ('fire'), to which the first hymn of the Eig-Veda is addressed; the Maruts (< winds '), to which the nineteenth is in part 8 INTRODUCTION addressed. Many are addressed to Indra, who presides over the firmament, and sends or withholds the fer- tilising shower. The valley of the Indus appears to be the locality where these hymns were produced. They reflect simple habits of life, and a primitive state of society* The sacrifices offered are in general not costly, and the blessings implored are principally the material advantages of the present life. The language of these hymns exhibits a nearer approach than any other to the first forms in which thought must have been expressed by the Indo-European branch of the human family. It serves to clear up many points left in obscurity by the fragmentary state of other languages, and is an essential element in any just appreciation of the questions discussed in Com- parative Grammar. The classical Sanskrit, however, though of a later date, contains a richer literature, and more abundant materials for linguistic investigation. The country in which it appears to have been first employed as a spoken language, and to which, as a spoken language in its purity, it was confined, is India. The words preserved by Ctesias, in the time of Artaxerxes- Mnemon, show that the influence of the language had extended over the south as well as the north. The people by whom it was employed are probably the same race as those who, centuries earlier, sang the Veda hymn on the banks of the Indus. Everything shows that they came from the north-west, and gained posses- sion of India by conquest. The simple mythology which they brought with them was soon remodelled under the hands of poets and sages, and, probably in part through contact with the primitive inhabitants whom they found in the country. India thus became the home, and 1. INDIC LANGUAGES 9 the Sanskrit language the organ, of the Brahman religion. The literature which is contained in this language extends over a considerable space of time. It must date its commencement, at least, soon after the invasion of the country. The earliest productions have probably perished, and it would be some time before the heroic deeds of the invaders would be made the ground-work of the voluminous epic poems Mahdbhdrata and Rd- mdyana. It ceased to be a spoken language in conse- quence of the popular agitations in behalf of Buddhism, and the prominence given to the common dialects in opposition to the language of the predominant religion. Sanskrit, however, continued to be the language of the Brahman religion, of learning, and of poetry. It con- tains numerous works in various branches of knowledge from the earliest to recent times. The utility of this language in our present inquiry results from the almost perfect preservation of forms which have only a fragmentary and perplexing appear- ance in modern languages. The English word came, for example, is used by us as a past tense, although nothing in its letters indicates past time, unless it be a instead of the o, of the present come. This, however, is not felt to be a tense-sign at all, for it occurs in just the reverse way in the present break, as compared with broke. Nor is there any termination to the word came, except in the almost obsolete second person singular earnest^ to show what person and how many persons came, whilst in Sanskrit a corresponding word not only indicates the tense, but has no less than nine differ ent endings, to show whether it applies to the first, second, or third person, and whether it includes only one or two or more individuals. The enigmatical second 10 INTKODUCTION person singular, earnest, is explained by one of them. Again, the sarhe language has eight different endings for as many cases of a noun. One of them presents the form out of which the English possessive 's has come. This termination thus appears to be the remains of a form older than the English language itself, and not to have come from adding his to a noun, and then con- tracting these together as some have assumed ; thus taking f queen's own ' to be for ' queen his own/ and ? men's ideas ' for c men his ideas.' That such expres- sions occur in English writers as 'the king his horse/ only shows that whoever first originated them found the form of the possessive case obscure, and could not rest without explaining what they did not understand. Thus an appeal to older records supplies the part of the inscription which time had obliterated. The Sanskrit will generally be appealed to in the following pages ; the Vedas sometimes as having a few remains of a still older form ; but other languages nearly related to them will not come within our scope. They are the following : 1. The Pracrit, or popular dialects assigned to subor- dinate characters in Sanskrit dramas. They are gene- rally distinguished by a considerable softening of sounds and decay of grammatical forms. 2. The Pali, which was conveyed by banished Buddhists to Ceylon. It became, like Sanskrit, from which it differed principally in the loss of grammatical forms, a learned language. It contains theological works on the Buddhist religion as early as the fifth century a.d. 3. The Kawi, which was preserved in the Islands of Java and Bali as a literary and poetic language. Its grammatical forms became mutilated, very much 2. IRANIC LANGUAGES 11 like those of the Pali, by contact with a strange people. 4. The Gypsy language belongs to the same class, for after the most varied theories had been adopted respecting their origin, as indicated in the names Gypsies (Egyptians), Bohemiens (Bohemians), it has come to be generally admitted that this, singular people came from India. Their language is the old Sanskrit, though very much corrupted and mixed with foreign elements. The modern languages of India belonging to the same family are reckoned by Pott to be twenty- four in number. The principal is the Hindostanee. They all bear the relation of daughters to the Sanskrit, and not that of sisters, like the four noticed above. They have each a peculiar deviation of features from the primitive type, and have only a subordinate relation to our subject. 2. IBANIC. 7- The second class of the Indo-European languages is called Iranic, and belongs principally to the country between the Indus and the Tigris. It is so called from the Persian word Iran — another form of the Sanskrit word Aryan — applied to the region which stretches from the chain of the Hindoo Koosh to the Persian Gulf. The term Iranian includes a considerable number of languages. 1. Of these the oldest is the Zend, the language of the Zend-Avesta, or sacred writings ascribed to Zo- roaster. The country where this language prevailed, and the people by whom it was employed cannot be exactly defined, and the language itself presents many difficulties. Early investigations in it were carried on 12 INTRODUCTION principally by means of translations, and presented, therefore, but confuse^ and unsatisfactory results. Of late, however, it has been subjected to more scientific investigation, and much assistance has been derived from the discovery and elucidation of the arrow-headed inscriptions belonging to the time of the Achsemenidse. The age of Darius, to whom some of these inscriptions owe their existence, is well known, and it lends some help in conjecturing the age of the Zend-Avesta. For on comparing the grammatical forms of the inscrip- tions with those of the Zend language, the latter are found to be of an older character, and they, therefore, belong to an earlier date. They are thus referred to a period earlier than the fifth century B.C. These writings are loosely spoken of as the writings of Zo- roaster, but the date of his existence is a problem not yet satisfactorily solved. Besides, it appears that only a small portion can justly be ascribed to him, and, therefore, if the time when he lived could be fixed, it would leave the chronology of a great part of the Zend- Avesta doubtful. The great difference between the grammars of the Sanskrit and the Zend might favour the idea that some centuries had elapsed, after the separation of the two peoples, before the Zend acquired the form in which it was written. Such would, undoubtedly, be a reasonable inference if we could assume that the Zend was de- veloped from the Sanskrit. But the two peoples may have spoken very different languages even before they separated, just as in England the people of two counties, or the educated and uneducated in the same county, speak differently one from another. It appears to have been a religious movement which caused the separation of the Iranian from the Indian branch. The oldest 2. IRANIC LANGUAGES 13 Iranians were fire-worshippers. It is in the nature of things probable that the zeal for this religion would manifest itself chiefly in some particular province, and principally affect some particular class of the com- munity, probably not the most educated. They would, therefore, already have a characteristic — a so-called provincial — language. So that if the Zend-Avesta were written immediately on the separation of the two peoples, its language would differ very much from that of the educated classes whom they left, and whose writings have become known to us as Sanskrit. The word Iranian itself furnishes one illustration out of many which might be adduced. The Sanskrit word Aryan appears in Zend as Airyan, in accordance with a general practice that, when y follows a single consonant, i is inserted in the preceding syllable. Thus what one (perhaps the educated) called aryan (ar as in far), the other (perhaps the uneducated) called airyan (air like ir in fire). Even now, in English, where one says 6 are you coming,' another says, ' aire you coming.' Whatever caused the difference, it may have existed before the separation of the peoples, and, therefore, supplies no argument against the antiquity of the Zend-Avesta. We find here two circumstances which render the Zend valuable in a work on Comparative Grammar. One is that it presents another instance of a very ancient and almost primitive language of the Indo-European family, side by side with the Sanskrit. The second considera- tion is that it presents grammatical modifications pecu- liarly its own, and thus serves strikingly to elucidate the genesis of language. Pott seems inclined to fix its locality in Bactria, and Professor Schleicher calls the language Old Bactrian. 14 INTRODUCTION But it is not well to build too much upon what is merely hypothetical. 2. The next Iranian language in historical order is that of the arrow-headed inscriptions belonging to the reign of the Achsemenidse, which therefore belongs to the fifth century B.C., and to the land of Media. The phonetic decay of its form shows that it belongs to a later period than the Zend. 3. The modern Persian strikingly illustrates the destructive effect of time upon the framework of lan- guage. Like the English it has preserved but few remains of the inflections which are so abundant in the older languages of the same class. It was a hasty generalisation which led to the idea that the Persian was closely related to the modern languages of Ger- many and England, and that the fuller forms of older dialect w T ere only excrescences which disfigured them. A juster analysis has shown that in regard to gram- matical inflection these modern languages are but scanty ruins, whilst the Sanskrit presents a grand and almost complete edifice. 4. To the east of Iran there are a few scattered populations, which have issued from the same stock, and speak languages belonging to the same class as those mentioned above. They have thus far rendered but little service to Comparative Grammar, and here require but a brief notice. The Pushtoo, in Affghanistan, is nearly related to the Persian, but exhibits some peculiar grammatical forms and has a large admixture of foreign words. The Beloochee, of which but little is known, is spoken at the mouth of the Indus, and the Parsee by the Parsees in Guzerat and elsewhere. 5. To the westward we meet with three other mem- 3. LETTO-SLAYIC LANGUAGES 15 bers of the same family. The language of the Koorcls in Koordistan ; the Ossetic, spoken by a small mountain tribe on the Caucasus, who call themselves Iron, and their country Iron-sag, thus preserving the tradition of their origin ; and the Armenian, which possesses an important literature, including a translation of the Septuagint, made at the beginning of the fifth century, and of the works of some of the Greek fathers. It seems likely, when further examined, to render more extensive service upon questions of Comparative Gram- mar than any other language of this class, except the Zend. Almost all the modern literature of this class of languages contains a considerable admixture of Semitic words, in consequence of their geographical position, and of the spread of Mohamedanism. The Iranian class of languages is remarkable for the number of characters employed in writing them, the Zend, the Arrow-headed inscriptions, and the Armenian, having characters quite distinct one from the other. 3. LETTO-SLAYIC. 8. The third class consists of the Lettic and Slavic languages. 1. The Lettic includes the Lettish, the Lithuanian, and Old Prussian. The Lettish is spoken in Kurland and Livonia, but has been much corrupted by the influence of other languages. The Lithuanian is now spoken by only a small popu- lation in the north-east of Prussia and in the neigh-' bouring districts of Russia. Its literary store is very circumscribed and of recent date. It has, however, considerable value on account of its almost perfect pre- 16 INTRODUCTION servation of some of the original grammatical forms. The Old Prussian, which has now ceased to be a spoken language, bore a very close resemblance to the Lithua- nian. They were neither of them much exposed to the influence of those external causes which generally modify the language of a people. The Lithuanians occupied a flat and marshy country, and never attained a very high state of civilisation. The character of their home, together with the inactive political life which they have led, will help to account for the unaltered state of their language. The Sanskrit, for instance, has probably preserved the interrogative pronoun in its primitive state. It there appears as has, so written three thousand years ago. In the Lithuanian, as spoken at the present day, it is likewise has, whilst the Greek almost three thou- sand years ago had changed it to hos, the Latin above two thousand years ago, to quis, the Gothic, fifteen hundred years ago, to hvas, and we have it in the still further reduced form of hwo (who). The Lithuanian hatras, again, is almost identical with the Sans, hdtaras, whilst the Greek is poteros, the Lat. liter, the Go. hvathra, and Eng. whether. There is perhaps no other people who have spoken the original Indo-European language with so little alteration in many of its grammatical forms for the same length of time. 9. 2. The second branch includes a large number of languages stretching from the Adriatic to the Gulf of Finland, and from the coast of the North Sea to the Ural mountains. They are condensed together, how- ever, principally in Eussia and Austria. History cannot trace the Slavic population back to their original abode in Asia, but it finds them at a very early period in the extreme East of Europe. Herodotus, in the fifth cen- tury B.C., speaks of the BovSivoi, iv. 21, and Larcher, in 3. LETTO-SLAYIC LANGUAGES 17 his note upon the passage says, ' The opinion of Eennel and Heeren, who assign the Budini to the vicinity of Voroniej near the southern border of the government of Tambof, is entitled to preference.' The description of them accords with the idea of their being a part of the Slavic people, and Shafarik, in his important work upon the antiquities of the Slavic race, does not hesitate to claim them. This people, therefore, who probably left the primitive abode at about the same time as the Iranians, are found on the north of the Caspian at the time when the Medes under Darius were leaving a written memorial of their presence on the south of the Caspian in the arrow-headed inscriptions. They appear in Dacia, north-west of the Black Sea, in contact with the Eomans under Trajan, at the beginning of the second century of the Christian era. We find the evi- dence of their presence later in Pannonia, for Buda is one of the names of Pesth, the capital of Hungary, and in more recent times they gave its name to the town of Bunzlau, more properly Budissin, in the Prussian pro- vince of Lusatia. Thus, as time passes on, we find them proceeding further westward. The Slavonians began to attain political consolidation and importance after the death of Attila, who with his Huns had held them in terror and subjection. The Bo- hemian nation appears as early as 650; the Bulgarian about the same time, but in greater power under Boris in 850 ; the Moravian under Eastislawand the farfamed Swatopluk during the ninth century; the Polish as early as the seventh century ; the Kussmn under Eurik in 862 ; and the Servian in the eleventh century. Some of these states for a short time extended their dominion along the south coast of the Baltic and westward to the mouth of the Elbe. The only independent Slavic c 18 INTRODUCTION nation at present is Russia, the rest being under foreign dominion, and the imperial family even of Russia is of German origin. The literature of the Slavonians is modern. The earliest remains go no further back than the ninth cen- tury. The first impulse was given to it by the two brothers Methodius and Cyrillus (Constantine), who came from Constantinople as Christian missionaries to labour among the Slavonians, as Ulfilas had done four or five centuries before among the Goths. They made an alphabet founded upon the Greek character, which, with some modifications, is still employed in Russia and Servia. They translated the Gospels into the language of the people among whom they dwelt, who were probably the Bulgarians. The language is therefore sometimes called Old Bulgarian as well as Old Slavic and Church Slavic. The last name is employed in consequence of the language in which the missionaries made their trans- lation being still employed in the services of the Greek Church ; so that, like the Latin in the Roman Catholic Church, it has acquired an ecclesiastical character. Nes- torius in the eleventh century also wrote his Chronicles of Russia in the same language. No less than fifteen languages are enumerated as be- longing to this class. Several of them, however, have had but little literary development. Those which contain the most important literature are — 1, the Bohemian, whose remains begin with a collection of national poems be- longing to the thirteenth century ; 2, the Polish, which begins with the Psalter of Florian, belonging to the thirteenth and fourteenth centuries ; and 3, the Russian, which has made rapid progress since the time of Peter the Great, and now possesses an extensive literature, especially upon scientific subjects. 4. GRyECO-ITALIC LANGUAGES 19 The Old or Church Slavic is the most serviceable of this branch for the purpose of Comparative Grammar, but this entire class is not quoted in the following work because the Sanskrit and Zend elucidate the points to which it would apply, sufficiently for our purpose. 4. GE^CO-ITALIC. 10. The fourth class is the Grceco-Italic, spoken principally in the two great peninsulas which run south- wards into the Mediterranean, and in the neighbouring islands. That neither the Greek nor the Latin owes its origin to the other, but that they are related together as sister-languages becomes obvious on an examination of their grammatical forms and the roots of words. Their resemblance to one another is somewhat disguised by their being written in different characters, and by the circumstance that different means are resorted to in expressing the analogous changes which time has pro- duced in both. The defective analysis of former years caused the Latin to be spoken of as derived from the Greek, whereas recent investigations have shown that, in some instances at least, the Latin has older forms than the Greek. Such, for example, is the preservation of the ablative singular which is wanting in Greek, and the fuller ending -has in the dative plural. Greek, on the other hand, undoubtedly has many older forms than the corresponding Latin ones. Evidence also supports the theory that the Italic tribes did not enter Italy by sea from Greece, but by land from the north. The Hellenic and Italic races seem to have parted company in the neighbourhood of the Danube and to have taken a southerly direction previously to the approach of the Slavic race, which we have seen in the same region. c 2 20 INTRODUCTION The first three classes probably left the original abode of the family at the same time, the Indian taking a south-eastern direction, the Iranian a south-western, whilst the Slavic went directly westward. We now come to those classes which, from their more westerly position and from the greater deviation of their lan- guages from the original type, seem to have left the common home at an earlier time. Their history also exhibits them in a more adventurous and victorious character, which agrees with the idea that they were the first to break off the associations of home and dare to invade distant lands. 11a 1. The Greek language claims-.our first attention from its occupying a more easterly position and pos- sessing an older literature than the Latin. It flourished principally in the eastern peninsula of the Mediterranean, in the islands of the Archipelago, and on the western coast of Asia Minor. It there produced the most perfect literature in the world — a literature which continues greatly to influence the opinions, the character, and the institutions of all the members of the Indo-European family throughout Europe and America. The variety with which Greek was spoken as a living language is illustrated in three principal dialects, the JEolic, Doric, and Ionic. The last differs considerably from the other two. This difference is illustrated in the national legend by iEolus and Dorus being repre- sented as sons and Ion as a grandson of Hellen. The Ionic grew up in a foreign land, Asia Minor ; a fact represented in the legend by the name of Ion's father, Xuthus, meaning c banished.' The Ionic far surpasses the other dialects in the abundance and perfection of its literary remains. The oldest form in which it appears is the Epic dialect of 4. GIUECOITALIC LANGUAGES 21 Homer and Hesiod, whose date is variously fixed from the twelfth to the ninth century b. c. It next appears in the New Ionic of Herodotus, which belongs to the middle of the fifth century. But its richest productions are in the Attic dialect, brought to marvellous perfection by dramatists, historians, philosophers and orators, from the middle of the fifth to the middle of the fourth centuries. The JSolic was spoken chiefly in Asia Minor, Boeotia and Thessaly, and includes the celebrated names of Alcaeus, Sappho and Corinna. The range of its litera- ture is very limitecL but it preserves some very old forms in its grammatical construction, and has a special inte- rest from its close resemblance in several points to the Latin language. The Doric was spoken chiefly in the north of Greece, in the Peloponnese, in Crete, and in Sicily. Its princi- pal representatives are Pindar and Theocritus. The Greek furnishes us with striking examples illus- trative of the effect accomplished in a certain length of time by the influences which are continually producing phonetic decay in living languages. The language of Homer may be regarded as five hundred years later than that of the Vedas ; and this difference of time corre- sponds with the difference of form in the one language as compared with the other. For instance, the genitive case singular of the a stems in Sanskrit ends in a-sya. In Homer it is o-io 9 in which we see that every element has undergone a change ; for a the lighter vowel o is substituted in both cases, for y the vowel i 9 and s is lost altogether. Five hundred years later, again, the Attic dialect presents to us the same form reduced to ou ; that is, oo is reduced to the weaker form ou, and i is lost altogether. These changes are not sporadic instances 22 INTRODUCTION which might be owing to accident. They affect the entire mass of the language to which they belong, and rest upon general principles. There is no example of the older a-sya in the language of Homer, nor any instance of the Homeric o-io in the Attic dialect. These effects appear as if they were accomplished at once and therefore arti- ficially, but this only results from our not possessing literary records during the time which intervened be- tween these epochs to illustrate the gradual approaches towards the final result. Any one will see how gradu- ally such modifications are effected who examines the change which is going on from th to s, from hath to has, for instance, in the third person singular of the present tense in English verbs. It is long since it began, and it is not yet completed ; but if when hath is exterminated and has is universal, all the intervening literature be- tween the first employment of s and the last use of th were to disappear, we should have an instance similar to those noticed above. 12. 2. The Italic branch of this class of languages belongs almost exclusively to the western peninsula of the Mediterranean. The Latin tribe gradually gained the upper hand in the political constitution of the coun- try, and the Latin language became the organ of public life, of education, and of literature. It was not, how- ever, the only language spoken by the Indo-European population who entered the peninsula from the north. It is evident that before their arrival, or soon after their settlement, their speech was marked by varieties as dis- tinct as the Greek dialects, and though only one has become embodied in a considerable literature, yet some important remains belonging to others have recently been discovered. The science of language has thrown considerable 4. GR^ICO-ITALIC LANGUAGES 23 light upon the character of the ancient population of Italy. It seems to have been invaded in succession by very different races. Some of them probably were not Indo-Europeans. But several tribes of which remains have been preserved evidently belong to the same family as the Latin. In the extreme south-east of the country inscriptions have been discovered composed in a lan- guage which, for want of a better name, has been called lapygian. It appears at one time to have prevailed more or less throughout Apulia and Calabria. The remains of this language have not yet been sufficiently deciphered to determine the exact ethnological position of the people by whom it was spoken, and they appear to have presented but little if any resistance to the superior civilisation of Greece; for Apulia, which is spoken of in the time of Timseus the historian (400 A.u.c.) as inhabited by barbarous Iapygians, in less than two centuries appears to be an entirely Greek district. Clearer evidence is supplied of the relation of two other peoples, or branches of the same people, who early occupied the middle of the peninsula, i.e. the Latin and the Umhrian branches. The latter, including the Marsians and Samnites in the south, comprised a considerable population. The dialects which prevailed amongst them have a close resemblance to one another, but in many points they form a contrast to the Latin. Distinctions appear which are also found in other classes of the Indo-European languages. Thus, where the Roman employed q, as in the interrogative pro- noun, the Samnite and Umbrian employed jo, sounds which also distinguish the Ionic from the Attic dialects in Greek, and the Celtic languages in Lretagne and Wales from the Gaelic and Irish. The Latin language 24 INTKODUCTION has, upon the whole, some such relation to the Umbro- Samnite, as the Ionic has to the Doric, whilst the va- rieties in the Oscan and Umbrian, as well as other dialects related to them, in distinction from the Latin ? are similar to those of the Doric in Sicily and in Sparta. Of all the Italic languages, the Latin only has fur- nished us with any considerable literature, and from it, therefore, our examples for this branch of languages • will be principally derived. It presents no great lite- rary works in a perfect state earlier than the second century, b. c. What precedes that period is fragmen- tary or only brief. These older remains exhibit some archaic forms which are of great value. But Soman literature is several centuries later than that of Greece,, which, of itself, is sufficient to account for the termi- nations of Latin words being, upon the whole, much more curtailed than the corresponding Greek ones. Thus the genitive singular, which appears in Sanskrit as a-sya, in Homer as o-io, and in Attic Greek as ou (from od), is in Latin, two or three centuries later, reduced to i (from oi). The classic literature of the Latin language is not only of a later date, but is also far inferior in extent and variety to the literature of Greece. The style of the two differs materially. The Latin is distinguished rather by a sonorous majesty and exactness of expression than by the graceful elegance and endless versatility of her Grecian sister. But their close relationship to one another is, neverthe- less, undeniable. This is rendered evident by an ex- amination of their grammatical structure, and it would be unnecessary to quote two languages so nearly allied in order to illustrate Comparative Grammar, but for the circumstance that they supply each other's defects in a 5. CELTIC LANGUAGES 25 remarkable manner. Thus, for instance, the letters s, iv, y 9 are generally either lost or disguised in certain positions in Greek, but are more or less fully preserved in Latin. On the other hand, the aspirates, which are numerously represented in Greek, are generally lost or altered in Latin. Again, the Greek preserves short vowels when final, but the Latin drops them, whilst final consonants, lost in Greek, are preserved in Latin. So, also, the Greek distinctly preserves the important aorist forms, but has greatly obscured the reflexive pronoun in verbs, whilst in Latin the latter is unmis- takable and the former almost absent. 5. CELTIC. 13> The fifth class consists of the Celtic languages, This name appears, variously modified, in application to the Galatians of Asia Minor, the Gauls (Galli) of northern Italy and France, the Celt-lb eri of Spain, and the Gael of Scotland. The earliest notices of the people represent them as occupying a considerable part of the south-west of Europe and the British Isles, but we have no historical evidence to connect them with the original abode of the family in Asia. The settlement in Asia Minor appears to have been the result of a migration eastward from Europe. Bohemia owes its name to them, for the Boii were one of their tribes. The Celtic languages are now spoken in Ire- land, the Isle of Man, the Highlands of Scotland, Wales, and Bretagne. Some remains also are preserved of the language of Cornwall and of the ancient Gauls. The principal literature of the Celts belongs to a recent period, and the inflections of the language have, conse- quently, become greatly abridged. No doubt, however, 26 INTRODUCTION remains of the fact that the Celtic belongs to the Indo- European family, and that it is very nearly related to I the Italic languages. The effect which time has had in abridging or destroying the grammatical inflections will be seen on comparing the Irish ech, ' horse,' with the Latin equ-us, and the Sanskrit aziv-as ; ir. deich, f ten,' I. dec-em, s. da'z-an. The Celtic languages are also subject to some pecu- liar phonetic changes. One of them is the ' assimila- tion of vowels,' by which a weak or strong vowel causes the insertion of one of its own class in the preceding syllable if there be not one there already, as in echaire, echire (' mulio,' ' muleteer '), which requires for its full explanation the help of the corresponding Latin word, equarius, ' groom.' The % after r has caused i to be inserted before r, which then ultimately causes a to disappear. A similar influence is exerted in other cases upon a subsequent syllable ; for example, the a of labra in labra-tar (1. loquu-n-tur) requires the ending to be tar, whilst the i of labri in labri-tir (1. loqua-n-tuv) requires tir. Another striking peculiarity is the ' aspiration ' of initial consonants, that is, h is added to an initial con- sonant in connection with certain changes of inflec- tion or derivation, thus, gaiv is f voice,' and fo-ghur, 6 sound.' These and other peculiarities render it difficult to introduce the Celtic languages into a work like the present. The necessary explanations would greatly increase its bulk, and the peculiar phenomena of this class of languages would render the subject more com- plex than is desirable for an elementary work. 6. TEUTONIC LANGUAGES 27 6. TEUTONIC. 14- The sixth class is the Teutonic, including three principal branches, the Gothic, the Low German, and the Scandinavian. It is distinguished from the Ger- manic, which comprises the High German language. These terms are employed not as being scientifically accurate, for that would be difficult at the present stage of the subject, but as being intelligible and definite. The sixth and seventh classes form the second and third divisions mentioned in section 5, and are distin- guished from the classes of languages previously enume- rated by a remarkable feature. The operation of what is called Grimm's Law of Consonant Changes separates the sixth and seventh classes from the other five, as well as from one another. At present the cause of these changes is hidden. Why should the sounds k, t, p have been changed by the Teutons for the aspirates kh (ch, gh), th (dh), ph (/, v) ? Not from any preference for aspirates, for the original aspirates at the same time were changed to g, d, b. Nor are these latter special favourites, for in their turn they have to give place to &, t, p. All the languages of the sixth class have undergone this change, though in many respects they differ greatly from each other, and many of them, as long as history has known them, have been entirely independent one of the other. Of course, however, the presumption is that when the change was made they all formed but one language. We will take the following instance, illustrating the effect upon letters belonging to both the labial and dental organs : the Icelandic word fotr, Swedish fot 9 Gothic fotus, Anglo- 28 INTRODUCTION Saxon fot, English /oof, Old Saxon fot, Friesian fot, Dutch voet, all have the aspirate (/, v) for the first consonant, and the mute (t) for the last, the final r and 5 in two cases being the nominative sign. Now the classes of languages which we have considered before, agree in having a mute (p) for the first consonant, and a vocal (d) for the last, e.g. Sanskrit pad-as, Greek po us for pods, gen. pod-os, Latin, pes for peds, gen. ped-is. It is, moreover, singular, that from the sixth class, as a starting point, a perfectly analogous change is made in producing the seventh class, or the High German language. Thus the word mentioned above is in High Grerman fuss, which does not preserve the final t of the sixth class, nor return to the original d of the other classes, but changes the mute into an aspirate. The aspirate, however, has become s in modern Grerman generally, as it has in modern English, in the third person singular present of verbs, e.g. has for hath. It was also the practice to preserve the characteristic t of the sixth class, when final, thus producing ts ; hence, by assimilation is formed ss. In other cases, the compound is written z and pronounced ts. Initial aspirates appear to have resisted the change, and, therefore,/ remains in the above word instead of becoming b. But the regular change is seen in the word leib, as compared with the English, loaf. The time when these changes took place cannot be exactly determined, but Grimm identifies the Goths and Geice, and the latter, as far as their language is known to us, seem to have retained the original consonants. Indeed, the names indicate the difference, for Getse has the mute (t) in place of which Goths has the aspirate (th). The consonant changes generally, therefore, 6. TEUTONIC LANGUAGES 29 probably accompanied this change of name, and took place as the Gretae disappeared, and the Groths acquired importance, that is, a short time before the beginning of the Christian era. Not only the Gretae, but also the Thracians, are identified by Grimm as belonging to the same race of people with the Groths. We have thus the means of tracing them to an early period, and to a remote easterly position. The Thracians are mentioned by Homer, and are described by Herodotus as being more numerous than all others except the Indians. He mentions also that Darius encountered the Getae on his march to Scythia. We thus find evidence of the presence of the Teutons in considerable numbers on the north-west of the Black Sea, at a time when a large body of the Slavonians were north of the Caspian, and the Celts were in the west of Europe, whilst the Graeco- Italic race already possessed the north coast of the Medi- terranean. The order in which they entered Europe appears, therefore, to have been the following : 1. The Celts; 2. The Graeco-Itali : 3. The Teutons and Ger- mans ; 4. The Slavonians. 15. 1. Gothic. A remarkable passage in Strabo con- firms the view taken by Grimm as to the relation of the Gretae, Thracians, and Groths ; for he states that in the time of Augustus, 'iElius Catus brought from beyond the Ister (Danube) 50,000 persons of the Gretae, a people speaking the same language with the Thracians, into Thrace, and they now live there under the name of Moesians.' Now, it was for the inhabitants of Moesia, that Ulfilas, 300 years later, made his trans- lation of the Scriptures, which is hence called the Moeso-Grothie. This work, though preserved only in fragments, is, in some respects, the most important of 30 INTRODUCTION all the literary remains s the Teutons for the purposes of Comparative Grammar. It belongs to the years 360 — 380, A.D., but preserves many traces of the lan- guage of a much earlier period especially in the use of the short vowels, (B) pleasure tr sh sh shed ¥ s s said ^ h h head __ 1 11 (in Welsh) cto Remarks. 18. In the examples given a vowel is to be considered short when not marked long ; e. g., in rajan, the first vowel is long and the second short. The pronunciation of the letters is that given under ' power.' It will be seen that the vowels are employed rather with the Continental than the English sounds. D 2 36 ALPHABETS Judging from modern pronunciation in India, the Sanskrit a seems to have had the sound of u in but, or a in the second syllable of readable. In pronouncing the aspirates it will be seen by the examples that the aspira- tion (h) must be sounded separately from the mutes after which it is written. I have not attempted to supply examples of the pronunciation of the fourth class of consonants, because the sound of them is so utterly different from anything in our own language that it seems impossible to convey a notion of it by written characters. The peculiar modification of sound is produced by bending the tongue upwards and as far back as possible. A hollow sound is then produced, which seems as if it proceeded from the upper part of the head. The letters are, therefore, in Sanskrit gram- mars, called murdhanya (capitalis), from murdhan, ' head.' By Bopp the term ' cerebral ' is employed, as being of similar meaning. 19. It is supposed that the palatal consonants, ch,j 9 acquired the sounds thus indicated at a later period, and that there intervened between the pure guttural pronunciation, k, g, from which they sprang, and the palatal pronunciation, such sounds as those of k and g in the English words kind, guard, that is, ky and gy. See Schleicher, Compendium, pp. 13, 14. THE ZEND ALPHABET 37 b) THE ZEND ALPHABET. 20. The Zend Alphabet consists of the following characters : Characters Example. had far apres (Fr.) tres (Fr.) made hid heed hood food hod bode bake bakehouse log loghouse ring reach judge right right hand red red hand Form. Characters employed in the following pages. Power. A5 a a AXJ A a A a 8 e e A e e A a a i i u 00 9 u A 00 \ A o A 3 k k ctf kh kh «? qh qh e g g 2. n g gh ng f ch ch iU J J t th t th d dh d dh 38 ALPHABETS Form. Characters employed in the following pages. Power. Example. / n n then «) P P loop \ f f roof _i b b job 9 m m room 61 initial 1 medial f y y yea 7 r r ray- » initial 1 medial j w w way oxf after th w, V W, V way, var s z z haze US i z *(») pleasure ^ sh sh shed ■*o s s said w h h head V d d, dh (th) then Remarks* qh represents a deep guttural sound, somewhat rougher than the German ch. Of the two characters for ng, the second is only employed after i or e. Further remarks upon the Zend alphabet will be found in III. THE GREEK ALPHABET 39 €) THE GREEK ALPHABET. 21. The Greek Alphabet consists of the following characters : Form. Large. Small. Power. Example A a a a had, made B /3 b bad r y g gate A 2 d red E t e led z I z haze H V e here e 9 th thin i i • A 1 1 hid, hide K K k take A X 1 lay M n m room N V n> ng then, ring I x(ks) box hod n TT P loop p P r ray s a c s said T T t right Y V u, u bud, bude $ f f roof X X k(kh) take ¥ * ps lips Q hi 6 abode 40 ALPHABETS d) THE LATIN ALPHABET. 22. The Latin Alphabet^ consists of the following characters : Large: ABCDEFGHIJKLMNOPQES TU VXYZ. Small :abcdefghijklmnopqrstuvxyz 23. They are pronounced by us as in the English language. It must, however, be borne in mind that the Eomans probably uttered the vowels with the Continental and not the English sounds of those letters, j repre- sents the half-vowel y, and was doubtless pronounced like y in yea ; whereas the Latin y> being in fact the Greek i/, is always a vowel, and was probably pro- nounced something like the French u in une. The Latin u, in some positions, stands for the half-vowel w 9 as, for instance, in sanguis, where it is pronounced just as in the English word sanguine. In c and g, only the sounds in cot and got should be employed for the ex- amples adduced in the following pages. The soft sounds of these consonants were of later introduction, and are of less service for the purpose of comparison with the other languages here treated of. e) THE GOTHIC ALPHABET. 24. The Grothic Alphabet consists of the following characters : Form. Power. Example A a had B b bad r g log THE GOTHIC ALPHABET 41 Form. Power. Example. <\ d red 6 A a made l= f roof Q h7 judge, yea h h head II i hid 61 ee heed K k bake A 1 lay M m room N n then X 6 b6de n P loop O hw when (hwen) K r ray s s said T t right th then, thin n u hood u cw quantity (cwantity) V w way X kh (ch) (ger.) lachen z z haze Al A e a led, made JYfl ° ° n °d> bode IT (gg) ng ring 42 ALPHABETS Remarks. 25. The characters employed are the same as those which express the powers of the letters in the above table, except that y is represented by j 9 w by v, and ee by t or ei. The same character was probably pro- nounced w in some instances, and v in others. 26. The pronunciation of the Anglo-Saxon letters, especially of the vowels, is very uncertain, but the following list is sufficient for comparing the grammatical forms and roots with other languages. /) THE ANGLO-SAXON ALPHABET. 27. The Anglo-Saxon Alphabet consists of the fol- lowing characters : Form. Power. Example 3E a had B b bad L ch chide D d red e e red F f roof n J judge V h head i i hid K k work L 1 lay CD m room N n then O hod THE ANGLO-SAXON ALPHABET 43 Form. Power. Exampl P P loop R r ray 8 s said T t right D dh (th) then P th thin U 00 hood b w way X X box Y 7 yea Z z haze 44 SOUNDS III. SOUNDS. 28- The articulate sounds which can be produced by the human voice are very numerous, and merge one in the other by almost imperceptible degrees. In the original construction, however, of the Indo-European languages, only a limited number of these sounds were employed, and they are such as are clearly distinguish- able from one another. The system on which they are arranged is remarkably simple, and they may be traced with surprising distinctness through a great variety of languages during a period of more than three thousand years. 29. On comparing the Alphabets now in use with those of the oldest Indo-European languages, we find that the vowels have undergone greater modifications than the consonants. They are of a feebler construc- tion, and less able to resist the violence of impetuous utterance or overcome the hindrance occasioned by climate etc. in the organs of speech. Hence, in some countries they preserve a free open sound, but in others are compressed and indistinct. In one language they are few and simple, in another numerous and difficult to distinguish. There was probably, at first, only one vowel sound, and this, being considered the natural accompaniment of the consonant, had no written character assigned to it. Before writing came into use it doubtless acquired SANSKRIT VOWELS 45 some variety of pronunciation, and as speech came to be fixed in written forms, it was increasingly found necessary to adopt a corresponding variety of characters to represent these modifications in the vowel sound. In course of time more complex sounds were formed by combining and contracting together the simpler ones already in use. 30. The original vowel sound is that represented by a in had. In organic formation it corresponds to the guttural consonants, being a simple sound emitted from the throat. The first modification to which it was sub- ject was probably the development of i as in hid, which corresponds in organic formation to the dental con- sonants. There was then produced u, sounded as oo in hood, which corresponds in organic formation to the labial consonants. 31. These three sounds we find represented in ancient as well as modern alphabets. The sounds, however, represented in different languages by the same letter, are not identical. The a, for instance, has a perceptible difference in German, Italian, and English pronun- ciation. This has probably resulted from varieties of social habits, climate, etc., after the various races had separated. New modifications afterwards arose, which must be considered in connection with the individual alphabets. 1. SANSKEIT. a) VOWELS. 32. In Sanskrit the original vowel a combines with the derived ones i and u, making with the former e, as in neigh, and, with the latter, o, as in nd. These, as simple sounds, are still represented in the French language by the original letters, ai and au, as 46 . SOUNDS in mais, rnaux; whilst their originally diphthongal character is indicated not only by this circumstance, but also by the fact that in Greek they appear not as s and o, nor as 77 and co, but as s 1 and sv, or as 01 and ov. By prefixing another a to e and 0, the diphthongs ai and an are formed, having the vowels a-i 9 a-u, pro- nounced separately, with the principal stress on the first vowel, and resembling i and ou in hide and loud. The short vowels e and seem not to have been part of the primitive alphabet, but to be later modifications of the original sound, for they are wanting alike in the Sanskrit and Gothic alphabets, and appear in Greek and Latin as representatives of the Sanskrit a. For example : Sanskrit Ushtau Greek o«rw Latin octo ' eight ' nsLvan kvvea novem 'nine' navsis vsog novus 'new 33. The following three characters also occur in Sans- krit. 1. Anuswara. It is represented by a point over the preceding letter, and is pronounced like the final n in French. At the end of a word it stands in place of an original m, and in the middle of words in place of an original n before sibilants, e.g. ta sunum for tarn sunum, ' the son ; ' hdsa for hansa, ' goose.' 2. Anu- ndsika. It is represented by a point over a curve above the preceding letter, and denotes that a final nasal has been assimilated to I, y, or w, at the beginning of the following word, e. g. pakshal lundti for pakshdn lunati, ' cuts off the wings.' * The Lithuanian and Old Slavic retain the sign of a nasal in the preceding vowel, the former without, the latter with, the corresponding sound. 3. Visarga. It is represented by two points at * Bopp, Kritische Gram. 2nd ed. 66, 70. SANSKRIT VOWELS 47 the end of a word, and stands in place of final $ or r, which then is pronounced as a soft aspiration, e.g. puna: for punar, ' again;' dwikha for duskha, 'pain.' The modifications of sound represented by these three characters are euphonic, and are generally occasioned by the consonants which follow them. 34. r and 1, as vowel sounds, seem to have grown out of the syllables ar and al (cf. -Greek a-fyepros with Sanskrit bhrtas, and Latin art for cart with Sanskrit krtis). 1 occurs only in one root, kip for kalp. 35. We may here introduce a reference to the Com- parative Weight of Vowel Sounds. In his second edition Bopp has given a detailed account and fuller illustrations of this subject. Of the original vowels a is the heaviest, i the lightest, and u intermediate. The principle on which this point is determined is that when a preceding or following syllable, on account of its own weight, requires the one next to it to be lightened, the vowel introduced for this purpose is considered lighter than the one for which it is substituted. For instance, the termination of the first person plural in verbs, -mas, is heavier than the termination of the first person singular, -mi ; therefore the syllable preceding the former must be lighter than the syllable preceding the latter. Now the last syllable but one inyunafmi, 'I bind,' contains a, whilst the corresponding syllable in yuni'rnas contains i, Hence i is lighter than d. Similarly, in Q'mi, 6 1 go,' and imas, 6 we go,' i is lighter than e ; in etfu and 'l/jlsv, i is lighter than si. A similar result is occasioned in Latin by prefixing prepositions, or by reduplication. Thus, jacio, 6 1 throw,' becomes abjicio, ' I throw off,' and cano, ' I sing,' becomes cecini, i I sang.' i is therefore lighter than a. The concurrence of two consonants has the 48 SOUNDS effect of partly (abjectus, e thrown off ') or wholly (con-* tSLctus, ' touched') preventing the lightening of the root vowel. The same phenomenon appears in the past and present tenses in the strong verbs of the Germanic languages. In Grothic the heavier terminations of the present tense are affixed to the lighter stem, as sU 9 whilst the lighter terminations of the past tense are connected with the heavier stem sat In English, where the inflections which originally caused the difference have disappeared, the distinction itself is retained, as in sit, sat Another illustration is derived from a com- parison of older and later forms of languages. The effect of time is to render grammatical forms lighter and easier of enunciation.* Hence the older Sanskrit a (dsidami, 'I give') becomes i in the later Greek (8/8o>/m). We have a in the older Sanskrit ch&twa'ras, 'four,' and i in the later Greek irlavpss, and G-othic fidvor. The Latin quatuor is in this respect of an older character than the Greek. From similar facts it is inferred that u is lighter than a, e. g. s. karomi, ' I make,' kurmas, ' we make ; ' 1. cSilco, ' I tread upon,' and conculco, 6 1 crush ; ' e. came and come. In this last word the older pronunciation has now become pro- vincial. Coom has changed to cum, for in come, though o is written, the u sound is preserved. For older s. ndktam, ' night,' we have later gr. vvfcra. That i is lighter than u appears from such cases as, s. aundidam, a later form for aundudam, and 1. fructifer, ' bearing fruit,' for fructufer. b) CONSONANTS. 36. Sanskrit grammarians arrange the consonants in five classes ; an arrangement which it will be convenient * See Diversions of Purley. SANSKRIT CONSONANTS " 49 to follow here, though we shall afterwards see that a more accurate division for the purposes of Comparative Grammar would be into three principal and two subor- dinate classes. There are also four half-vowels, three sibilants, and h. This arrangement is determined by the organs used in pronunciation. Each of the Five Classes contains five letters — a mute, a vocal, the cor- responding aspirate of each, and a nasal. In the aspi- rates the aspiration is pronounced distinctly from the letter to which it belongs ; e. g. the aspirated p is pro- nounced as ph in haphazard, and the aspirated b as bh in abhor* The later substitution of a single sound and character, as / and v, is analogous to the change of the diphthongs a + i and a + u to e and 6. 37. The first class of consonants, the Gutturals, contains the letters &, kh, g, gh, ng. They occur in most of the cognate languages with great regularity, and require little elucidation. The aspirates, however, especially Jch, are rare. Yet the following examples are met with : « nail ' s. natelia gr. ow'V' 1. 'light' laglius IXa'Vvg levis 1 warm ' glxarmds ^repfir] (uro) In the last example there is a change of aspirates, the Dental $ (for dh) being used in Greek for the Guttural gh in Sanskrit. That the Guttural was the original appears likely from the evidences of it in other languages. In Latin, although the Guttural is lost from uro, as it is in many other Latin words, yet the xi indicates its previous presence. The same may be said of the w in the Germanic languages, e.g. go. varm> a. s. wearm, e. warn, ger. warm. For these letters are E 50 SOUNDS usually developed from a Guttural, e. g. s. ka-s, 1. qui, e. who (for hwo). In the second example above there is another variation of aspirates, v (for bh) being used in the Latin for the Sanskrit gift. Indeed, we shall find that such variations in the aspirates are not un- frequent. 38. The second class, the Palatals, contains the let- ters ch, chh, j,jh, n. A modification in the sound is pro- duced by pressing the tongue against the palate, whence the class obtains its name. These letters are derived from the first class, and are used only before vowels, half-vowels, and nasals. They were probably originated after the separation of languages, which would account for their being entirely absent from some alphabets. The following examples illustrate their absence from the Greek and Latin alphabets : four ' s. c^atwaras gr. 7Ti(Tvpeg 1. qmtuoT cut' chh id th, d, dh, n. The dh is represented in Greek by #, whilst th appears as t, from which it is probable that # represents the former as well as the latter sound. The Latin sometimes omits the aspiration, and sometimes substitutes the aspirated letters of other organs. These letters are often inserted inorganically in Greek after similar letters of another class, e. g. 7tt6Xl9 for 7ro\^, and X&ss f° r X s9 ' Instead of d in Sanskrit, we often find in other languages I or r. Examples : s. aiiumds ma&Tiya pun Jiyas CLah 40. The fourth class, the Cerebrals, contains the let- ters t, thy d, dh, n. They are derived from the Dentals, and are similar in sound, except that in pronouncing them the tongue touches the roof of the mouth, point- ing perpendicularly towards the brain, from which they take their name. They seldom begin a word, n never; and they rarely appear in the cognate languages. They are substituted for Dentals after sh, and are very common in Pracrit, probably as the result of local influences. E 2 gr. ^JVfiog 1. fumus * smoke ' — f melius \ meridies 1 middle ' 1 mid-day ' ttToXiq 1 city ' x&k heri 4 yesterday ? Tiiyvvg lignum 1 firewood ' 52 SOUNDS 41. The fifth class, the Labials, contains the letters p, ph, by bh, m ; ph rarely occurs, bh is represented in Greek by or /3, and in Latin by / or b. The Greek may have represented both aspirates in sound. Final m generally becomes Anuswdra in Sanskrit, is preserved unchanged in Latin, and appears as v in Greek; in Gothic it is sometimes m and sometimes n. Examples : s. -pliena g r - 1. lahli eXapov -pdnchan 7Tsvts quinque 'five' hliaran Q'epujv ferens ' bearing ' dewdm Sr'soV deum. 'god' As Labials are used in Greek for Sanskrit G-utturals, so we have the converse in the Latin gidnque, i. e. a Gmttural qu for a Sanskrit Labial p. 42. The Half- Vowels are y, I, r, w. They are more frequently interchanged with one another than the conso- nants, and are therefore less easily traceable in the sister languages. This is more especially the case with I and r 9 which correspond to the class of Dental consonants, and differ from each other only in the circumstance that in pronouncing I the tongue touches the roof of the mouth once only, while it does so more than once in pronouncing r. Hence only one of them is found in some languages, y corresponds to the G-uttural conso- nants in organic formation. In Pracrit it has become j, as it has also in the transition from Latin to Italian ; e. g. in 1. jugum, { yoke,' the j answers to English y, and in it. giogo, gi answers to English j. In Greek it be- comes £ or c , or is assimilated to a preceding letter. V) corresponds, in organic formation, to the Labial con- SANSKRIT CONSONANTS 53 sonants. In Pracrit it is often assimilated to a preced- ing consonant. In Sanskrit it seems to have become v, except after a consonant. In Greek it is either assimi- lated or changed to a vowel, or appears as (from the digamma), or is lost. In Latin it is changed to a vowel or is lost. It assumes the character of a mute Guttural in 1. victum, factum, and in e. quick. The following list will serve to illustrate most of the above state- ments : 1 shine p s. ruk gr. Titvicog 1. luceo * leave ' rih TutVw linquo 4 another ' anyds dXTiog alius * join ' TV tpeuyvvjJLi ^un go ' liver ? ydkrt ■qirap jecur 1 four ' chatwd'ras TiG&aptg quatnor 4 horse •' azvsa 'iTTlTog eqnus * thou ■ twam tVL 'his' &was ctyog smis 'sail' pldvami ir\s(x> (7r\tFu)) 1 sheep ' avis v'ig ovis 43- The Sibilants are & 3 s, sh. z corresponds, in organic formation, to the Guttural consonants, and has, in fact, in almost all cases sprung from an original Jc. Hence it is generally represented by a Guttural in the sister languages. In Lithuanian, however, it appears as sh, and in Slavic as s. Its sound resembles that of s in the word pleasure, s corresponds, in organic forma- tion, to the Dental consonants. It seldom remains un- changed at the end of a word. Final as followed by a word beginning with a vocal letter becomes o ; probably through the changes as, ar, au, o. Instead of this 54 SOUNDS letter we frequently find r in other languages, sh cor- responds in organic formation to the Labial consonants. It takes the place of s in certain cases where the latter is not admissible in Sanskrit. 44. h is a feeble aspirate. It appears in Greek as %, and in Latin as h. Sometimes it is only the aspiration of another letter which has been dropped, as han for dhan. The above statements are illustrated in the fol- lowing list of words : 'dog' s. iwan gr. Huwv 1. canis 'ten' daxan dsKa decern 'right' da.ksh.inas ds£i6g dexter for 6tK(TlOQ for dec-ster ' winter ' Ydmam X 6 ~ L i ia litems 'kill' l&an- trJ avov funus In the Latin word funus the Labial aspirate bh (f) is used instead of the Dental aspirate dh. 45. The following table exhibits the relation of the letters in the Sanskrit alphabet. The larger characters represent the three older classes, viz. Gutturals, Dentals, Labials ; the smaller characters represent the two later derived classes, viz. Palatals and Cerebrals. Mute. Vocal. Nasal. Half-vowels. Sibilants. Vowels Gutturals k kh ggh ng y Z a Palatals ch chh J jb n — 00 e ai Dentals t th ddh n lr s l Cerebrals t th d dh n It 6 au Labials p ph bbh m w sh u ZEND VOWELS 55 2. ZEND. 46. The Zend has a close relation to the Sanskrit, and from its grammatical forms, as well as from remains of the two subordinate classes of consonants, it may be supposed to have been separated from Sanskrit at a later stage of development than was the case with Greek and Latin. The same conclusion is also supported by the local proximity of the peoples by whom Sanskrit and Zend were spoken. a) VOWELS. 47. a, and in some cases, through the influence of particular consonants, e (resembling the French e in apres), occupy the place of a in Sanskrit, a, i 9 i, u, u, generally answer to the same letters in Sanskrit. Some- times, however, 6 occurs for Sanskrit u, as in kerenabt for s. akrnaut, which is written akrnot. This form probably arose when the original diphthong was pro- nounced a + u, not 6. The vowel 6 or the diphthong eu answers to the Sanskrit 6, eu having arisen probably out of the original au. Generally do, but sometimes also du, stand in place of the Sanskrit au. Thus gdus for s. gaus. Generally Sanskrit e is represented by e, but by 6i after y or before final 8 or d, e. g. yoi for s. ye, ' who.' At the beginning and in the middle of words the combination ae occurs, which Bopp regards as containing a gunaed form of i in the original di- phthong ai. Thus a + e would be for a + a + i, and this instead of a+i or e. The sound, however, remains the same as that of the original diphthong. Thus there occur in the same word rathive and rathvjaecha. 56 SOUNDS b) CONSONANTS. 48. Of the GtUTTUKALs, Sanskrit k remains k before vowels and v 9 but becomes kh in other cases. Sans, kh remains unchanged, qh is in sound nearly the same as kh, but of different origin. It has grown out of qw 9 and answers to Sans, sw 9 e.g. qha-dhdta = s. swa-dhata, ' self-made ;' qhafna = s. swapna, ' sleep.' g and gh correspond to the same Sanskrit letters. 49. The Palatals have in some cases become sibilants, but there remain the two unaspirated Palatals ch and^'. 50. The Dentals are t 9 th 9 d 9 dh 9 &n& correspond to the Sanskrit letters of the same organ, t 9 however, sometimes stands in place of Sans, th or th, e. g. sta for s. sthd ; ishta for s. ishfha. The aspiration in such cases seems to be prevented by the preceding sibilant. At the beginning of words d sometimes stands for s. dh 9 e. g. da for s. dha, gr. $77. 51. Sanskrit Ceeebeals are represented by Dentals in Zend. On the other hand, d is used for Sans, t at the end of words and before case-endings beginning with b. 52. The Labial p corresponds to Sans, p, except that when immediately followed by r, s, or n 9 it becomes /, e. g. ira for s. pro, gr. 777*0. / corresponds in other cases to Sans. ph 9 and b to Sans. b. 53. The Half- Vowels are y 9 r 9 iv. y is represented by the three characters given in the table, of which the last shows by its form that it has arisen out of the vowel i. y and i as well as e have an assimilating power, which causes the insertion of an additional i in the preceding syllable, e. g. maidhya = s. madhya, 6 middle ;' tuirya = s. tv!rya 9 ( fourth.' The combina- ZEND CONSONANTS 57 tion of two consonants, however (except nt), and some- times a single consonant, prevents this assimilation from taking place. Hence we have azti and not aizti for s. asti, ' is.' Bopp also ascribes a like assimilating power to y over a succeeding syllable, in changing a to e (= ai). In all the cases adduced, however, there is an i in the syllable which follows the e, and this will account for the change without introducing any new principle. 54. r is usually followed by e 9 in order to separate it from other consonants, e.g. dadareza for s. dddarza, gr. SeSop/ca, ' I saw.' 55. w is represented by three characters, as in the table. Of these the first occurs only at the beginning, and the second only in the middle of words, e. g. $&m(j waem for s. wayam, ' we ;' a5»aj^o tawa for s. tava, ' of thee.' The form of the second shows that it originated from the vowel u, being, like the English letter, a 1 double u. y The third character occurs after th, and sometimes after dh. Bopp puts all three together under the designation of Half-vowels, but represents the first two by v, and the last by w. Now v is not in any sense a half-vowel, any more than /, v having the same relation to &, both in origin and sound, as / has to p. The combinations, also, which are given of the last character, with y and r, in wyo, wra, aiwyo, 6 aquis,' zuwrd, ' sword,' supposing the first letter = e. iv, if not phonetically impossible, at least are strange. These terminations correspond to the s. bhyas and bhra, and would therefore be more correctly represented by vyo and vra, seeing that v is the aspirated 6, just as / is the aspirated p. We should be inclined to follow Eask in giving the sound w to the first two characters ; while the difference between Eask and Anquetil in regard to 58 SOUNDS the last — the one assigning to it the sound w, and the other the sound v — would lead us to the conclusion that it had either the one or the other sound, according as it arose from a consonant or a vowel, w and u, like y and i 9 have an assimilating power over the preceding syllable, e.g. haurva = s. sarva, ' whole.' 56. y, r, w, m, n, and the sibilants, cause a mute before them to become an aspirate; e.g.merethya,ug}Lva 9 kazethwahm, takhma, pathni, correspond to the Sans- krit mrtyu, ugra, Jcas-twam, tank, patni. 57. The Sibilants are z, sh, z, s. z has the two characters given in the table, which differ little in sound, but are not of the same origin. The first answers generally to the Sanskrit z. In some cases, however, it stands in place of Sanskrit s, e.g. ztdro for s, Stdras, ' stars.' It has the effect of changing w, immediately after it, top, e.g. zpd = s. zwan, gv. kvcop, ' dog ' (hound). The second character for z answers to Sanskrit y and j (a Latin y becomes j (z) in French — Lat. jocus, Fr. jeu), e. g. yuzem = s. yuydm, ' you ;' zenu = s. jdnu, 'knee.' sh is represented by two characters, which, however, are of the same origin. The second is used before vowels and the half-vowels y, w. They both answer to Sanskrit sh, e. g. a&hta = s. dshta, c eight ;' aitaishwa = s. ete'&hu, ' in these.' z answers etymologically to Sanskrit h, e.g. azem= s. aham, gr. iycov, ' I,' In some cases it stands in place of Sanskrit j, e.g. yaz- = s. yay 'worship;' and in a few cases for Sanskrit g, e. g. zdo = s. gdus, gr. yrj, < earth.' 58. h answers etymologically to Sanskrit s, e.g. ha = s. sa, gr. 77, ' she.' n is prefixed to hr for s. sr. Xkhr = s. sr. 59. The Nasals are numerous, ng has two charac- GREEK 59 ters, of which the second is evidently only a modification of the first, and is used only after e or i. They answer to Sanskrit ng, e. g. yenghe, 6 who ;' anghdo, ' of this.' n has two characters, of which the second is used before strong consonants. They correspond to Sanskrit n. m also is the same as the Sanskrit letter, n is repre- sented in two characters, of which the second = an. 60. The Zend Alphabet, as compared with the Sanskrit, has already sustained considerable loss. The Cerebrals are entirely wanting except cZ, which, how- ever, answers to Sanskrit Dentals and not Cerebrals. There are no Palatal aspirates, bh and I are also missing. On the other hand, the sibilants and di- phthongs are more numerous than in Sanskrit. 61. The following table exhibits the Zend Alphabet : Mute. Vocal. Nasal. Half-voweli. Sibilants. Vowels. Gutturals k kh (qh) g gh ng y z a Palatals ch — j — n — z eeai Dentals t th d dh n r si Cerebrals d — — — — 6 au ao 6i Labials pf b — m w shu 3. GEEEK. 62. The following list of Greek letters consists mainly of what was called the Cadmus Alphabet, traditionally derived from Phoenicia. The aspirates were at first represented by the un- aspirated letters. %, £, <£, however, are found on the earliest monuments. There appears to be no aspirated k or t, unless, like the English th, % and # represent two sounds each. The two subordinate classes of Sanskrit consonants are altogether wanting. The long vowels 7j and a>, which usually answer to the Sanskrit a, 60 SOUNDS were represented by s and o. The formation of the diphthongs is easily understood. As s, o stand for Sans- krit a, so si, 01 answer to ai, and sv, ov to au, which in Sanskrit appear as i, 6. Similarly at, av, being formed of aai, aau, correspond to Sanskrit ai, au. The im- proper diphthongs introduce no new elementary sound. £ and yfr are merely later inventions, used in place of mutes combined with a Sibilant: £ for kg; which repre- sents k, 7, or x followed by followed by , * I hasten,' forms . 63. Six consonants are wanting in the three classes of the Greek as compared with the Sanskrit, viz. : Gutturals, kh, y, z ; Dental, th ; Labials, w, sh. The w, however, may have been represented by ov in ovai= go. vai, e. woe. Mute. Vocal. Nasal. Half-vowels. Sibilants. Vowels. Gutturals K (x) V X 7 (0 — ays Si 01 at Dentals t (5) S 5 v \ p a l sv ov av Labials it as TRANSMUTATION OF CONSONANTS 75 in trahere, and the Dental aspirates are changed to Labial aspirates, e. g. ^vfios, fumus ; %rjp, fera ; %vpa, fores. 103. A singular fact is observable in Grerman, which, illustrates the dependence of the later upon the earlier classes of languages. When Dental or Labial aspirates occur, the mutes out of which they have arisen, and which characterise an earlier class of languages, are also preserved, e.g. pfeffer, zehn (z = ts), for English pepper, ten, where p and t of the second class are pre- served, and / and s of the third added. Sometimes the preceding letter is assimilated, and the aspirate becomes doubled, e. g. hoffen, ha&sen, for English hope, hate, from hopfen, hat&en. 104. For a fuller discussion of this law, the reader is referred to Grimm, Geschichte cler Deutschen Sprache, i. pp. 392—434, 1st ed. 76 BOOTS IV. EOOTS. 105. The Latin word dicitur, 6 it is said/ is found, on examination, to consist of several parts of different origin. It may be thus divided, dic-i-t-u-r. The letter r is in Latin distinctive of the Passive voice, and the u by which it is here preceded is a connecting vowel which has no effect on the meaning of the word. When these two letters are removed, there remains dicit, { it says.' Again, t expresses the Third Person Singular, answering to the English pronoun 6 he,' 'she,' or 'it;' and i is another connecting vowel. When these are removed, we have the monosyllable die, which is called the < root ' of the word. 106. All primitive verbs in the Indo-European lan- guages may be similarly reduced to monosyllabic roots. These roots, however, have no meaning, and, as far as we know, were never thus used in ordinary speech. Such words as the Imperatives die, fac, etc. form no real exception, seeing that this is not their original form, but the result of abbreviation. For the sake of rapid utterance the endings of the Imperative have been dropped, and the words have retained the meanings which the terminations gave them. The roots, not having had such terminations, have acquired no dis- tinctive meaning, and are therefore incapable of being used in such languages as those of the Indo-European family. THEIR DISTINCTIVE CHARACTER 77 107. It has sometimes been assumed that before terminations were used the roots were placed one after another, as in Chinese, and acquired a meaning from their position. The system of affixes is supposed to be of a later date. No doubt modern languages of the Indo-European family tend towards the state of the Chinese. For instance, the phrases ' I have seen,' ' you have seen, 5 'they have seen,' express three different meanings, not from any change in the terminations, but from the use of different words. In Latin, , the second to verbs in -/jli. The first conjugation inserts a between the root and the ending, e. g. s. hodh-a-mas, ' we know ; ' gr. tvttt-o-/jlsv, ' we strike :' the second appends the termination immediately to the root, e. g. s. dwesh-mi, ' I hate ; ' gr. fyrj-fii, ' I say.' 11®. The processes of Gtuna and Vriddhi need here to be noticed. The former, Gtuna, in Sanskrit, consists in prefixing a to another vowel, thus changing i or i to e (a-j-i), u or u to 6 (a + u), and r or f to ar (a + r). The latter, Vriddhi, prefixes d in a similar way, and changes a or a to a (a + a) ; i or i to di (a + i) ; u or u to au (a + u), and r or f to ar (a +r). 117 H The First Conjugation inserts a between the root and the ending. The four classes of which it con- sists are distinguished principally by the use or omis- sion of guna and the position of the accent The First Class both gunaes and accents the root vowel. Thus budh becomes bo'dh-a-ti, ' he knows.' The Sixth Class does not guna the root vowel, and. places the accent on the connecting vowel: tud becomes tud-a-ti 3 'he strikes.' Those verbs which have a as their root vowel, since it is not affected by guna, can of course be known as be- longing to the first or sixth class only by the position of the accent. Lup, vid, etc. of the sixth class have no THEIR DISTINCTIVE CHARACTER 81 guna, but insert a nasal in the root, e. g. s. lump-a-ii, ' he splits/ 1. rumjp-i-t ; 8. vind-a-ti, ' he finds. 5 118. In Greek the connecting vowel has become s (before nasals o) : Xsltt-s-ts, ' you leave;' favy-s-rs, 'you flee ;' Xdir-o-iiev, ( we leave ;' svy-o-/jLsv, ' we flee.' The guna-vowel has also become s, as is seen by comparing the above forms with s-Xur-ov, 6 I left ; ' a-cj)vy-ov } s I fled.' These being gunaed belong to the first, but 7X^-0-/^11, without guna, to the sixth class. There is not the same distinction of accent between these two classes in Greek as in Sanskrit. The third conjugation in Latin fur- nishes analogous examples. The connecting vowel has become i (before nt it has become u), e. g. leg-i-mus, 6 we read ; ' leg-u-nt, ' they read.' In Gothic the con- necting vowel is a or i. Thus haitan, ' to be named/ has in the Sing, hait-a, hait-i-s, hait-i-th, and in the PI. hait-a-m, hait-i-th, hait-a-nd, where the a appears before a nasal, like o in Greek, u in Latin, and a in Sanskrit. The guna-vowel appears as i. Thus kin becomes keina, 'I germinate' (ei being for i + i), and bug becomes biuga, f I bend.' A radical a, being inca- pable of guna, as in Sanskrit, either remains unchanged, e. g. far-i-th, 6 he wanders,' for s. chdr-a-ti ; or it be- comes i, qvim-i-th, ' he comes,' for s. gdm-a-ti. In some cases this derived i 3 being looked upon as primi- tive, is gunaed, e. g. greipa, f I grasp,' for s. grabh-a-mi. Almost all the Germanic Strong Verbs belong to the first class in Sanskrit. 119. It is interesting to observe analogous deviations from general rules in different languages. One of these consists in the radical vowel being lengthened instead of being gunaed, e. g. s. gu'h-a-ti, ' he covers,' from guh ; gr. rpi'^coy ( I rub,' from Tp*/3-; 1. duco, ' I lead,' from due-; go. us-luk-i-th, , s. d-ya-mi. Again, we meet with this y in the form of s 9 which, in some cases, must have very closely resembled it in pronunciation : for example, in Triarsctis, ' of faith/ where aco is pronounced as one syllable. Thus we have goS-ecd, ' I push,' for coS-yco. Latin verbs of the third conjugation in -io belong to this class, e. g. 1. cup-io, ' I desire ; ' s. kup-ya-mi, f I am angry;' 1. cap-io, ' I take;' go. haf-ja, ' I lift.' Such forms as pi-yu-n, ' I drink,' in Old Slavic, may belong to this class. But as the y occurs between two vowels, it may be only euphonic. Gothic verbs in -ja ( = ya) are of this class, e. g. vahs-ja, ' I grow' (wax) ; bid-ja, ' I beg ' (bid). When Sanskrit 6 is changed to a in Gothic, y is also changed to i 3 and forms with a the diphthong ai 3 e. g. vai-a, 6 1 blow,' for va-ja ; lai-a 3 ' I despise ; ' sai-a, ' I saw.' Probably no roots originally ended in diphthongs. Those which now appear in that form resulted from con- traction, and belong to this class, e.g. gafyati^ from ga, not from gai ; dhayati, from dha, not from dhai. So also dydti is from da, as is clear from the Participle dd-tas, cut off,' and the Substantive dd-trum, ' a sickle.' 122. The Tenth Class gunaes the root-vowel like the first class, and, like the sixth class, it accents the part inserted between the root and the ending. This class, instead of a of the first and sixth, and ya of the fourth, inserts aya between the root and the ending, and o 2 84 ROOTS places the accent on the first vowel of this part. It is thus identical in form with the Causative verbs, e. g. chor-dya-mi, 'I steal/ from chur. From this form have arisen most Greek verbs in -a£&>, -aco, -say, -oft), and Latin verbs of the 1st, 2nd, and 4th conjugations. Slavic verbs in -ayun belong to this class, e. g. si. rud-ayu-n, ' I lament,' for s. rod-dya-mi, ' I cause to weep.' In Gothic we have ja, the first a of ay a being dropped (just as in the Latin form -io of the fourth conj.). Sometimes the last a is dropped, and y changed to i, as in hab-ai-s. This i also is dropped before nasals, e. g. hab-a-m. When y is dropped, a+a produces 6 in Gothic, which corresponds to a in the Latin first conjugation, e. g. go. laig-o-s, s thou lickest,' for s. leh-dya-si ; 1. laud-a-s, Q thou praisest.' The intermediate step is found in Pracrit, where gan-aa-di, ' he wanders,' is for s. gan-dya-ti. 12i3. We now come to the Second Conjugation of Sanskrit verbs, including the other six classes. They all affix the ending immediately to the root, without any connecting vowel. Four of the classes, however, add an inorganic syllable or nasal letter as an enlarge- ment of the root 124. The Second Class accents the ' heavy ' termi- nations, but before the € light' terminations it gunaes and accents the roctf-vowel, e. g. e'mi, ' I go ; ' imds, ' we go.' This difference of accentuation is not observed in Greek, e. g. sl/jli, l/jlsv. Almost all the Greek roots which belong to this class end in a vowel, I, rj>d, /3a, 8co, srd, $77. The only root ending in a consonant, which immediately adds the terminations, is h, e. g. s. ds-ti, gr. Jo--™, 1. es-t, go. is-t, ' is.' In Latin ?', cZa, sta 9 fa, fid, qua (in-quam), as well as some forms of fer, and vel, belong to this class. THEIR DISTINCTIVE CHARACTER 85 125. The Third Class prefixes a syllable of redu- plication, and places the accent on this syllable, e. g. ddda-ti, 'he gives;' dadhd-ti, 'he puts.' The Greek language has many forms belonging to this class, e. g. SlBco-n, tIBij-tc, ftifir]-Ti. In Latin the second i in sisti-t, bibi-t, is a shortened form for Sanskrit a ; seri-t for sisi-t, with the common softening of s to r between vowels, and the usual change of i to e before r, is another instance of a reduplicated form of this class, as is indicated by the participle sa-tus. 12*6m The Seventh Glass inserts in the root the syllable na before the ' light ' endings, and this is re- duced to n before ' heavy' endings, being changed, when necessary, to a nasal of the same organ with the final consonant of the root In the last case words of this class coincide with those of the sixth, except in having no connecting vowel. Most of the corresponding words in Latin insert a connecting vowel, and agree with the first, instead of the second, Sanskrit conjugation. The accent in Sanskrit is placed upon na in the first case, and upon the endings in the second, e.g. yundj-mi, 6 1 bind ; ' yunj-mas, f we bind ; ' bhindd-mi, ' I split ; ' bhind-mds, 'we split;' chhinad-mi, 'I cut;' chhind- mds, 'we cut.' In Latin the corresponding words have the syllable in both cases reduced to the mere nasal, and a connecting vowel inserted, e. g. jung-o, 'I join,' jung-i-mus; find-o, 'I cleave,' jind-i-mus ; scind-o. 'I cut,' scind-i-mus. In Greek some verbs exhibit the characteristics of two classes, a nasal inserted and another appended to the root, e. g. XapfSdvo), ' I take,' Xifiirdva), ' I leave,' fMavSdvco, ' I learn,' from the roots Xa/3, \nr, fiaS, as in s-Xa/3-ov etc. In Gothic a nasal is inserted in the Present tense of standa, Pret. stoth ; a. s. stande, stod ; e. stand, stood. In this word 86 ROOTS the final consonant d appears to be inorganic, as it does not exist in Sanskrit, Grreek, or Latin. A similar phe- nomenon appears in mita, e. mefe, compared with s. ma, 'to measure. 5 127. The Fifth Class adds nu to the root, and this syllable is both gunaed and accented before ' light' endings, but ( heavy' endings both prevent guna and themselves take the accent, e. g. str-no'-mi, ' I spread ; ' str-nu-mas, ' we spread ; ' gr. ar6p-i;v-/M (where the vowel is lengthened instead of being gunaed), arop-vv- 128. The Eighth Glass probably ought to be in- corporated with the fifth. It is said to add only u to the root ; but as all the roots, except one, terminate in n, this may easily be supposed to have caused the omis- sion of the second n. Gruna and accent are the same as in the fifth class : tan-o'-mi, 'I stretch,' tan-u-mas, 4 we stretch,' gr. rav-v-fiac; av-v-fii 9 ' I complete ;' ydv-v-fiaL, i I delight in ;' o\-\v-/m 9 ' I perish,' from 6\-vv-ijll 129. The Ninth Class adds na before 'light,' and ni before ' heavy ' endings, and accentuates like the fifth class. The Greek has vtj before Might,' and va before heavy endings. This is irregularly shortened in such forms as Sd/c-vo-fjusv, ' we bite;' e.g. s. yu-na'-mi, ' I bind,' yu-ni-mas, ' we bind ; ' gr. 8d/j,-v7j-/j,i, 6 1 tame,' Sdfi-va-fjLsv ; s. str-na' -mi, ' I strew,' str-ni-mds, < we strew,' 1. ster-no, ster-ni-mus. 130. The following is a brief list of roots which may serve further to illustrate the relation of these languages one to another. It will have become obvious how the modifications of the different classes of roots are almost entirely lost in English, thus illustrating the progress of phonetic decay. THEIR DISTINCTIVE CHARACTER 87 Sanskrit. Zend. Greek. Latin. Gothic. English. A (3rj-vai gagg-an g° ga dha da $&-vai do-n, do-ff jna zna yvuj-vai gno-sco know wa vi-nds wi-nd stlia sta OTrj-vcu l-T£ kv-eiv sta-re i-re sta-nda sta-nd 1 zwi hau-hs bi-gh smi-le smi LL£l~C—ClU) pri fri (pi\-uv frij-on frie-nd zi KfX-fiai qui-eo plu fru irXs'slv plu-o floo-d zru lu bhu zrav k\v-£IV Xl/'EIV (pv-eiv clu-eo' lu-o fu-i liu-san bau-an loo-se be ad dru i-dpa-jx-ov ed-o it-an eat bandh band £-7T&-0V fid-o bind-an bind stig S'ffTLX'OV steig-an sti-le ruh rudh rood, rod bliraj (pXs-yEiv flag-rare bairh-ts brigh-t raj raz reik-s rich sach £77-0(40.1 sequ-or seek as as ko-ri es-t is is iksh OTT'Og oc-ulus aug-o eye jush zausha y£v-(t) gus-tare kius-an choose diz diz C£lfC-VV{ll dic-o teih-an teach j iv jva /3to-co viv-o qiu-s quick 88 STEMS V. STEMS. 131. We have seen that in some cases letters or syllables intervene between the root of a verb and the endings which denote person and number. Thus, in the Latin regit, ' he rules/ t, meaning 'he/ 'she/ or 'it,' is not added to the simple root reg, but to the com- pound form regi. So, in the Perfect tense rexit, ' he ruled/ t is added to another compound form consisting of reg+s + i. So also, in nouns, the endings which denote case and number are similarly added to a com- pound form ; thus, in regem the sign of the Accusative Singular m is added, not to reg, but to rege. This com- pound form, in distinction from the root, is called a stem, in harmony with the same figure of speech. The complete word, conveying intelligence, is not the bare root, nor the root and the connecting medium, or the stem, but the entire tree with its branches and fruit. 132. Nominal Stems of this kind are used with case-endings to form primitive nouns, e.g. Gren. S. regis, c of a king ; ' and without case-endings as the first member in compound nouns, e. g. regi-fugium, < king's flight.' Different genders are sometimes indicated by dif- ferent stems of the same word: aya$-6s, 'good/ ayaS-6-v, Mas. and JNeut., have the same stem ayaSo, but ayaS-rj Fern, a different one. So in Latin bon-u-s, 'good/ Mas., bon-u-m Neut., have the same stem bonw, but bon-a Fern, a different one. This distinction is pre- THEIR CONSTRUCTION 89 served somewhat incompletely in Gothic, confusedly in Anglo-Saxon, and in English the word, e. g. good, is re- duced again to its root form, all indication of stem, case, and number being lost. The Neuter gender at first doubtless was employed in reference to things which had no natural distinction of sex. In course of time it has, in some languages, acquired a wider application, as in English ; and in others has entirely gone out of use, as in French. The number of nouns is usually denoted by a modifi- cation of the case-ending. In Sanskrit bhyam is some- times the ending of the Dative Sing., bhyam the Dat. Dual, and bhyas the Dat. Plural. The Dual number, which was very carefully employed in earlier times, gradually lost its power, and then entirely disappeared, so that it no longer exists in the principal living lan- guages of the Indo-European family. The Sanskrit has it most perfectly, both in the noun and the verb. The Zend has it rarely in the noun, but frequently in the verb. The Greek preserves it extensively, the Latin only in duo and ambo. It is in the Gothic pronoun and verb, as well as the Anglo-Saxon pronoun, but apparently in no other Germanic language. Terminations of Nominal Stems. 133. In Sanskrit all the three vowels a, i, u, occur at the end of Nominal Stems. They are usually of the Masculine gender, a is always either Masculine or Neuter. It is represented by a in Zend, and in a few cases in Gothic ; by o in Greek and Latin. In later times the Latin o was in some cases changed to u, e. g. \6y-0-9, 'word;' h&p-o-v, 'gift;' domin-u-s, 'lord,' domin-o-rum; regn-u-m, 'dominion,' regn-o-rum; s. 90 STEMS vrk-a-s, ' wolf,' gr. Xv/c-0-9, 1. lup-u-s. i occurs in all three genders. It is i or s in Greek, ttoXl-s, 7roXs-co9, and i or e in Latin, facil-i-s, ' easy/ facil-e-m; s. vd'ri, ' water ;' 1, mare, 6 sea,' Nom. PI. mari-a, u occurs in the three genders. The same letter is preserved in the other languages : s.*sun-u-s, go. sun-u-s, Q son ; ' s. swad-u-s, 6 sweet,' gr. rj8-v-s. The fourth declension in Latin furnishes examples. 134. The stems endiDg in the long vowels d, i, u, in Sanskrit, are generally Feminine, seldom Masculine, and never Neuter, a is shortened to a in the other languages. But the Grothic has 6 in some oblique cases, and in the Nominatives : so, ' that ' (she), for s. sa, and hvo, ' who,' for s. Jed. i is often employed in Sanskrit and Zend to form Feminine derivatives, e. g. mahat-i\ ' great,' from mahat. A further addition is made to it where it is preserved in Greek and Latin; a or S is added in Greek, and c in Latin. Thus, s. sivad-u-s M., swdd-u N., becomes swad-i f in the Fern. ; but in gr. r]h-v-s M., rj8-v N., becomes fjh-sl-a in the Fern. ; s. jani- tri, gr. ysvsTStpa for ^svsrpia^ 1. genetrix (genetri-c-s) ; \rjpv-$, ' eyebrow.' 135. There are a few instances of stems with diphthongal terminations : rai, e riches,' becomes rd before consonants, and answers to the Latin re ; dyo is from the root div, and forms some of its cases from dyau. The Norn. Sing, dyaus corresponds to Greek Tasvs, d being dropped, y changed to £, and a shortened to s. The Latin has added i to the stem in Jov-i-s. s. gd, 6 ox,' is in Zend gau, gr. (3ov, 1. bo or bov ; s. nau, 6 ship,' gr. vav-Sy 1. nav-i-s. The stem is preserved, with- out the additional i 9 in nau-fvagus, ' shipwrecked.' 136. In Sanskrit Consonant Stems, i. e. stems end- ing in consonants, are confined to the letters n, t, s, r. Several other consonants occur at the end of roots, which are used in the formation of nouns. In Greek and Latin a consonant appears sometimes in addition to the vowel stem in Sanskrit, e. g. the patronymics in -jS, and 1. pecu-d-is compared with s. *pazu, go. faihu, 92 STEMS 6 flock ' (e. fee), s frequently occurs in Sanskrit, and serves to explain some otherwise obscure forms in Greek and Latin. Thus it appears that the opo9, ' com- pleting,' preserve the original a in the first member of the compound. In Latin this s coming between two vowels is softened to r, e. g. genus, ' race,' generis. An analogous modification of the preceding vowel also takes place in these two languages; the Nom. Sing, having o for s in Greek, and u for e in Latin, a change similar to that which occurs in the connecting vowel of verbs before nasal consonants. 137. In Sanskrit the stem sometimes assumes three different forms — the ' strong,' ' middle,' and * weak.' The ' strong ' forms are used with the lightest case- endings, the ' weak ' with the heaviest, and the ' middle ' with those of intermediate weight : e. g. rud, 6 weep,' has a reduplicated preterite, the participle of which has Ace. Sing. M. rurud-wa'ns-am, Loc. Plu. M. and N. rurud-wat-su, Gen. Sing. M. rurud-ush-as, where the stem ends in the strong -wans, the middle -wat, and the weak -ush. More generally the stem has but two forms, when the c weak ' includes the ' middle ' and ' weak ' in the previous classification. The ' weak,' in the division into three, only includes the Gen. M. and N. of the three numbers. The eight cases of Sanskrit (and Zend) are therefore divided into two classes — the 6 strong,' those which have the strong stem; and the c weak,' those which have the weak stem. The follow- ing example will show which they are : — THEIR CONSTRUCTION 93 Strong Cases. Weak Cases. Sing. — Xom. Voc. Ace. Instr. Dat. Abl. Gen. Loc. bhar-an(t), * bearing bhar-ant-am bhar-at-a bhar-at-e bhar-at-as bhar-at-as bhar-at-i Dual. — Norn. Voc. Ace. bhar-ant-au Instr. Dat. Abl. bhar-ad-bhyam Gen. Loc. bhar-at-os Flu.- — Nom. Voc. Ace. Instr. Dat. Abl. Gen. Loc. bhar-ant-as bhar-at-as bhar-ad-bhis bhar-ad-bhyas bhar-at-ani bhar-at-su This arrangement is not carried out so fully in Zend as in Sanskrit. The accentuation of words with mono- syllabic stems, where this difference of strength in the stems cannot be made, sometimes coincides with this arrangement of cases. And, singularly enough, the Greek retains the same position of the accent : e. g. s. vak and gr. 6tt-$ are of the same origin, and are thus accentuated. 94 STEMS Strong Cases. Weak Cases. Sing. — Nom. Voc. vak (' voice ') OTT-Q Ace. vacb-am OTV-Ct Instr. vacb-a Dat. vacb-e Abl. vacb-as Gen. vacb-as ott-oq Loc. (gr. Dat.) vaeb-i' on-i Dual — Ace. Voc. va cb-au 07T-6 Instr. Abl. vag-bbyam Dat. vag-bbyam 6ir-oiv Gen. Loc. vacb-6's Flu. — Nom. Voc. vacb-as 07T-SQ Ace. vacb-as 07T-EQ Instr. vag-bbis Dat. Abl. vag-bbyas Gen. vacb-am ott-wj/ Loc. (gr. Dat.) vak-sbu or-tri The Ace. Plu. is here placed among the strong cases, because of the position of the accent. Again, in Trarep, fjirjTsp, Svyarsp, the s is dropped only in the weak cases. In Gothic also a is dropped before r and changed to i before n in the weak cases : e. g. broth ar becomes in the Dat. brothr, Gen. brothrs ; ahan, Dat. ahin, Gen. ahins. 138. When a case-ending which begins with a vowel has to be added to a stem which ends in a vowel, a euphonic consonant is inserted between them, e. g. in the Instr. Sing, and Gen. PI. in Sanskrit, n ; in the Gen. PI. of three c^dVnsibn^ in Latin, r. s; 95 VI. FOEMATION OF CASES. 139. In Sanskrit and Zend there are eight cases, of which the Greek preserves only^e; the Latin, six ; the Gothic, five ; Anglo-Saxon, five ; and English, none. Some isolated instances remain in different languages, of a case which no longer forms part of their grammar ; as, for instance, single words referable to the Locative case in Greek and Latin. THE NOMINATIVE SINGULAK. 140. The sign of the Nominative Case Singular, in the Masculine and Feminine, is s. It forms 6 with a preceding a, sometimes in Sanskrit, always in Zend. It is omitted at the end of consonant stems, and if the stem ends in two consonants, the latter of these is also dropped, e. g. Nom. bibhrat, ' bearer,' for bihbrats ; tuddn Nom., ' striker,' for tudants. In Zend the s is preserved, e. g. Nom. afs, ' water.' Stems in n omit this letter as well as the Nominative sign, lengthening the preceding vowel as a compensation, e. g. Nom. dhani', 'rich,' from dhanin; Nom. ra'ja, from ra/jan. The same occurs in Zend, except that the vowel is lengthened only in monosyllables, e. g. Nom. ipd, * dog,' from zpan ; ashava, ' pure,' from ashavan. Stems in -ar, -as omit both their final consonant and the Nom. sign, lengthening the precedi vowel when short, e. g. Nom pita', ' father,' from pitar; data', ' giver,' 96 FORMATION OP CASES from datar. The Zend omits the lengthening of the short vowel, e. g. Nom. brata, data. Stems in -as omit the Nom. sign, and lengthen the preceding vowel, e. g. Nom. durmanas, ' evil-minded,' from durmanas. Of vowel stems the Feminines in -a always, and those in -i generally, omit the Nominative sign, e. g. Nom. ziva from ziva, Nom. nadi from nadi, but Nom. bhis from bhi. The Greek and Latin languages preserve the Nom. sign in consonant stems, omitting the stem consonant if it be a Dental, e. g. x^P L " s f° r X a P LT ~ 9 > virtus for. virtut-s. So also tlSsl-s, amans, for tiSsvt-s, amant-s. But Gutturals and Labials in the stem are preserved, e. g. tcopa/c-s, \at\a7r-s, lex (leg-s). Greek stems in -v sometimes preserve this consonant, and sometimes the Nominative sign, but never both, and generally a pre- ceding short vowel is lengthened, e. g. iizka-s (/leXav-), rkprjv {repev). In Latin there is the same diversity, and in many cases both consonants are omitted, e. g. sanguis (sanguin-), fiumen (flumin-), homo (homin-). -p is generally preserved and the Nom. sign omitted: sap, ' spring ' (eap-), but ixaprv-s, ' witness ' (/jLaprvp-). The stem consonant r is likewise preserved in Latin, and the Nom. sign omitted, e. g. marmor, ver. Stems in s lengthen the preceding vowel, as in Sanskrit, to com- pensate for the omission of the stem consonant or of the Nom. sign, e. g. hvapBvris- from Svcr/jLevecr-, Gen. 8vafjL6vs(a)-o9. The same remark holds good of the Latin words mos, flos, etc., where the s may be regarded either as belonging to the stem and softened to r between two vowels in the oblique cases, or as being the Nom. sign before which the stem consonant is dropped. In Gothic a and i are omitted before the Nom. sign where it is possible, that is, in all but monosyllabic THE NOMINATIVE SINGULAR 97 words and such words as harja, 6 an army.' The a is weakened, however, to i in the Gren. harjis : ji thus formed are often contracted to ei ( = i) 5 in other cases diminished to i or altogether dropped, e. g. ivulf-s, ' a wolf ;' gast-s, ' a guest ; ' althei-s, ' old ; ' suti-s, ■ sweet ; ' gamein-s, ' common.' After r the Nom. sign is some- times omitted, e. g. vair 9 e man ; ' fcngr-s, 6 finger.' va after a long vowel loses a in the Nom., and after a short vowel the v is also changed to u, e. g. snaiv-s, 6 snow;' qviu-s, ' living.' n is dropped, but nd is fully preserved before the Nom. sign, e. g. Nom. ahma, 6 mind,' from ahman; bairand-s, ' bearing.' The n in Feminines, preceded by 6 or ei 9 is inorganic, e. g. viduvdn, s. vidhava, 1. vidua ; quivon, s. jivaf. The Neuter has no special form for the Nominative case; the Accusative form is used in its stead. 141. The following list contains illustrations of the Nominative case Masculine and Feminine : K 98 FORMATION OF CASES ^ * % i "& ^q d a o ^ & o o u o © bq .-c ^ o «G ,jq to & 'to bo o 3 c 3 c3 <3 a o -a .2 «J aj .23 o 'T* -O ag .j-h G C c$ T5 ^3 S3 ^2 ^5 O Sh S- *_, c3 ^3 -I- X CQ. 3 5 3 .2 15 a rO c3 rQ T3 T3 ej £ *-4-3 *^ N •N N03 43 vo3 A & 3 <3 CO 09, rspas, for Tsrv "o o o £ 3 X •r* CO o o c < 33 T3 c2 'So CO fee < JO 5+-I 2 %H Pi o3 oS -^> O "o "3 ^ CO 03 m O > T3 'Sb fcJQ a Sh .O <+H a £ S S 3 3 CO ^ cr 1 OQ „ , J* ft ft ft I ft cS 0) o o o3 Pi 03 1 M ^*/ *a o CO o Jh Pi u pj a a 03 a 03 a o3 a 03 I a a a a a 3 a -1-3 S3 Pt 03 Pi vo3 CO 03 53 <^M a and with Feminine stems in a, and poly- syllables in i, u, it becomes ai, preceded by ay instead of a, e. g. azvjay-ai from dzvjcl. Masculine stems in ?, u, are gunaed, and Feminines, if the case-ending is e. Neuter vowel stems insert n. The same sign is pre- served in Zend, but ai is preceded by ay instead of ay. Masculine a stems make dya (=a-fay + a = a + e-fa) in Sanskrit, and ai ( = a + ai = a + e) in Zend. The use of guna is also only partial in the i and u stems. What is called the Dative in Greek and Latin corresponds in 104 FORMATION OF CASES origin to the Sanskrit Locative, and will be considered in connection with that case. In Latin the ending is i. If in Latin as in Greek this case was originally the Locative case, the long quantity of the final vowel must be referred to a general tendency in the Latin language which increases the quantity of this vowel at the end of a word without organic cause. To the general rule that final i is long there are but very few, and those isolated, exceptions. Although, therefore, the quantity of this vowel induces Bopp to regard this case in Latin as originally a Dative, yet the great probability that Greek and Latin both adopted the same course, and the existence of a suffi- cient reason in the genius of the Latin language to account for the subsequent change of quantity in the vowel, justify our regarding the Latin Dative as origi- nally a Locative case. In Gothic this Dative ending is entirely lost, unless the i in gibai (for giba-i) be regarded as a remains of it in the Feminine a stems. The gunaed forms of the stems in i } i, u, u, are retained ; but in the first of these cases the final i is dropped, e. g. gasta (for gastai), anstai, sunau, Jcinnau, from gasti-, ansti-, sunu-, kinnu-. 146. The Pronouns are in so far peculiar that in several cases, of which the Dative is the first that comes under our notice, they insert the syllable sma between the stem and the case-ending. This syllable is capable of many modifications by omission of one or other of the letters, and by euphonic changes. It appears con- sequently in many fragmentary forms in different lan- guages; e. g. the Dat. Sing, of ka is kdsmai, 'to whom? ' ( = lea + sma 4- e). The syllable appears in Zend as hma, in Pracrit as mha. This inversion of letters resembles that in the English word ivho for the Anglo-Saxon hiva, THE DATIVE SINGULAR 105 and it renders less startling the connection between the Sanskrit sma and the Gothic nsa : the s, being a Dental, changes the preceding Labial m to the corresponding Dental n. It thus becomes easy to identify the English us and the Sanskrit asmdn. For, as s. asman is pro- bably for asmans, the vowel being lengthened to com- pensate for the s, it is clearly the same as the go. unsis (for u-nsa-s), where s is preserved and n dropped. But this unsis has already become us in Anglo-Saxon as well as English, merely by that system of abbreviation which Home Tooke so convincingly showed to be an inherent characteristic in the history of language. This particle, however, assumes no less than six different forms in Gothic, viz. nsa, zva, gleet, gqva, mma, and s. The first, nsa, occurs in the Accus., Dat., and Gen. Plur. of the first personal pronoun, and the second, zva, in the same parts of the second personal pronoun, i. e. u-ns-, ' us,' u-nsi-s, ' to us,' u-nsa-ra, ' of us ; ' i-zvi-s, € you,' i-zvi-s, ' to you,' i-zva-ra, ' of you.' In the corresponding cases of the Dual the first person ha&gka, and the second gqva, i. e. ii-gki-s, ' us two,' u-gki-s, ' to us two,' (u-gka-ra) ; i-gqvi-s, c you two,' i-gqvi-s, ' to you two ' i-gqva-va, ' of you two.' The g here stands for n (ng before Gutturals). The fifth form, mma, occurs in such Datives as i-mma, c to it,' hi-mma, ( to him,' hva-mma, ' to whom ? ' where mm is by assimilation for sm. The sixth form, s, appears in the Datives, mi-s, ' to me,' thus, 6 to thee,' sis, ' to one's self.' Bopp also ascribes the same origin to the s in the Nom. Plur. vei-s, ' we,' and ju-s, e you.' In the Feminine forms of the third personal pronouns in Sanskrit, the Dative, Genitive, and Locative Sing, end in -sy-ai, -sy-as, -sy-am, of which the first part sy may be for smy, and this for smi, an ordinary Feminine equiva- 106 FORMATION OF CASES lent in Sanskrit for sma. A confirmation of this view is obtained from the Zend, which preserves forms like yahmya for s. ya-sy-am. For Zend hmy presupposes a Sanskrit smi. In Gothic, 6 is a Feminine termination, which would give the form smo to this particle. If then m be dropped as in Sanskrit, the remainder so serves to explain such words as Gen. Sing, thi-zo-s, the s being regularly softened to z between two vowels, as it is in Latin to r. In Anglo-Saxon a further change has taken place. The Gothic nsa has perhaps entirely disappeared in the Ace. and Dat. us, Gen. u-re. The ow in the Ace. and Dat. eow, Gen. eower 9 may be an equivalent for the v (or iv) in the Gothic zva. In the Dual of the first person (Ace. and Dat. u-nc 9 ' us two,' Gen. u-nce-r 9 ' of us two ') and the Dual of the second person (Ace. and Dat. i-nc 9 ' you two,' i-nce-r, ' of you two'), the remains of this particle are almost as complete as in Gothic. Why has the Dual preserved fuller forms than the Plural ? Probably because the Dual had, for a long time, been of rare use in common language, and the antique forms were therefore preserved, whilst the every- day use of the Plural caused it to be still further abbre- viated. This conjecture is confirmed by the existence of u-se-r as an older poetical form for u-re. The fifth Gothic form, mma 9 is represented by m in Dat. him, thdm, zvham. The s of mis, thus 9 has disappeared in me 9 the. In the Feminine hire, thcere, as compared with the Gothic thi-zo-s 9 the s appears weakened still further to r, and the vowel from 6 to e. In English, as the Dual is lost, the third and fourth Gothic forms of course disappear. The rest is nearly in the same state as in Anglo-Saxon. First person plural, us, our ; second person plural, you, your. The Anglo-Saxon e before THE DATIVE SINGULAR 107 vowels was pronounced y, and the ow probably 6, as in the vulgar pronunciation of yd, yo-er, for you, your, at the present day. Him, which was both singular and plural in Anglo-Saxon, is now restricted to the Singular, and the Plural Hhern' borrowed from ' that,' which has no variation of cases. Whom answers to hwam. The Feminine r is preserved in her for Anglo- Saxon hire. In Greek rj/justs or afiass for a-afjuz-s, and vpsis or vfifjuBs for v-Gfjbs-s, exhibit the same particle, almost as complete as in Sanskrit ; whilst in the Latin forms no-s, vo-s, it is much abbreviated. 147. The following instances of the Dative case in Pronouns illustrate the use of the particle sma : Sanskrit. Greek. Gothic. Norn. Phi. Ace. Plu. Instr. Plu. Gen. Plu. Kom. Sing. Abl. Sing. Loc. Sing. &sm6 Vl! AelQ \ as man rjfiag asmabhis asmakam t/fnov kasmai yasmat tasmin — — 4 by us ' unsara ' of us ' hvaninia (to) whom? ' from whom ' { in that ' 108 FORMATION OF CASES The following list illustrates the Dative case : Sanskrit. Zend. Gothic. English. m. azw-aya azp-ai vulfa wolf f. azway-ai hizvay-ai gibai 'gift' m. patay-e paithy-ai gasta guest f. pri'tay-e afritay-ai anstai ni. sunav-e pazv-e sunau son f. hanav-e tanu-y-e kinnau chin m. ? . gav-e gav-e cow f. vach-e vach-e m. bharat-e barent-e fijand fiend m. azman-e azmain-e ahmin n. namn-e namain-e namin name m. bhratr-e brathr-e brothr brother f. duhitr-e dughdher-e dauhtar daughter m. datr-e dathr-e n. vachas-e vachanh-e — — - THE ABLATIVE SINGULAK. 148. Except in Sanskrit, Zend, and Latin, the Ablative has but few representatives. Its sign in the above languages is t 9 d, d, respectively. In Sanskrit it is preserved only in the a stems. The vowel is lengthened to a, e. g. vflca-t, i from a wolf.' In Zend a becomes lengthened, as in Sanskrit, before the ending, e. g. vehrka-d ; i is gunaed, e. g. dfritdi-d, ' benedictione ;' u assumes the forms p6v-cQ-9, e wisely,' etc. An instance of the preser- vation of 8 preceded by a short vowel, as in Latin, is furnished by acppo-S-LTT], if the meaning ' sprung from foam ' is correct. Gothic adverbs in 6 furnish examples of an Ablative case, for 6 = s. a, and final t is uniformly dropped, so that the termination -6 corresponds to s. at in such words as thathr-o, ' from there,' hvathr-o, ' from where,' from stems in -thara, containing the expression for the comparative degree. The two adverbs in Anglo-Saxon answering to the above in meaning are differently formed, viz. hwanon, thanon. From these we have the 110 FORMATION OF CASES English words whence, thence, containing an additional adverbial ending. But many Anglo-Saxon adverbs in -e, being equivalent in meaning to the Latin Ablative, appear to be of this case, e. g. the first word in micl-e mare = 1. multo magis, ' much more.' The following are a few instances of Ablative termi- nations : Sanskrit. Zend. Greek. Latin. Gothic. Anglo- Sax. m. azwa-t azpa-d ofjuo-g alto-d hvathro micle f. pri'te-s afrit 6i-d navale-d THE GENITIVE SINGULAE. 149. Unlike the Ablative, the Genitive case is very extensively represented in various languages. In Sanskrit it assumes four different forms, viz. sya, as, as, s. The first, sya, is employed with a stems, and one other word, the personal pronoun 'So P ** &fi -13 'feb C3 CO CO CO i en i 3 yi CO ^3 'So CO CO p 03 P CO P 3 p -, CO c3 P^ <:3 1 £ w £ •S3 vc3 •S3 Cc3 vc3 pi* vo3 P-, » ttH 114 FORMATION OF CASES THE LOCATIVE SINGULAR. 151. The Locative case is expressed in Sanskrit and Zend by i in the a stems and the consonant stems. In the first of these cases a and i are contracted into e in Sanskrit^ and to e or 6i in Zend, e. g. s. azwe, z. azpe ; s. rnadhye, z. maidhyoi ; s. na/mn-i, z. namain-i. 152. In Greek this form appears as a Dative case, and is indicated by the i added to consonant stems, and by i subscriptum with vowel stems, e. g. %wpa, Xoyw, iroh-L The Locative meaning is preserved in many expressions which have come to be regarded as adverbial, e. g. AcoSwvi, MapaScovi, XaXa/jblvi, ' at Dodona,' etc. ; a, ' afield,' 'in the field;' oil/cot,, ' at home;' ^apac, 6 on the ground.' The Latin Dative has i with the consonant stems, where the length of the i is probably the result of a general tendency in the Latin language, in which this letter is almost always long when final. The i is pre- served distinct in all the declensions in the older stages of the language, e. g. familia-i, populo-i, ped-i, fluctu-i, re-i. In later times the first declension re- duced this ending to e, making a diphthong with the stem-vowel as in familia-e (se), and the second in- corporated it with the stem-vowel, which consequently became long, as in populo for populo-i. Some writers also have fiuctu for fluctii-i etc. in the fourth declen- sion. The stem is not subject to the same amount of modification as in Sanskrit. In Latin the Locative form is said to be used for the Genitive case in the second declension, where t appears as a contraction of o+i. Both Bopp and Eosen adopt this view of the Locative origin of the Latin THE LOCATIVE SINGULAR 115 Genitive in the second declension. Yet it seems to do some violence to the general spirit of language. The meanings of the two cases lie very far apart, and the form i may be for o + i out of s.a-sya. This would make the proximate forms of both Genitive (i) and Dative (6) to be the same, o+% The reason for the difference in the ultimate forms may be that in the Genitive the case-ending represented by i was ' heavier' than the stem-vowel o ; and therefore the sound of the former predominated when the whole was reduced to one syllable, and i was the result. On the other hand, in the Dative, the case-ending i being 6 lighter ' than the stem- vowel o, the sound of the latter predominated, and the i had no other effect on it than that of lengthening it to 6. In this view the Greek and Latin forms harmonise together. In the Genitive the consonant stems have -o$ in Greek and -is in Latin. The Masculine and Neuter a stems have ov for oo from o{at)o in Greek, and i for oi from o(s)i(o) in Latin. The Feminine a stems have -s preceded by a in Greek, and -s preceded by a in Latin (paterfamilias). In the Dative the consonant stems have i in Greek, and i (for i) in Latin; the Masculine and Neuter a stems have ©from o + 1 in Greek, and 6 from o + i in Latin; the Feminine a stems have iota subscription in Greek, and e (for i) in Latin. 153. One of the most unsatisfactory rules of Latin syntax is that which Zumpt (§ 398) expresses as fol- lows : — ' In answer to the question where ? the names of towns in the Singular, if of the first or second de- clension, are in the Genitive; if of the third, in the Ablative case.' The rule would be much simpler and more satisfactory if it could be thus expressed : — 4 In answer to the question where ? the names of towns in the Singular are in the Dative case.' At first sight i2 116 FORMATION OF CASES there are two objections to this — one in regard to the meaning, and the other in regard to the form, of the words in question. The usual meaning of the Dative is not adapted to answer the question where"? If, however, the Latin Dative, like the Greek Dative, is in reality the Locative case substituted in place of the Dative, the first objection disappears, because it is very likely that with the old form some remains of the original meaning should be preserved. The objection that the words in question are not in the form of the Dative does not apply at all to the first declension, for Romce is as much a Dative as a Genitive in form. In the third declension the difference between e and i can hardly be looked upon as determining the case in the Singular, for these letters not unfrequently change ; e. g. hosti-s and hoste-m have the stem-vowel as i in the one case, and as e in the other. Besides this, the words in question sometimes are found with the usual Dative form i 9 e. g. Tibur-i, Carthagin-i, meaning ' at Tibur,' 'at Carthage.' There remains the second declension. But in the Singular the Datives Abydo, Corintho, etc., not unfrequently occur. It is only, therefore, in some instances of the second declension that any real diffi- culty occurs as to the form, and these are doubtless the result of a false analogy which led Eoman authors to write words belonging to a case (Locative) of which they had no consciousness like a case (Genitive) with which they were well acquainted. All these names of towns, therefore, of whatever- declension they appear, with a Locative meaning may be regarded as Latin Datives, that is, original Locative cases. A few other words are similar in meaning and admit of the same explanation, i. e. domi, f at home ; ' ruri or rure, ' in the country ;' humi or humo, ' on the THE LOCATIVE SINGULAR 117 ground/ etc. Corresponding words in the Greek lan- guage which are clearly Datives (originally Locatives) confirm this view of the Latin words : olkoi, 6 at home/ %a/W, ' on the ground/ compared with domi, Uumi. Comp. Sect. 152. It is easy to see how this confusion arose. When the Locative case was generally employed as a Dative, the meaning appeared inconsistent with the idea that the words in question belonged to that case. They were, however, manifestly case forms, and were assigned to such other cases as they resembled. For instance, in the passage, Romce Consules, Carthagine Sufetes, sive judices, quotannis creabantur ( ' At Home Consuls, at Carthage Sufetes, or judges, used to be appointed yearly '), as Romce and Carthagine were supposed by their meaning not to be Datives, they were referred to the other cases which they resembled, i. e. Eomse to the Genitive, and Carthagine to the Ablative. The diffi- culty as to the meaning, however, was only altered, not removed, by this method ; whilst referring these words to the Locative case fully justifies the sense in which they are employed. 154. There are three other forms of the Locative case in Sanskrit. The first, au, is used with Masculine i and u stems, and sometimes with Feminines ; but the stem-vowels i and u are dropped (except in paty-du, sdkhy-au), e. g. pri't-du, sun-au. This Bopp regards as really a Genitive ending, viz. au for as. The Zend has 6, which is also a Genitive form. The second additional Locative form, m, is used only in the pronouns of the third person, e. g. tdsm-in, ' in that ; ' kasm-in, ' in whom ? ' The third, am, is used with Feminine stems ending in a long vowel, and sometimes with those ending in i or it, e. g. bhiy-am, 118 FOKMATION OF CASES 4 in fear.' Perhaps this form of the Locative is pre- served in such Latin expressions as ante diem quartum Nonas Januarias, ' on the fourth day before the Nones of January;' where the preposition ante governs Nonas, and diem means ' on the day.' The employment of the Locative in regard to time is not without example in other languages, e. g. s. divase' , ' in the day/ nizi, ' in the night ;' gr. rfj avrfj fjixspa, ' on the same day,' vvktl, ( at night.' The similar use of the Latin forms die, node, interdiu, noctu, ' by day,' ' by night,' makes it probable that they were originally Dative, that is, really Locative cases. The Dative and Ablative forms fluctuate, not only in the cases quoted above, but also in the employ- ment of both vespere and vesperi for 6 in the evening ; ' luce and luci for 6 during the day,' etc. The Ablative gradually absorbed these and similar meanings to itself, and the Ablative form was substituted where no pre- position occurred, whilst the use of a preposition in such expressions as ante diem quartum Nonas, etc., caused forms in m which look like an Accusative to be preserved. 155. The following is a list of instances of the Locative case : Sanskrit. Zend. Greek. Latin. m. azwe azpe 'lttttuj equo f. azway-airi hizvay-a X upy equa3 m. paty-au TTOdl'L hosti (i + i) f. pri't-au 7TlpTl-l n. va'ri-n-i ISpt-i fideli (i + i) m. sun-aii VSKV-l pecu (u -f i) f. han-au y'tvvi socru (u -f i) n. madhu-n-i fitdv-i pecu (u -|- i) m.f gav-i (ioP'i bov-i f. vach-i 7T-1 voc-i m. azman-i azmain-i caifxov'i. sermon-i THE VOCATIVE SINGULAR 119 Sanskrit. Zend. Greek. Latin. n. namn-i namain-i TaXaV't nomin-i m. bhra'tar-i brathr-i (ppdrop-i fratr-i f. duhitar-i dughdher-i Srvyarp-L matr-i n. vachas-i vachah-i £7T£(7)-l gener-i THE YOCATIYE SINGULAK. 156. The Vocative has no distinctive sign. A few instances occur of the Nominative form being employed for the Vocative, i. e. in Latin neuters and in such words as Ssos, deus, where the familiarity implied in the short Vocative form is not allowable. In some instances the stem-vowel is lightened : e. g. gr. s and 1. e of the second declension for o or u. In Sanskrit the accent is drawn back to the first syllable, and in some Greek words it is placed as far back as possible. 157. The modifications of the Vocative will be easily understood from the following list : Sanskrit. Zend. Greek. Latin. Gothic. m. azwa azpa 'Linre. eque vulf n. dan a data duipo-v donu-m daur f. azwe hizva y&pa equa giba m. pate paiti TTOGL hosti-s gast f. pri'te afriti TTOpTl turri-s anstai n. vari vairi Id pi mare m. suno pazu V8KV pecu-s sunau f. hano tanu ■ykvv socru-s kinnau n. madhu madhu fi&v pecu . m.f. gau-s gau- s f3ov bo-s f. vak vaksh-s ? 07T-Q voc-s m. azman azman daifiov sermo ahma ? n. naman naman rdXav no men namo ? m. bhratar bratare (pparop frater brothar f. duhitar dughdhare Svyarep mater dauhtar m yachas vacho F.7TOQ genus 120 FORMATION OF CASES THE NOMINATIVE PLURAL. 158 > The Nominative Plural in Sanskrit has the sign as, e. g. dzman-as, vfk-as, pdtay-as. The a and a stems, of course, make the Nom. Plur. end in -as, and the i and u stems are gunaed. In Zend the original ending is represented by -az when the conjunction cha is added. In other cases s appears as o, and as as 6, e. g. azman-azcha, azman-6, 6 stones ; ' vehrk-ao, € wolves.' The gunaing of i and u stems is arbitrary. One word preserves the original s when final, viz. geu-s. In Greek the ending is -ss. The i and u stems are not gunaed. The a and a stems exhibit i, making with the stem-vowels ol and at, e. g. iraTsp-es, ( fathers ; ' ttoctl-ss, ' husbands ; ' lttttol, c horses ; ' x^P a h * lands.' In Latin consonant stems have -es, with the vowel long. The i and u stems are not gunaed, but form with the ending -es and -us, for i-es, u-es. The a and a stems form i and ce (for ai), *which resemble the Greek ol and at, e. g. patr-es, ' fathers ; ' host-es, c ene- mies ; ' exercit-us, < armies ; ' animi ' minds ; ' familice, 6 families.' , In Gothic the ending is reduced to s in the consonant stems. The i and u stems form, with the endings, is (eis) and jus (yus\ where the stems may be regarded as gunaed, since the guna-vowel in Gothic is i ; the ending would then be s only, as in the consonant stems. Or i and ju may be regarded as lengthened forms of the stem-vowel, like the Latin e and u, caused by the vowel of the ending. The a and a stems have -os, which answers to the Sanskrit -as, e. g. ahman-s, gastei-s, sunju-s, vulfos, gibos. THE NOMINATIVE PLURAL 121 In Anglo-Saxon -as is preserved in Eask's second class of the second declension, which includes all the Mas- culine nouns not ending in a or u. Elsewhere the Nominative Plur. ends in -a, including the stem-vowel, e. g. dag-as, ' days ; ' suna, 6 sons.' In English s still appears as the sign of the Nominative Plural. The ending of the Nom. Plur. Neuter presents some difficulty. It appears generally as -a, e. g. 1. gener-a, etc. With a stems it would form -a, an equivalent for which appears in some pronouns, e. g. 1. quce, ' which,' hce-c, ' these ; ' go. hvo, tho ; a. s. hwa, thd. In the Latin forms quce, hce-c, the original ending -a is reduced to e. In nouns, however, the long quantity thus obtained is reduced to a. In Sanskrit this a appears already weakened to i 3 which is connected with vdwel stems by n. If the stem-vowel be short, it is lengthened, e. g. dd'na-n-i, va'ri-n-i, madhu-n-i, from da'na, va ri, madhu. This a Bopp regards as an abbreviation of the form -as which appears in the Masculine and Feminine. But the reason whic^ is given for s never having been used in the Singular will apply also to the Plural Neuter. This a may therefore be regarded as the original ending of Neuter nouns in the Nominative Plural. A peculiarity appears in the Nom. Plur. of the first and second declensions in Greek and Latin — viz. instead of -as we find i and i (e) added to a and a stems. The same form appears in Sanskrit, Zend, and Gothic, restricted, however, to pronominal stems in -a, e.g. s. te (for ta-i), z. te, go. thai, e. they. Bopp regards this i as an inorganic enlargement of the stem, and thinks the case-ending to be entirely lost. But if we suppose s of the original -as to be dropped, which is frequently 122 FOKMATION OF CASES the case with final s, and a then weakened to i, as in the Sanskrit Neuter nouns, we reach the same result without supposing any step so arbitrary as enlarging the stem without apparent reason. 159. The following list contains illustrations of the Nominative Plural : ™ o o B B 03 *T3 a B o o3 02 *B o3 c3 T3 B B B 02 o3 pB •^3 B B .B B T3 ^3 &0 T3 o3 bu B 03 rB B pB B T3 B <3 o pB Ui y B -2~ B 's B <03 eft o3 s o> Uj 1 1 Q_ o Q_ 8 ^ .2 Hb ^ o3 v B 03 +3 2 c r£ «) eg ^3 CQ eg i? o 3 <3 3 o s pJ3 -m o <* 02. 5^ eg 3-3 2 eg CO <£ £ 3 -» O ^ .a rO rC I cS rC rQ -» og r& >t rQ 1 5-.. rQ -U <6 , c of feet ; ' ltttt-oop, ' of horses ; ' T-cbv, ' of the ; ' /3o-wv, c of oxen.' In Latin the third and fourth declensions, embracing the consonant and i and u stems, add the ending imme- diately to the stem, e. g. ped-um, c of feet ; ' hosti-um, ' of enemies ; ' exercitu-um, ' of armies.' The first, fifth, and second declensions, embracing the a and a stems, insert r, which is the Latin equivalent for Sanskrit s, between two vowels : equa-r-um, ' of mares ;' re-r-um, ' of things ; ' equo-r-um, ' of horses.' In Gothic n appears only with the a stems, and s is THE GENITIVE PLURAL 129 softened to z. The latter occurs only in pronouns and strong adjectives. e.g.(cjibo-n-6), ' of gifts;' thi-z-e'MsiSC., thi-z-o Fern., ' of these ;' blindai-z-e Masc, blindai-z-6 Fern., ' of the blind.' The half-vowel v is developed in the u stems, which are gunaed, whilst the vowel of the d and i stems is dropped or incorporated with the termination, e. g. suniv-e, ' of sons ; ' vulf-e, ' of wolves.' In Anglo-Saxon -a is connected with Feminine vowel stems by n ; and the pronouns tha-r-a, this-s-a, exhibit remains of the original s as it appears in Sanskrit. 166. The Greek and Latin forms of the Genitive Plural supply a strong reason for regarding s as the consonant originally inserted in all instances. The change from s to n 9 as in Sanskrit, will be more fully discussed on a subsequent occasion. The absence of this consonant in Greek is consistent with the general tendency of the language which causes s to disappear between vowels. In Latin also, as a rule, s becomes r in the same position. Hence the use of r in the Geni- tive Plural of the first, second, and fifth declensions. The remains in Gothic and Anglo-Saxon also are in harmony with the view of s being the original consonant. 167. The following list includes the above modifi- cations of the Genitive Plural : K 130 FORMATION OF CASES ra o to - p to o 5 ^4 o o p 09 » .P -^ O P wV "2 . 3 p p go rP P ^3 O P <0 i ^3 P o3 pP 15 s S3 S-. S3 » CT 1 J 5 5 p- P- rn B p ■ o p 1 B B a p p 3 ^3 2- 5 8 2 •p C5 c a S3 a pj 1 as oa > tk a. .5 SI 15 *p S .2 <«* B a •p cS i P i P P -*-5 i P i P ^P ^3 a a g 0£ a •p C3 I o •p p o> u a •p o3 P cs a ■SJ C3 '3 si rP bO p P. C3 -P O p. 1 g r*

a rO VC3 ^2 •TO > v: P pI S «M p p <4-i* a a a «*-? p THE LOCATIYE PLURAL ]31 THE LOCATIYE PLURAL. 168. The Locative Plural has disappeared from several languages. Its sign is in Sanskrit -su, in Zend -hu, and in Greek -at. In Sanskrit the laws of euphony change -su in certain cases to -shu. The same takes place in Zend, where we find -shv a and -hva, which makes it probable that in Sanskrit also the original form was -siva. In Greek i is added to the a and a stems, as in Sanskrit it is to the a stems. A connecting vowel, s 9 is sometimes added to the % u, and consonant stems in Greek; and the forms ending in -acre perhaps contain the half-vowel which appears in Zend, assimilated to the preceding consonant. Therefore -crat is for aft,, cor- responding to Sanskrit -siva. In later times the final i was dropped in the first and second, i. e. the a and a declensions. In -aiv of the third declension v is inorganic. 169. The following is a list of Plural Locatives : Sanskrit. Zend. Greek. m. azwe-shu azpai-shva 'LxTroi-ai f. azwa-su hizva-hva y&pai-ai n. dane-shu datai-shva ? CWpOL-Gl m. pati-shu paiti-shva? TTOGl-ai f. pri'ti-shu afriti-shva 7TOf)TL-(JL i\. vari-shu vairi-sliva idpi-m m. sunu-shu pazu-shva vstcv-cri f. Mnu-shu tanu-shva ■ysvv-Gi n. madhu-shu madhu-shva p,'s5v-(ji m.f. go' -shu gau-shva (Bov-gi f. vak-shii vakh-sva ? OTT-GL m.n. bharat-su ,i. (pspov-ai K 2 132 FORMATION OF CASES m. n. ni. f. n. Sanskrit. azma-su nama-su bhratr-sliu duhitf-sbu vachas-su Zend. azma-hva nama-hva bratar-e-shva dughdher-e-shva vacho-hva Greek, daifio-CTL TaXa'cn (pparop-ai Suyarpa-Gi t7reo-(Ti DUAL. 170. The Dual has but three forms for the eight cases, and appears only in the declension of the Sanskrit, Zend, and Grreek. THE NOMINATIVE, ACCUSATIVE, AND Y0CATIVE DUAL. These three cases have for their sign in Sanskrit -au, in Zend -do, and in Greek -s. From the Sanskrit and Zend it appears probable that -as was the original form, and a confirmation of this is found in such Zend words as hurv-aoi-cha, where the s (necessarily changed to i before ch) is preserved. The Veda forms have only -a. This is further reduced to -i in the Sanskrit Neuter stems, as well as in the Feminine a stems, whilst the i and u stems merely have their vowel lengthened, except monosyllables, which develope a half-vowel and take au. In Zend -a appears as well as -do. The Masc. and Fern, i and u stems have the stem-vowel lengthened without any further addition, and the Neuters, as well as the Feminines in a, have only i added. In Grreek the stem-vowel in the a and a stems is lengthened only. Elsewhere the case sign is s. All these may be regarded as successive abbreviations of the original -as. 171. The following list exhibits the different forms : THE DUAL NUMBER 133 Sanskrit. Zend. Greek, m. azw-au or a azp-ao or a %7T7r(j) f. azw-e hizv-e X&pa n. dan-e dat-e dwpw m. pati paiti ? TTOGIS f. pri'ti afriti ? 7ro'prt-£ n. vari-n-i 'idpL-e m. sunu pazu VEKV-E £ hanu tanu ysvv-£ n. madhu-n-i madhv-i flk$V-E m.f. gav-au or a gav-ao or a &0-6 f. vach-au or a vach-ao or a 07T-B m. bharant-au or a barant-ao or a (pspovr-e m. azman-au or a azman-ao or a daifiov-E n. namn-i namain-i raXav-E m. bhratar-au or a bratar-ao or a cppdrop-E f. duhitar-au or a dughdhar-ao or a Svyar'sp-E n. vaclias-i e7te((t)-e THE INSTRUMENTAL, DATIVE, AND ABLATIVE DUAL. 172. The ending for these cases is in Sanskrit -bhyam, in Zend -bya, and in Greek -w. The stem-vowel -a is lengthened in Sanskrit and becomes -aii or 6i in Zend. In Greek the older form was - that it is for an original m, and making the word for ' one ' to be me'-nas, meaning ' little.' The origin of the gr. fjula, /jlovos, and the Armenian mino, is too obscure to be relied upon for the explanation of the Lithuanian we'nas. The second part of the Sanskrit ef-ka, ' one,' Bopp thinks^ is preserved in the following Gothic words as ha, viz. haihs, stem ha-iha, corresponding to the Latin ca-icu-s, cce-cu-s, ' one-eyed,' ' blind;' halts, from 142 NUMERALS Aa-litha, ' one^gfeed,' ' lame,' 'halt;' h&nfs, from ha-nifa, ' one- ^PSed ' (nifa = Scotch nieve, ' hand ') ; haZ6sj from /ia-liba, ' of one part,' ' haW !@& B (2.) The word for two is in Sanskrit and Zend dwa ; Greek Svco, Suo ; Latin <#mo, declined wholly or in part as duals ; Goth, tvai, and Anglo-Saxon twa, declined as plurals ; English two. In composition this word is reduced in Sanskrit to dwi ; Greek to Sr 9 Latin and Zend to bi, the d being dropped, and u changed to 6, as in 6-ellum for cfe^-ellum; Gothic tvi; Anglo- Saxon twi ; English twi, e. g. twi-ligkt. The adverbial forms are s. dwis, gr. 8ls, 1. bis (a. s. has twawa), e. tivice. In English the final e only indicates that c stands for s. 183. (3.) The word for three is in Sanskrit tri; in Zend and Gothic thri. In Zend the aspirate th is occasioned by the letter r. The declension is regular. The i becomes ij in Gothic before vowel endings, and in Sanskrit the Genitive traya-n-afm is formed from tray a. In both Sanskrit and Zend the Feminine is formed from the stem tisar. In Greek, M. F. rpsis, N. rpia; in Latin, M. F. tres, N. tria ; both declined as plurals. In Gothic ihri, and in Anglo-Saxon threo, as a plural. In English three. # 13^d (4.) Four is expressed in Sanskrit by chatwa'r, in weak cases chatur, and Feminine chatasar ; in Zend by chathwdr, weak cases chathru ; in Greek by rsrrapss, TEaaaps?, iEol. irscrvpss, Homeric iriavpss ; in Latin by qwattuor, quadru- in composition ; in Gothic by ftdvdr, fidur- in composition ; in Anglo-Saxon by feower, both separately and in composition; in English by fowr 9 sometimes for- in composition. The first three of these languages regularly decline this numeral ; but the Sans- krit and Zend insert n in the Genitive, as with vocal stems. In Greek the initial it corresponds to Sans, ch, w CARDINAL NUMERALS 143 but r is an unusual deviation. The 3 uble consonant results from assimilation, i. e. tt for Vt- In Latin the guttural initial stands for Sans, ch, and iv is vocalised to u ; whilst in the Ordinal quar-£us the whole is ab- breviated similarly to the English in for-ty. In Gothic the initial /is the regular representative of the gr. tt. 3.85- (5.) For five we have in Sanskrit pdnchan ; Zend panchan; Greek irkfiire^ 7rsvrs; Latin quinque; Grothic fimf ; Anglo-Saxon /£/; Engl. five. The final nasal of the first two is perhaps inorganic. In the Greek forms there are both tt and t for the Sanskrit ck, as in the number four. In Latin there is the guttural for ch, and also for the initial p. The Grothic follows the Grreek in having labial consonants. This wx>rd is de- clined only in Sanskrit and Zend, and there not in Nom., Ace, and Voc. 186* (6.) The word for six is in Sanskrit shash; in Zend khsvas; Grreek If; Latin sex; Gothic saihs; Anglo-Saxon six ; Engl, six — all undeclined. As sh is a derived sound and begins no other word in Sanskrit, it may originally have been preceded by the guttural which is preserved in Zend ; and Bopp thinks the other forms are transpositions for xes, etc. Hence s. kshash, . Latin Jcses, etc. The Greek aspiration is here, as in many other instances, for an original sibilant. Gothic h stands regularly for an older &, which in English has been reduced again to the earlier form, siks. A similar inversion to that here supposed occurs in a. s. acs-ian and asc-ian, e. ax and ask. 187. (7.) For seven Sanskrit has sdpta; Zend hapta ; Greek sirra ; Latin septem ; Gothic sibun ; Anglo-Saxon seofon ; Engl, seven. The declension of the Sanskrit and Zend shows that the stem is saptan. The Greek a also, as in the Accus. Sing, of the third declen- 144 NUMERALS sion, stands for a nasal preceded by a short vowel. In Latin m instead of n is an apparent deviation. In Gothic b probably stands, like the d in ftdvdr, for the vocal aspirate bh, since the regular law of consonant changes requires in Gothic dh for Sanskrit t 9 and bh for Sanskrit p. The English v, which represents bh, con- firms this view. Possibly the Grothic b was in this case pronounced like v. 188. (8.) The word for eight is in Sanskrit dshta, and the dual form ashtdu ; in Zend asta ; Greek 6fcrco ; Latin octo ; Grothic ahtau ; Anglo-Saxon eahta ; Engl. eight sh and s are derived from the original guttural. In Grothic and Anglo-Saxon h stands for the guttural aspirate, whilst in Engl, the aspirate is fully written gr A, though not pronounced. 189. (9.) For nine we have in Sanskrit ndva; Zend nava ; Greek svvka ; Latin novem ; Gothic niun ; Anglo-Saxon nigon ; English nine. The original stem appears to have been navan. In Greek the a indicates the final nasal, whilst the prefixing of a vowel and the doubling of the consonant at the beginning are not unusual. In Latin, again, m is in place of n, and in Gothic va is reduced to it, which in English is repre- sented only by the long quantity of the £. The g in Angl.-S. is irregular : perhaps it was pronounced y, and was developed out of i. 190. (10.) For ten we have in Sanskrit ddza; Zend daza ; Greek 8e/ca ; Latin decern ; Gothic taihun ; Anglo-Saxon tyn ; English ten. z is for an original guttural, which in Gothic is represented by A, and alto- gether dropped in Angl.-S. and English. In Greek a includes a final nasal, and in Latin m is instead of n. 191. The following is a list of the above numerals: CARDINAL NUMERALS 145 ,£3 ,4 c3 c3 ^4 c3 ^4 X d a o w Ci a ^d c < d d -4-> d CD u ?-( a> 3 .-^ IB eg o d G >^ ^ ■*3 f-l cd o a> 0) O -M <4=i d cd d £* c S3 d 1 o I .S "3 ^ d 3 CD ■d o •73 H3 f-i S3 d *o3 ■4-S > ■d d 03 p> *d 'd d 03 CD *j d 2 c3 ■ 13 w J3 •d in some instances, was pronounced bh or v. 194. The following are the numbers from twenty to a hundred : CARDINAL NUMERALS 149 — o IB To to fcJD to bC o oi c .3 £; c a 2 .to o o ad O "5 > a ■♦J -M <3 *n c 3 C 3 <3 *3 3 *3 c3 %J J 3 3 rr 'So '&0 i 14 .J!, •T r3 3 to o k ■* 8- t ■ -b -i i- . a ^ eg T3 ■> i > eg £> i 3 -4-3 CQ »*H *Q *3 P *a «*5 rQ -♦J •*> & ^ ^3 i eg a s a 2« 1 eg eg CQ eg *c3 veg BB veg veg i eg CQ 1 eg a eg ^3 "£> .>*> fl ^3 49 eg i "E T3 +3 i eg § a. 3 : veg NCg a a eg eg ■+-> -*J« cu ^3 eg CQ CQ eg s 55 ^ -S £ r^ eg cq eg veg » veg +* a eg •SJ C3 T3 "Z5 eg a a d> * a> g o E Ph PI o St *3 C4H o i o o *2> b *§. ^ OQ i=l P o a o 13 S o N a a a 1 a a J2 <0 a 5 7' a -5 a a i I a O FIRST AND SECOND PERSON SINGULAR 157 p 200. There are two stems in the above forms of the first person, one for the Nominative alone, and the other for the remaining cases. The Nom. has for the first person in Sanskrit ah, and for the second twa. The Latin and Greek seem to have preserved the original consonant in the first person; for the Germanic lan- guages have k and kh (germ. ch\ which presuppose g in the earlier languages. The Sanskrit h and Zend z, therefore, are corruptions of the original sound. Mo- dern English, i, has lost the consonant as compared with the old English ik, like the Italian io as compared with the Latin ego. The ending of the Nominative is s. -am, z. -em, gr. -ov in the archaic forms sy-cov, rovv, 1. o. It is lost in the other languages, and is a form which appears originally to have been confined to a few pronouns, i. e. ah-dm, tw-am, ay-am, sway-am, etc. The stem in the oblique cases of the first person is ma, that of the second person twa (or tu, changed to tiv in some cases). The Accusative, Instrumental, Ab- lative, and Locative coincide with the declension of nouns. The Dative and Genitive differ. The former cases have in several languages lost the case-ending. Even Sanskrit and Zend have Ace. md and Gen. me as well as the fuller forms. The k in Gothic and sometimes in Anglo-S. Ace. appears to be the remains of a demon- strative particle, as in Latin hi-c, hun-c, tun-c, etc. The ending of the Dative is -bhyam, which occurs with modified forms in the dual -bhydm and plural -bhyas of nouns. It is reduced in the Sanskrit first person to -hyam, in Zend to -by a and -6% in Greek to -i (unless the Greek forms are really Locatives), in Latin to -hi and -bi. It is quite lost in the Germanic lan- guages. 158 PRONOUNS The Genitive appears to be a reduplication of the stem, md-ma, ta-va for ta-twa, and this for twa-twa. The Greek and Latin are greatly abbreviated, and the Gothic as well as the Anglo-Saxon and English words have an adjective form, which is doubtless of later origin, and occasioned by the reduced reduplication being no longer understood as a Genitive sign. In English this adjective ending is again dropped before consonants, e. g. ' mine own,' but ' ray house. 5 In Greek the stem consonant a in the second person is the regular substitute for an older t, and many archaic forms preserve t, e. g. to/, to?o, etc. 201. The Plural forms of the First and Second Persons are : FIRST AND SECOND PERSON PLURAL 159 O o to S3 ftj ^3 ^O J^ (D £ fe i 09 S3 N3 00 v O 1 v o a> v 3 0> v 3 o i N 3 .i £ g 3 .i , and Latin Gen. as it. The s has become h in Zend, and spiritu-s asper in Greek. In Zend this pronoun occurs in the form of qha in THE THIKD PERSON 165* compounds (e.g. qha-dhdta, ( self-produced '), and oihwa as a possessive. In Latin i-pse is explained as being by inversion for i-spe, and sp for Sanskrit sw 9 as in sponte. A similar inversion occurs in the Doric yfrtv for cr(f)Lv. The change of Sanskrit w to Latin p is illustrated in 1. por-ta = s. dwd'ra, ' door.' The Eeflexive Pronoun is not preserved in Anglo- Saxon and English, unless it be in the word self, as in the German selbst, in which If as in loaf, and lb as in laib, are an expression for * body ' or ' person ; ' se-lf = ' one's own person.' Demonstrative Pronouns. TA. 204. The Demonstrative stem ta, Fern. td 9 is ex- tensively employed. Its forms are the following : 166 TRONOUNS 8 8 rS ~ c8 c3 c8 .S3 ^3 c3 _C3 r3 i O '3* ^3 r^ 09 BQ -2 v ■ „ • «. ' 1 <£> ^ /■N % -G 03 c3 O O r& rQ o £ S*> Irt * g S-, 3 * r ^ r I . . s s I e s aJ 172 PRONOUNS e. g. Dat. a-sma'i, Abl. a-sm&% Loc. a-smm, etc. It is used in the adverbial forms a-tra, ' here ; ' a-tas, c hence ; ' a-dya, e this day/ The usual Feminine form % has become iy, as in the Nom. singular iy-dm. Some cases combine a with na, e. g. Instr. singular a-ne'-na (a-na-i-na). There is the same compound in Latin e-ni-m, < for/ and the second part of it in na-m, ' for,' na-nc, ' now,' ne-mp^, ' surely,' nu-m, ne. The Greek words vi-Vj s him,' etc., vv, ' well,' vv-v> ' now,' as well as the Sanskrit nu, appear to contain the same root na. This root also furnishes the negative particles, s. na, gr. z/77-, 1. ne-, ni-, go. ni, a. s. ne, Old Engl. ne. The compound form ana appears in the conditional particle, gr. i-ti-dem, and s. i-ti 3 { so.' ov is a more usual repre- sentative of s. av than av : hence we have ov-v 9 ' then/ ou, ' not.' In go. BXL-h (a. s. a-c, ' but '), e. e-&6, the same pronoun appears, with a demonstrative par- ticle, as in mi-k, etc. Relative Pronouns. YA. The Eelative pronoun ya is declined in Sanskrit like ta. In the Greek o? 9 77, o, the spiritus asper, as in many other cases, is for Sanskrit y. In Zend this pronoun has a demonstrative meaning. It is preserved in go. ei and ja-bai for ja-ba, * if;' a. s. gi-/, pronounced yif, e. i-/. The ending is the same as in the Interro- gative i-bai, i-ba. Bopp supposes this ba (where b was probably pronounced v) to be for s. va, which brings it still nearer to the English if, where both syllables are greatly abbreviated, viz. i standing for ya, and / for va. Further, go. ja,-u, c whether,' for ja-v, and that for ja-va, preserves the same elements ; go. ju, ' now,' ' already,' 1. ja-m, are of the same origin. The Latin m for v corresponds to 1. mare for s. vari. Gothic ja-i, ja (e. yea), and jSL-h, ' also,' with an ending like the Latin que in quo-que, € also,' belong to the same root. Engl. ye-t is combined with the same particle as na is in s. net for na-it. Interrogative Pronouns. KA. 206. The root of the Interrogative has three forms in Sanskrit, ha, ku, hi, all probably modifications of the same original. The first is the more extensively applied, and was doubtless originally complete in Sanskrit as in Zend, Greek, and Latin ; for the Nom. and Ace. singular 174 PRONOUNS Neut. kat, instead of which kim is used, appears in kach-chit for kat-chit, and in Jcacl-adhwan, ' a bad street/ lit. ' what a street.' The ordinary Greek form would be /co, which is also preserved in the Ionic dialect, e. g. ko-ts, c once/ kg>-$« ' how ? ' Ko-rspov, ' whether ? ' ko-ctos, c how great ? ' ko-Io$, ' of what kind ; ' but in Attic Greek the consonant is changed to it, e. g. tto-ts, thS-s*, 7ro-rspov, iro-aos, Tro-Lo?. The Latin qui also, in some cases, belongs to the o and a declensions, which answer to the Sanskrit a and d. It has the Latin characteristic u after the guttural. In the Nom. singular Fern, the e in qua-e, as well as in ha-e-c, is obscure. The forms of the go. hva belong to the same root, as well as the Anglo-Saxon hwa, and English ivho (with an inversion in the order of the consonants) for hwo. The h in these languages regularly answers to Sanskrit k, and the w is developed similarly to the Latin u. The second form of the Interrogative, ku, appears in some adverbs, e. g. ku-£ra, s where ; ' ku-ks, c whence,' etc. The Latin forms which seem to belong to this stem, viz. cu-jus, c of whom,' cu-i, € to whom,' etc., are more probably abbreviations of quo-jus, quo-i, etc., where there is a similar curtailment of the first syllable as in the pronunciation of the English ivho for hwo, in which o is omitted and w sounded as oo. ku is used like kat in depreciatory compounds, "ku-tanu, c ugly- bodied,' lit. 'of what a body.' 1. u-ter has lost the guttural which is preserved in go. hvSL-thar, a. s. whse-dher, e. whe-ther. It is also wanting in 1. u-bi, u-n-de (but ali-cu-bi, ali-cun-de), u-nquam, VL-squam, u-spiam, usque. The third stem, ki, is used in Sanskrit to form the INTERROGATIVE PRONOUNS 175 Nom. and Ace. Singular Kent. ki-m, with the ending of Neuter nouns. That ki-t was earlier in use seems probable from 1. qui-c?. A Masc. ki-s is also employed in the Veda compounds na-kf-s, ' no one ; ' ma-ki-s, c let no" one.' This stem is used in hi-drza, ' like what ; ' kfy-a£, ' how much ; ' hi, ' for.' Here h is for k, as in Arc? compared with 1. cord-, ' heart,' A similar change appears in s. hy -as for hi-as, and this for ki-divas, gr. vSis with S inserted, 1. he-ri (hesternus), go. gi-8-tra, a. s. gy-sternlic dceg, e. ye-s-ter-day. Some confir- mation of this etymology of hyas is derived from s. iwas, perhaps for ku-as, 'to-morrow.' The abbre- viation of words causes less difficulty in these cases than the application of the interrogative pronoun in such a sense. The familiar designations of time, however, ge- nerally involve the strangest modifications of language, e. g. s. parut, ' in last year,' from para vatsa. qui-s and hi-c are Latin pronouns formed from this stem. The latter has h instead of the original guttural. The u is preserved in Gen. and Dat. h.VL-jus 9 hu-ie, and the origi- nal vowel lost as in cu-jus, cu-i. qui-a is the regular Neuter plural, and qiL-es the old Masc. Plural. Plautus has Gren. plural qui-um. The stem appears in ci-s, ci-tra, ' on this side,' retaining the guttural, but omit- ting the usual u. In hi-c, e this/ the affixing of the guttural at the end may have been a reason for chang- ing the initial to h. This final particle appears in different forms, as c 9 ce 9 que 9 quam 9 pe 9 piam. It is from the same interrogative ptem, and is often affixed to interrogative pronouns, ^e doubling of interrogatives destroys the interro- gation, just as doubling a negative destroys the negation, . quis 9 'who?' compared with quisque 9 ' whoever;' huid 9 'what/ compared w\\h.qviipjpe (forquid-pe), 'what- 176 PRONOUNS ever.' In Gothic uh is the representative of this parti- cle, e. g. hvas= quis, and hvaz-uh— quis-que ; hvo = quae, and A^o-^=qu8e-que. Does not the English whoso preserve a remains of this compound ? In that case s would be the Nom. sign, and the particle reduced to a vowel. As in the other examples quoted, who is interro- gative and whos-o indefinite. Of this compound form, e. whos = go. hvaz = 1. quis. e. 6 = go. uh = 1. que. e. whoso = go. hvazuh = 1. quisque. The G-othic Dat. hi-mma, Ace. hi-na, Neut. Ace. hi-ta, are from this same stem, with the usual h for an older k. The adverb hi-dre, a. s. hi-dher and hi-der. as well as the English hi-ther, has the comparative ending, go. he-r, in compounds hir (e. g. hi-r-i, hi-r- ja-ts,hi-r-ji-th, second person Sing., Dual, Plur. = ' come here '), and e. he-re, have the same ending, r, as hvar, ' where;' thav, ' there;' and exhibit the stem under consideration as hi, he. In Anglo-Saxon, he-r, thce-r 9 hivce-r have the same ending. In Greek tl-s, tl-vo-s, 6 who ? ' r/-y, Ti-vo-s 9 6 some one,' originally identical, have t for /e, like rsaaapss, ttevts, so that the guttural of this pronoun is in Greek represented by all three classes of consonants, e. g. k5)$, it, ' whence.' 1. perrmps (c)un-de, ' whence;' de, in-de, ' thence.' The full form appears in sivarga-tks ; L cosli-tus, ' from heaven ;' gr. iv-ro?, 1. m-tus, ' within.' da forms adverbs of time, e. g. ka-dk', ' when ? ' ta-d&V then ;' ya-dk\ ' when ;' eta-da', ' once ;' sa-da', 'always;' £a-da' -mm 5 'then;' i-da^iilm, 'now.' In the Vedas, i-dk'. gr. perhaps tto-te, 'once;' To-re, ' then ;'o-7*6 5 'when;' perhaps also rj-S?;, 'already,' for ^-^ = ya-da, with a long vowel, like rj-7rap for ya-krt. tham, thd, ti, form adverbs of manner, e. g. s. ka- tham, 'how ? ' it-thkm, ' so ;' anyd-tha, ' otherwise ;' £d-tha, 'so;' ya-thk, 'as;' sa?^d-tha, 'everyway;' i-ti, 'so;' d-ti (prep.), ' over/ 'beyond;' ic-t (prep.), j' upwards.' z. m-ti, 'so.' 1. i-tem, 'likewise;' au- ! tern, 'but;' i-ta, 'so;' aliu-ta, i-ti, i-ti-dem, u-ti, u-ti-nam, u-ti-que. Sanskrit d£i appears in 1. a-X-avus, ' primitive ancestor.' N 2 180 VERBS X. VEEBS. O) THE CONSTRUCTION OF VERBS. £09. The verb is the most important part of speech. It sometimes forms a complete sentence by itself, and no opinion or determination or wish can be expressed with- out it. It undergoes a greater variety of changes than any other word. In some American languages it is said that a single verb may appear in six thousand different forms. So great a variety does not exist in any Indo-European language ; yet in Sanskrit about a thousand possible forms, without including participles, may be assigned to one verb. In Greek, and still more in Latin, the number is very much reduced. In modern languages, the ana- lytical method has brought the verb to an almost Chinese simplicity. An English verb, for instance, does not assume more than half a dozen different forms ; and when we remember that the same word at one time had all the varieties which are found in the Sanskrit verb, we gain some idea of the great change which has been gradually made in the language of man. The method pursued has been the analytical, not the syn- thetical. The progress has been from the complex to the simple. History does not present to us a language growing out of a rude state, developing new -forms, and in process of time acquiring expansion and symmetry. On the contrary, we see that it is most perfect in its earlier history. We see its ornamental leaves gradually THE MIDDLE VOICE 181 fall off, its pliant branches broken, and ultimately but little remaining besides the gnarled trunk. What is thus illustrated in language as a whole, is specially illustrated by the changes which the verb has undergone. Our examples, therefore, will necessarily be drawn chiefly from the older languages, and be com- paratively few from those now spoken. A verb may consist of several distinct elements. The Latin verb amabantuv, for example, may be thus di- vided : am-a-ba-nt-u-r. Of these parts, am is the root which appears in all the forms ; amd is the stem of several tenses, etc.; ba is the sign of the Imperfect Tense ; nt is the sign of the Third Person Plural ; and r is the sign of the Passive Voice, whilst u is merely a connecting vowel inserted between the consonants nt and r, without affecting the sense of the word at all. Again, in amdrentur, re, like ba, represents the Im- perfect Tense, but differs from ba in denoting also the Conjunctive Mood. In verbs, therefore, besides the root or stem, there may be expressions for Voice, Mood, Tense, and Person. b) THE THREE VOICES. 210. There are three Voices, so far as the meaning of verbs is concerned. The Sanskrit has separate forms for all three ; in other languages there are only two forms, and in some only one. The three forms are the Active, Middle, and Passive. In Sanskrit the Active is called pavasmaipadam, f affecting another,' from parasmai, Dative Singular of para, 6 another,' and padam, from the root pad, f fall,' ( fall upon.' It is so called because the action expressed by the verb is not aimed at the acting person, 182 YEEBS but at some one else, e. g. ' I strike/ i. e. not myself, but some one else. The Middle is called in Sanskrit atmanepadam, ' self-affecting/ from atmane, Dative Singular of atman, 'self,' and padam. In this case the person acting is also the object acted upon, which, as there is no such verbal form in English, has to be expressed by the addition of a pronoun, e. g. ' I strike myself? The terms Active and Middle are clearly inappro- priate, for the Middle is as active as the so-called Active ; and the word Middle, in itself, conveys no notion of the thing intended, but merely that, as in Greek, where this term is chiefly applied, since it is in some things like the Active, and in some things like the Passive, it may be conveniently supposed to be half- way between them. But these terms are so widely used and so generally understood, that we shall adhere to them. £1I B In Sanskrit the Middle is distinguished from the Active by the endings affixed to the stem ; e. g. the Third Person Singular has -ii in the Active and -te (for ta-fi) in the Middle. The Passive has the same ending as the Middle, but ya is inserted between it and the root. Hence for Mid. -te we have Pass, -ya-te, e. g. from the root dwish, 'hate,' 3 Sing. Act. dwe'sh-ti, 'he hates' (some one else); Mid. dwesh-te\ f he hates himself;' Pass, dwesh-ya-te, 'he is hated' (by some one else). In Grreek the Middle form is also used to express the Passive voice, e. g. 3 Sing. Act. tvttt-u (for tvttt-s-ti), ' he strikes' (some one else) ; Mid. and Pass. TinrT-s-rai, 6 he strikes himself and 'he is struck' (by some one else). There are, however, a few special Passive forms. In Latin likewise one form serves for both Middle THE MIDDLE VOICE 183 and Passive verbs. It consists in affixing the reflexive pronoun se to the Active. We thus from the 3 Sing. Act. amat obtain amat-u-se, the u being introduced as a connecting vowel between the consonants, or tu may be a modified form of the pronoun used to express the Third Person Singular, which in Sanskrit is ti 9 in Greek ai. These latter instances make it probable that in Latin also, at an earlier period, a vowel followed such forms as amat Which of the above explanations of u is adopted will depend on the period in the develop- ment of the language at which it is supposed the Middle and Passive were formed, whether before or after the loss of the vowel in the personal endings. It is, further, one of the euphonic laws of the Latin language, that s between two vowels is softened to r, and hence from amatuse we obtain amature, and from this again, by the very common loss of the final vowel, amatur. Such forms ending in r occur both as Deponent verbs, which exhibit instances of Middle verbs (i. e. verbs with both Active and Reflexive meaning), and as Passive verbs. 212. The above will perhaps throw some light upon the formation of the Middle verbs in Greek. For whilst in Latin s between two vowels is usually soft- ened to r, in Greek, when occurring in the same position, it is entirely dropped. Indeed, the reflexive pronoun itself, in Greek, is already changed from as or d i o I v3 d ► i c3 cd i d d i c5 d ^ s«5 \C3 o d '-2 d .h es d d C3 '£* o O rd -a * CD rd bJD ^ THE FIRST PERSON 193 The primary forms always lengthen the characteristic vowel of the first conjugation to a 9 whilst the secondary forms leave the vowel short. The reason seems to be, that the heavier ending mi requires a stronger vowel to support it than the lighter ending m. The ending is made lighter in the secondary forms probably in consequence of the word being lengthened, here by the augment, and in the moods by the mood vowel. Though the connecting vowel of the first conjugation is not lengthened in the secondary as it is in the primary forms, yet, on the other hand, a connecting vowel is introduced in the second conjugation, where m would otherwise follow a consonant or half-vowel. In Greek the primary forms have -/u in the verbs cor- responding to the Sanskrit second conjugation, and co in the verbs corresponding to the Sanskrit first conjuga- tion, where the connecting vowel may be supposed to be lengthened, according to the analogy of the Sanskrit, and -in dropped. Thus we have "aTTj-fju, ' I stand,' and <£e/o-o), t I bear.' In the secondary forms -v occurs as the usual Greek equivalent for the Sanskrit m when final, e. g. in the Imperfect wtt^-z/, ' I was standing,' and s-fap-o-v, ' I was bearing.' In Latin this ending is almost universally lost, and -o left, like the o> in Greek verbs, in the primary forms, e. g. Present st-o, 6 I stand ;' fer-o, i I bear.' There are a few exceptions, such as su-m and inqua-m, and even there only m, not mi, is preserved. In the secondary forms m is preserved as in Sanskrit and Greek (v), e. g. Imperfect staba-m, ' I was standing,' and fereba-m, J J I was bearing.' In G-othic the primary forms present only one in- ! stance of the preservation of -m for mi in the substantive verb i-m, Q I am ; ' and thus the work of destruction has o 194 VEKBS gone further than even in Latin. In the other instances, e. g. bair-a, ' I bear/ etc., a is weaker than co and 6 in Greek and Latin. The secondary forms, however, pre- serve u for m, and in this vocalising of the consonant the language has again proved weaker than the Latin. Compare baiv-a-u, i I may bear,' with 1. fer-a-m. In Anglo-Saxon eo-m, and in English a-m, we have likewise the consonant of the original ending in a single word only. Elsewhere there is no trace of it. The following list illustrates the formation of the First Person Singular : THE FIRST PERSON 195 a 8 bO «v a •r-i >> fl rO rO S3 fc ° K fe> c3 *P c3 eg DQ « bo s 2 c5 o <3 bo .2 i c8 eg & a ^ 3 -g" "^7^ i •e- ^^i ^ i Ji n; 3 NCg. 196 VEKBS The First Person Plural. 2=15- The nominative plural of the pronoun in the Vedas is asme, probably for masme, from ma-snia-i, including the pronominal particle sma. If this be the origin of the verbal affix, it has in most cases been greatly curtailed, which would not be surprising ; for the addition of it as a whole would render the verbs very cumbersome, and we have seen that language re- sorts to many devices to prevent this. The Vedas have the ending -masi, and the Zend -mahi, e. g. v. dad- masi and z. dadernahi, ' we give.' They are the nearest approach to the supposed original masme. In Sanskrit the primary forms have -mas, sometimes -ma, and the secondary forms regularly -ma. The connecting vowel of the first conjugation is lengthened as in the singular, e.g. 6Mr-a-mas and bhdr-d-ma, 'we bear;' sdrp-d- mas and sdrp-a-ma, 'we creep;' a-bhar-a-ma, 'we were bearing.' The Greek has -/jls? in older and dia- lectic words, but elsewhere -/jlsp in both primary and secondary forms. The change of s to v is unusual, but not without example in other parts of the language, e. g. epTT-o-fie?, 'we creep;' (frep-o-fxe?, ' we bear;' i-cfrsp- o-[ie$, ' we were bearing.' The Latin likewise has but one form of this affix, i. e. -mus, e. g. serp-i-mus, fer-i-mvLS,fer-e-ba-mMS. In Gothic the primary forms, as in the singular, have suffered greater loss than the secondary. For the Sanskrit -mas of the former we find only m, whilst the Sanskrit -ma of the latter is preserved entire, e. g. bair-a-m, 'we bear,' aud bair- ai-ma, ' we may bear.' We see here the same relation between the endings and the previous syllable as in the Sanskrit forms bo'dh-d-mi and d-bodh-a-m ; i. e. the THE FIRST PERSON 197 stronger vowel ai sustains the heavier ending -ma, and the weaker vowel a the lighter ending -m. In Anglo- Saxon there is n in some forms, which may, however, be the third person used for the first. The English has entirely lost this affix. The following list illustrates the First Person Plural : P* •^ c3 CD CD ^ »* r^ ^ ^ m CD h S 2 r^ a a cd &JD 8 .a r fcJD 55 c CD CD CD CD CD H m £ £ ► ► ► V* *a c3 i-H c3 • r-f o CO pO cc c3 CO "■+3 c3 CO § I cc p" s 3 CO s CO 1 2 i CD ^ P CD > 1 < co CD ^ A uj $ 4* f 3. i O • r-l c*^ r^ e<3 cS <3 a i a i •i-i 2 i Ul c C r^ t >■ c Q rO ^ 198 VERBS The First Person Dual. 216. The Sanskrit has -vas for the primary and -va for the secondary forms, differing from the plural affix in substituting v for m, a change which takes place in other parts of speech as well. Indeed, the plural of the first personal pronoun itself presents an instance of it in the nominative : vayam may be supposed to be for mayam, of which am is an affix as in yuy-am. We have, then, may or ma + i, which would be written me, and differs from masme (ma-sma + i, the supposed original of the Veda form asme') only in the absence of the particle sma, which is uniformly absent from the nominative in Sanskrit, me would then correspond to te, Nom. Plur. of the third personal pronoun. If, then, v has resulted from an earlier m, the dual may be regarded as sprung from the plural. The first conju- gation lengthens the connecting vowel to d, as in the singular and plural, e. g. 6Mr-d-vas, 'we two bear;' bhar-e-V2i, ' we two may bear.' In Grothic the Conjunc- tive has -va like the Sanskrit, e. g. bair-ai-va, c we two may bear ;' the Present Ind. -6s from a-as, and this from a-vas, e. g. bair-6&, c we two bear,' of which, how- ever, properly only -as belongs to the affix ; the Prete- rite has -u for uu, and this for u-v, and again for u-va, e. g. maj-u, ' we both could.' Here, again, only one u, belongs to the affix. These cases also furnish instances of the stronger syllable supporting the heavier ending, and vice versa. In Anglo-Saxon the verbs have no dual forms. In Greek the place of the First Person Dual is supplied by the forms of the First Person Plural. In Latin there is no dual. THE SECOND PERSON 199 The following list illustrates the First Person Dual : Sanskrit. G-othic. English. s-vas sij-u 4 we two are * bhara-vas bair-os 6 we two bear ' vaha-vas vig-6s ' we two carry ' sya-va sijai-va ' we two may be ' bhare-va bairai-va c we two may bear ' avaha-va vag-u 1 we two were carrying ' The Second Person Singular. 217. The second personal pronoun is in Sanskrit twa (probably from tu\ a weakened form of which, -si, is employed as the verbal affix. The change of the conso- nant to -s is shown in the Greek pronoun, which is sv. The secondary forms further reduce this affix to s, and euphonic laws in some cases change si to shi. The cor- responding forms in Zend are -hi (for si) and -s, which is sometimes represented by -o as in the nominative of nouns, e.g. s. bhdva&i, z. bavahi, 'thou art;' v. krno'- Shi, z. kerenui&hi, 'thou makest;' s. ctbros (abravis), z. mrauSj ' thou spakest ; ' s. pra'zravayas, z. fra'zra- vayo (6 for as), ' thou didst cause to hear.' In Greek the full form -at is found in old and dialectic forms, e. g. Doric sa-cri, ' thou art ;' but usually the i is drawn back into the previous syllable, which is frequently the case in the G-reek language when the consonant be- tween it and a preceding vowel is not dropped, e. g. %r)a$a 9 ' thou saidst,' for sjyqs, ^> S3 a c3 S d 53 pj o O o o ^ pd r£ ^a -M +-> •4^» ■*a © 02 cp ' "T 1 02 • i— i 3 c3 ' 1 bo •i—i fed "3 r^ !> 0Q ^p .CD I CD O Co Co ^•o «a Co 1 Co b Ui •o Co O 1 b 1 i ■6- S3 c3 C3 Fh SI cS c3 r^ J> o I THE SECOND PERSON 203 The Second Person Plural. 218. The Sanskrit and Zend both have -tha in the primary and -ta in the secondary forms ; the Greek has -T£, the Latin -tis, and the Gothic -th, in both primary and secondary forms. The Latin has -te in the Im- perative. There are no remains of it in English. This ending seems to be formed from the singular of the personal pronoun, for the stem of the plural is yu. The aspiration in the Sanskrit and Zend may have originated in some euphonic peculiarity of these lan- guages. It appears as a vocal aspirate in the Middle forms dhwe and dhwam. We have thus both aspirates in the plural, as we have noticed them in the singular. The Gothic th is the regular representative in that lan- guage of an older t. The Anglo-Saxon has likewise th (dh), weorthadh, ' ye become.' In many cases this form is lost, and the Third Person is used in its stead. The s in the Latin forms is difficult to account for. Bopp suggests that it may be a remains of the particle sma in a plural form, twasme', similar to the asme' (masme) of the First Person Plural. The following list contains examples of the Second Person Plural : s 204 YERBS b0 CD CD CD CD b/D d •ft H %? •* c3 £ CD CD <3 rQ ^ o ^ fc* CD CD Ph ce 03 C3 S a CD CD CD CD i>^ N> ►» >^ O £ ^ 1 1 3 1 !S • rH c^O ■a .& • r— s "3 tl^ 5 "cS •£f r^ > cc ^ m CD '■v «3 «D •S3 <4-l c3 i % c3 5 i 02 OQ c3 c3 I ^ ** O (D O 4-3 •3 o O c8 bo 1 a • 1— t £P c3 m r^ O a o a h o 1 H m <*. 01 ^Uj >< -e- *c o i o o I N c3 rC +a -»? 1 *h m cS 5- ^ a a rJG N<3 1 ■+3 ^3 m GQ r^ m t *C3 THE THIRD PERSON 207 The Third Person Singular. 220. The third personal pronoun is ta. This, in Sanskrit, becomes -ti in the third person singular of the primary forms (-iu in the Imperative), and -t in the secondary forms, e. g. ddd&Xi, ' he gives ; ' dadya'% ' he may give.' In Zend the ending is -ti and d, e. g. dadhditi and daidhydd. In Greek the primary forms have for the verbs in -/u ai, and for the verbs in -co i, the t being dropped in accordance with a general eu- phonic law. The forms in which t is changed to cr, and so preserved, are older than those in which the t is lost. Both the first person and the third (-a> and i\ when compared with the earlier forms -/m and (-n or) -crc, show the destructive effect of time. The original -tl is preserved in the substantive verb sari, owing probably to its being preceded by a hard consonant. The ending has altogether disappeared from the secondary forms. According to the analogy of the other languages, and of the other personal endings in Greek, the secondary forms would end in -t, but this letter, when final, is almost universally dropped, e. g. SISgxtl, € he gives ;' tvtttsl, c he strikes ' (for tuttts-tl) ; Stuttts, ' he was strik- . ing ' (for stv7ttS't). In Latin both forms have -t 9 the primary forms having dropped the final vowel, e. g. dat (comp. s. dadati) and deX (comp. s. dadya't). The Gothic has -th in the primary forms, and nothing in the secondary, th is the regular Gothic representative of an older t. It thus agrees with the Latin in the primary forms, but has suffered greater loss in the secondary forms, e. g. itith., ' eateth,' and sijai, compared with 1. siet, 'he may be.' In Anglo-Saxon the primary forms have likewise th, which is lost in the secondary. The 208 VERBS older English agrees with the Grothic and Anglo-Saxon, e. g. eateth in the Present and ate in the Past tense. The th has in later times become reduced to s. This change of sound has become generalised in the German language, where s usually stands in place of th, which would be the proper equivalent for an English t and for an older d, e. g. 1. quod, e. ivhat, ger. was. The following is a list of verbs in the Third Person Singular : THE THIKD PEKSON 209 bO r^ m •i— i v* ra rCl o ■a 1 "3 3 rO '> kp *c§ 50 rQ w $ t« ® r- **» *a *o ■e. *£ f :§; ^ s '2 I -+^> *■£ •-3 •i-i c3 ■a i ^2 § > 2 ^8 3 I ' ■ A 3 fl <4 ^ ■ 4 - V g >-, r^J g J> o5 -^ 210 VERBS The Third Person Plural. 221. The Sanskrit has in the primary forms for the Third Person Plural -nti (-ntu in the Imperative) ; n is dropped in the reduplicated verbs of the third class, probably to lighten the word at one end, as the redupli- cation makes it heavier at the other. The secondary forms have -n only in some cases and -s in others. Where the stem ends in a consonant the connecting vowel a (u before s) is inserted, e. g. tishthanti, * they stand;' dadati, ' they give;' abharan, € they were bearing;' tishtMyus, 'they may stand.' The Zend has -nti and -n in the two forms, with a connecting e where necessary, e. g. histenti, ' they stand,' and his- tayen, ' they may stand.' In Greek the primary forms have -vri in earlier examples, with a connecting o after consonant stems. In later examples ri is changed to crt, whilst v is vocalised, so that by it the preceding vowel is lengthened or changed to a diphthong. Hence the forms -ovai, -acre, e. g. Doric StScW*, Attic StSovcrt, 6 they give ;' Doric lardvrh Attic laraah ' they stand.' In the Epic svtl and the Attic slal, a alone repre- sents the verbal root -e? ? whilst, in slal,.v of the ending is represented by i. In the Doric sovrt the connecting vowel o indicates the presence originally of the full verbal root, making iaovrc. The secondary forms have -i/ 5 e. g. sfapoi/i ' they were bearing.' The Latin has preserved -nt in both forms, e.g. stent, 'they stand;' stent, € they may stand.' The Grothic has -nd in the primary forms, in which d may have been pronounced dh (the Gothic representative of an older t), as we have seen to be probable in other cases; or, a.s Bopp suggests, the change may be referred to a preference for the combination nd. The Gothic Preterite has -u-n, THE THIRD PERSON 211 analogous to the Sanskrit -u-s in the Keduplicated Preterite, e. g. haihaitxm, ' they were called,' com- pared with s. a'sus, ' they were.' It is important to notice the relation of the nasal and sibilant (n and s) in these cases. We have had other instances in the Greek forms for the First Person Plural {-fisv and -jiss). We shall find another illustration of it in the Third Per- son Dual. The secondary forms have -na, in which they agree as usual with the Greek, but are inferior to the Latin in the loss of the t. No very satisfactory reason appears for the final a. e.g. sind, 'they are;' bed- rains,, ' they may bear.' In Anglo-Saxon nd is only partially preserved, and in modern English not at all. The following is a list of verbs in the Third Person Plural: p 2 212 VERBS a "P. Pi CO a "a Sh O a >-> V* a >> c3 13 1 ^ <3 d d Pi i 1 ^ <3 • i— t A i "3 c3 5h d b/D <3 1 1 c3 :z? pQ CQ r^ "£ CQ 1 -e 43 o3 i5 1 d i d 1 d d 3 ^ CQ >8 t> ►3 •5? ^» A ? =^ S or* rs O i . rr*- SI 31 ^ <3 - d 1 «? i & * N ,d r<2 > % -fc -* I * -R CQ I d s ? d ja ^8 THE THIRD PERSON 213 The Third Person Dual. &2>2. The Third Person has -tas for the dual in the primary and -tarn in the secondary forms of Sanskrit verbs, e. g. bhdratas, 6 they two bear/ and abhara\km 9 * they two were bearing.' m and s seem here to be inter- changeable, as we found n and s in the Second Person Plural. The long vowel in -tarn may be owing to a similar cause to that which produces it in the First Person Singular and Plural Present, i. e. -a-mi and -a-mas, though it seems as if there must at some time have been an addition to -m in the dual to give it this power ; for in the secondary forms of the first person -m has not this power, but leaves the connecting vowel short, as in abharam. In Zend the primary forms have -to, e. g. vakhsayato, ( they two (i. e. the cloud and the rain) cause to grow.' There are no instances of the secondary forms. In Greek we find -rov in the primary and -tt}v or -to)v in the secondary forms, e. g. (pspsroi/, 'they two bear;' sfyspLrrjVi 'they two were bearing;' fapsToov, ' let the two bear.' The nasal v (the usual representative of the Sanskrit final m) occurs in all these instances. In Grothic there is no Third Person | Dual. The following list contains instances of the Third Person Dual : 214 VEKBS 6fl o cS CD o i CD 3 o to bo a • iH ^ S>~> F-4 C3 | 45 'I 02 03 1 5 c8 c3 "So 'So > •So 'So 2 o 73 <3 13 O 3 J 3 oo OQ OO CO 1*1 b I O OQ o A ? o 60 o oo N 73 73 i 1 •rj «3 «D -H i 73 N3 ^ c3 c3 c3 73 73 73 c$ vc3 ■+3 \c3 >c$ 00 00 -*3 s r^ 1 p rJ3 vcS ■43 N<3 \c3 vc8 13 9 c3 c3 vc3 c3 c$ c5 13 13 13 73 13 73 73 13 13 s P THE WEIGHT OF THE PERSONAL ENDINGS 217 The root- vowel is preserved, though, in a shortened form (o), in the Greek plural and dual, whilst in Sanskrit the corresponding forms have entirely lost the root- vowel. This is accounted for by the fact that in Greek the endings are lighter upon the whole than in Sanskrit. A similar difference appears in the Imperfect. In the Aorist, however, in cases where the reduplication is not employed, the Sanskrit preserved the root-vowel long, the Greek short, as follows : o CO ■£ O 3 £0 I 3 I o o 1 I o £0 o O 1 i bvind-e and bund-urn. The same THE WEIGHT OF THE PERSONAL ENDINGS 221 cause admits or prevents guna in Gothic, just as we saw to be the case in Sanskrit and Greek, e. g. from the root vid: Sanskrit. Greek. Gothic. Ang.-Sax. Sing. ve'd-a, ' I know ' olS-a vait wat ve't-tha ola-Sa vais-t was-t ve'd-a olS-s vait wat Plur. vid-ma i8-fjt,sv vit-u-m wit-o-n vid-a-(tha) fa-re vit-u-th wit-e vid-us la-a-ai vit-u-n wit-a-n Dual. vid-va vit-u . vid-a-thus %t\-h-T£, c ye love,' 8t]\- os-ts, s ye show,' and Tip-ds-re, ' ye honour,' are for (f)tX- a(y)a-T£, S^\-a(y)a-rf, and r^-aya-rf. But the -co con- CONJUGATIONS 225 jugation also includes such words as Sdfcvco, < I bite,' which belongs to the ninth class, with the vowel, how- ever, short, e. g. Sd/c-vs-rs. In tvtt-tz-tz we have the addition of ta to the root, which is found in none of the Sanskrit classes, as well as several other forms which are peculiar to the Greek, showing that in the arrange- ment of their conjugations each language has acted independently. The general character of the -co con- jugation is that e(o before nasals) connects the ending to the stem, as its equivalent a does in Sanskrit. The -fjbi conjugation omits this connecting vowel, and adds the ending either (1) immediately to the root, with- out or with guna, e. g. J rd CO a i rd | OS 1 .s c? • i— I 1 ce 1 • r-f eg <£ k c3 «3 ck £ «3 ^ >> t>> ^ ^ T* ^ ■^ rd M rd rd rd rd rO rO T& r£> r^> r^ r& • *2 *P *2 , ^ *2 ^ • ^ be r-H Fh r— 1 I— 1 13 PS T-H H £ X; s ft rd +-> k i rd rO Q2 228 VEEBS /) FORMATION OF TENSES. 228. In the original construction of language it is evident that there was the intention of marking the difference of time by a corresponding variety in the forms of expression. In the Indo-European languages slight modifications in the forms of the same word were at first made to answer this purpose. What was the primitive meaning of the letters or syllables added to form tenses it is not easy in all cases to determine ; but it is possible to classify the forms actually in use, and by comparing them one with another to throw some light upon this difficult question. Whatever the primitive signs of tense were, they have in progress of time been gradually mutilated and de- stroyed, so that they are almost as entirely obliterated from the leading languages of Western Europe as the original inhabitants are from its soil. The analytical method has completely established its power in this respect over a large part of the English language ; but its operation has been so symmetrical as well as com- plete, that the English language has a considerable advantage over most others with regard to the precision with which the time of an action can be expressed. We mention the subject somewhat in detail here for the purpose of reference when examining each particular tense. 229. The three main divisions of time, viz. Past, Present, and Future, are presented in at least four different modifications, expressed by distinct forms of speech. For, in regard to each of these tenses, an action is represented as incomplete or complete at that time, Each of these incomplete or complete actions is either narrated or described. Thus there are furnished four ^ FORMATION OF TENSES 229 varieties for each tense. We will take the word 'go' as an example : Present Tense, Incomplete. Narrative : ( I go ' Descriptive : f Iam going' Past Tense, Incomplete. Narrative : ' I went ' Descriptive : ' I was going ' Future Tense, Incomplete. Narrative : ' I shall go ' Descriptive : ' I shall be going } Complete. ' I have gone ' 4 1 have been going ' Complete. 6 1 had gone ' ' 1 had been going ' Complete. ' I shall have gone ' ' I shall have been going' The difference in regard to time between the primary divisions is very obvious : e went ' cannot be used of to- morrow, nor < shall go ' of yesterday. The distinctions expressed by the secondary divisions are not so great, but still they are considerable. For instance, the Narrative forms may express a single action, and the Descriptive a succession of actions; the former may denote what is done at once, the latter may express the continuance of an action through a long space of time. Again, the idea expressed by the Incomplete is very different from that expressed by the Complete forms. I gain an advantage ' denotes that I am at present receiving the advantage, but may not have yet acquired ts full amount. On the other hand, < I have gained an advantage' denotes that I have no more of that ad van- 230 TEKBS tage to acquire, but that at the present time the gain is complete. There are many ways in which the difference between these four modifications of the same tense might be illustrated, but the above will sufficiently answer the purpose of showing how methodically and completely the analytical method expresses the relations of time. We have no less than twelve different forms for as many varieties of tense. The end is gained in modern languages by means of auxiliary verbs, which is charac- teristic of the analytical method; but in most of the languages under our present consideration the synthe- tical method is employed for the same purpose, and seems to have been employed exclusively in their original structure. It consists in making various inflexions of one verb answer the end which we have seen accom- plished by means of auxiliaries. The same primary distinctions of time, viz. Past, Present, and Future, belong to both systems. We begin with the Present, as containing the simpler forms. The Present Tense. 230. The incomplete and the complete are distin- guished as Present and Perfect, but no distinction of form is made in the older languages between the narrative and descriptive of the Present. The first of the above twelve forms has no auxiliary, therefore properly no indication of time. It is doubtful whether it is the same in the synthetical system. Some suppose the strengthened forms of the Present tense were intended to denote the present time ; others suppose that they contain no such meaning, but that the present is sufficiently indicated by the absence of all reference to THE PKESENT TENSE 231 any other time. We will first take an example of the Present Tense Indicative from the first Sanskrit conju- gation, in order to illustrate this difference of opinion : o b rH o QQ > "P *H o £ rH O r3 75 a bo bx) # bjo V '> V o o -r3 o 1 r* O o is .£P .S° o I H-=> r3 H-3 9 ^ 3 > r3 > '•3 *o 3 i 1 Ul i? i* £ 5 -^ T 1 * • i i -a > rd CD § >£ & >tf *T3 PI SI I i N c5 i 5 i •a ■a c3 i QQ bo 33 9 «iH .rH rd t f— i c3 c3 I 232 VERBS The Anglo-Saxon wceg-an or weg-an has little trace of the connecting vowel. In examining these forms, we see that between the root of the verb and the personal terminations different vowels are inserted, which, however, have probably all originated from the same. In Sanskrit they are a and (before Labials) a ; in Zend a, (before Labials) a, and e ; in Greek s and (before Nasals) o; in Latin i and (before n) u ; and in Gothic i and a. The origin of them all is the a which appears in Sanskrit. Is this merely a ' connecting ' vowel, or is it intended to express the Present tense ? It can hardly be the latter, because it is ,used also in the Past, i. e. the Imperfect tense. Again, in other verbs, syllables containing also a consonant are similarly in- serted, and with a greater variety in Greek than in Sanskrit. Are we, with Pott, to regard these as the grammatical expressions of the present time? If so, why are they also used in Past tenses (e. g. the Imper- fect), and why is such a variety of forms adopted to express one idea ? On the other hand, if they are not intended to express the relations of time, for what pur- pose are they inserted ? In regard to guna, Bopp thus expresses himself in the second edition of his Vergl. Grammatik, vol. ii. p. 378 : 4 1 cannot ascribe a grammatical meaning to the guna in the conjugation of Sanskrit and its sister languages, but explain it as resulting merely from a desire for fulness of form which causes the lighter vowels i and u to be strengthened — to be propped up as it were — by prefixing a, whilst a itself, being the heaviest vowel, needs no foreign help. If; as is done by Pott (Et. Forsch. i. 60), guna be regarded in the Present and Imperfect as an expression of the continuance of an action, a difficulty will be felt, which he also found, in the fact that guna is not confined to these tenses, but, in verbs with the lighter stem vowels i and u, accompanies THE PRESENT TENSE 233 the root through almost all tenses and moods, not only in Sanskrit but also in its sister languages in Europe, wherever this mode of forming diphthongs is preserved at all. As, for example, in Greek, \h7tw and (psvyu cannot free themselves again from the s inserted in the roots Xitt, bo bo 5h g g "£> Id Tab i o3 i ► 03 1=1 £ O rQ HP % +3 C3 I CD I Jh .»M a 5s b - bjo Pi •r-i s 1 OQ *3 f o3 s 1 c3 +3 PJ o3 > a 03 f c5 i 03 I > ^ THE AORIST TENSE 241 5 ^ 02 C3 § § iii iii CQ ^J - J= ' J ^ r S r^l r3 ,-3 ii mm # cq ^qq ^02 m 02 *02 *CQ CQ CQ "cQ *CQ *02 > >> ^^^ ^ ^ l>> s I It--? " -S 3 3 £ ^R ^ ^ 1 1 1 £ r3 r3 r3 o CQ | 02 | 02 CQ eM ^ • r-i M ^3 ^ nd v» vc3 vc3 vc3 c5 |' CO 1 f a 1 r3 ri r3 A Jk 1 02 CQ Ul CQ CQ CQ • i— ( "T •r-i 'T 1 •r-( ^ pd Jl ^ ^d rC T3 T3 TJ ^ nd T3 ^ o3 3 m 5 ft 242 VERBS The Zend is but scantily supplied with examples of the Aorist tense. The following, however, are instances in the first class: mahita, 'he spoke,' like the s. Mid. amansta, 'he thought,' from the root man; rasta, 'he rose,' from rudh. _ In Greek the 'Weak' or First Aorist furnishes nu- merous examples formed upon the same principles as those in Sanskrit, but apparently by an independent action; e.g.*8«£«, 'I showed' (i&»*-re sent tense, but the pure or verbal root. The fifth form affixes the personal endings to the root, e.g.d-da-m, 1 gave, from da; the sixth inserts a between the roof and the personal ending, e.g. a-Mar-a-m, 'I bore^ from bhar or bhr ; the seventh reduplicates the root and inserts a, e. g. d-papt-a-m, 'I fell,' from pat. In Greek the fifth is represented by such forms as **»* the sixth by such as i-^-o-,, and the seventh by such as g-rrsv-o-v. The last form is doubly represented in Latin by such In Zend we have an instance of the seventh form in fmrudhusha, 'thou didst grow,' from rudh, where the 244 VERBS initial u- is for the augment. The in* ances of the other forms are hardly to be distinguished from the xm- ^Tht following are instances of the fifth and sixth forms: o \ o i a- I 1 1 ^Uj «j -e- -e- •>Ul *iu o I l ~2- P v ttJ I a n © I I c3 i NC3 bo 35 ^ OJ Dor. saav 1 ^ <+H Ph M c3 -1-3 a A 02 1 1 c8 © A r a 5 c$ *f +-> «3 , 'they gave'), and in the Latin perfect (e. g. fuerunt, Hhey were'= fu-es-unt), at first sight suggests a doubt whether the a in the third person plural of the Pluper- fect implies its previous existence in the rest of the tense ; yet the Latin Pluperfect again removes the doubt, for it has er in all parts of the tense. The absence of the substantive verb in the middle and passive may have been occasioned by the greater weight of the endings. In Latin the uniform correspondence of the latter 254 VERBS part of the Pluperfect with the imperfect of the sub- stantive verb eram, etc., at once suggests that er is for Sanskrit as in dsmi. The general practice of dropping a preceding in favour of a succeeding vowel supports the division amav-er-am, in preference to amave-r-am, whilst in the Conjunctive amav-is-sem is may be as exact a representative of s. as, as er in the Indicative, for s. a often appears both as i and as e in Latin ; so that the difference between es-sem and the latter part of amav-is-sem would be unessential. Examples. (< I had struck ') (< I had loved ') Greek. Latin. Sing. sTSTv-si-v amav-er-am ststvc})-sl-9 amav-er-as sTsrv(p-st, amav-er-at Plur. irsTiKfr-sc-fjLev amav-er-amus ststv(J)'H-ts amav-er-atis sTSTvcp-scr-ap amav-er-ant The connecting vowel a is needed in all the forms in Latin because the • consonant of er is everywhere pre- served, but it is needed only in the 3 Plur. in Greek because a of so- is preserved only there. The Future Tense. 240 a. The Future is one of the principal modi- fications of time expressed by verbal forms. In Sanskrit there are two modes by which it is expressed. 1. The first consists in affixing the present tense of the substantive verb to the abbreviated form of the THE FUTCKE TENSE 255 Norn. Sing. Masc. of the participle in -tar,e. g. data' si (data-asi), < thou wilt give.' In the third person the sub- stantive verb is generally omitted, but the participle retains its full form in the dual and plural, e.g. : Sing, data'-smi Plur. data'-smas Dual, data'-swas data^-si data'-stha data'-sthas data'ras data'rau This participle is rarely used separately in a Future sense. 2. The second method of expressing the Future is by affixing sya, a form of the substantive verb, which does not exist independently in Sanskrit as a Future tense but is found in the a Potential sydm, syds, sydt, etc!, and in the Latin siem, sies, siet, etc. (later dm, etc ) and as a Future tense in era, evis, etc., for eso, esis, etc. I he latter forms also preserve the root vowel of the substantive verb, er = s. as. The change of 8 to r between two vowels is a common occurrence in Latin and appears in the imperfect tense of this same verb' emm etc. In era, erunt, the i is dropped, probabty through an imitation of the present tense, e.g.reao yap sr' e/xeXXeu sir' aXysa Tpwab, < for he was going to inflict further woes upon the Trojans ' (E. ii. 39) ; also s 258 VERBS with the- Present Infinitive, e.g. fxeXXovriV awry nroXuStrov, ™ X ™ °W as K W V % ^ ' 1kcov 1kST °> ' but there came to Mm from home, as he was going to pre- pare multifarions sacrifices, a herald.' iSi^-e. g. « 5* SeXvcret, roijov rsXnrjvavros, h t^ Svyarspa ravr V v foam™ V rvpavpk, 'but if, when he dies, the royal power shall come to this daughter ' (Her. i. _109> In the following examples of the synthetical luture, an archaic Latin word is given (faxo). In both Greek andLatin, £ and i must be supposed to be the remains of yd; or they are the ordinary connecting vowel, and ya is entirely lost, leaving only CQ 1 S3 +3 d «0 r3 t 1 rd CQ rd rQ-' 3 rd eg rd rO 1 rd i— i 03 32 5 3 ft CQ > t rC CQ c3 f J- CQ rd I > rd rQ 1 I CQ •t •* 4 3 S2 260 TEEBS In Grothic Ulfilas translates Greek Futures by the Conjunctive mood, just as in Latin the third and fourth conjugations have Conjunctive forms for the Future tense. But the auxiliary Q have ' is also employed in Gothic for the Future, e. g. taujan haba (' I have to do ' = I shall do) for TToir) gr. 8i8o-lr)-v 9 1. d-d-m, c 1 may give.' An older form exists of a few Latin words, which brings that language sufficiently near to the others ; e.g. du-i-m preserves i 9 and s-ie-m preserves ie, for the original yd. All take part, more 262 VERBS or less, in the gradual abbreviation of this Mood-sign, In Sanskrit it becomes yu for yd in the third person plural, and i in the first conjugation, forming e with the class vowel a. In Zend it is ya in several places. In Greek it becomes is in the third person plural '(8i$o-ie-v 9 'they might give *), and i in the co conjugation (rvTTTo-L-fM, 'I might strike'). In Latin it is reduced generally to % (s-i-mus, ( we may be *), and in the third person singular to i (s-i-t). In Sanskrit Atmanepadam (Middle) it is uniformly i ; in the Greek Middle and Passive, i (8i8o-L-firjv 9 tvttto-c-jlltjv)* In the past tense of the Gothic Conjunctive this syllable assumes three forms, in which it appears successively abbreviated from ya to ya, i (ei\ and % according to the general laws of the language: — 1. et-ja,-u, for et-ja-m, 'I might eat.' 2. et-ei-s, ' thou mightest eat.' 3. eM, ' he might eat.' In Anglo-Saxon it remains only in the singular as e y e. g. ic et-e, ' I might eat.' Besides these three different Moods, the same element -ya is used to form the Imperative in Old Slavic and Lithuanian, which adds further evidence that these verbal forms, if at all existing, were but imperfectly determined before the entire separation of languages. The following list will serve to illustrate these Moods, viz. the Sanskrit and Zend Potential, the Greek Opta- tive, and the Latin and Gothic Conjunctive. THE POTENTIAL, OPTATIVE, CONJUNCTIVE 263 o O $ ? m C3 r^ ca ? +3 a cS i •iH •iH "S • i— 1 0) •i-H • r-4 "S +2 <0> -»-i «© +3 c3 r-3 CD > "So •M <^ .M 3 2 3 'd 'd 'd 1 •iH o • e JO CD 1 5 •r-( to O O S N CD i cc T ? o o o «o ^ to -» 3 o CO CO CO rd .T3 0 a ? Sk rx o & r ~? v f -a 1 «8 O CO i O CO o CO CO CO co c3 | 3 i c3 "3 'I '3 • r-l "3 • i— I 1 o U Sh Fh H Sh Fh u %-i o •i— i • i-i •i-H • i—i •i-H • l-H • r-l c3 c3 c8 *3 c$ c3 c3 c3 *v ,2 rQ r^ rO rO r^ rO r^ J-* c3 o r^ >> c3 a C4 CO I i I i 3 «3 I +3 I a u a T* c3 I «x> I r-l v3 r3 rQ e3 3 a c8 I • r-l • r-» P>-» <3 c3 a H *H h 3 c3 3 pO rQ rO e3 a rd S I «D I Jh rS rO i r^ I I r3 § I r3 I I l rS THE IMPERATIVE 265 242. The tenth class in Sanskrit forms aye from aya-i (Mm-aye-s). In corresponding Greek verbs the c appears separately, e. g. rifido-L-9, l<\-€i, S^X-ou ; in Latin, a, e, i, e. g. am-a, hab-e, aud-1 ; go. % (ei), 6, ai, e. g. tam-ei, 4 tame;' laig-6, 'lick;' Aafe-ai, 'have.' The second person plural in Latin perhaps preserves the secondary form in not having the final s which appears in the indicative. The forms with to, both in Greek (-to>) and Latin (-to), to which the Veda -tat has some resemblance, may have been suggested and occasioned by the mood which appears as Let in the Vedas and as conjunctive in Greek. The long vowel, which is characteristic of this mood, appears also in the three first persons of the second Sanskrit conjugation. The following are examples of the Imperative : Sanskrit. Zend. Greek. Latin. Sing. 2. bhar-a bar-a ip-s fer 3. bhar-a-tu bar-a-tu — - Gothic. bair Plur. 2. bhar-a-ta bar-a-ta (f)ep-s-TS fer-te bair-i-th 3. bhar-a-ntu bar-a-ntu? DuaL 2. bhar-a-tam ep-e-Tov bair-a-ts 3. bhar-a-tam fop-i-row Let. 3. S. vah-a-tat eX'S-TO) veh-i-to THE CONDITIONAL 267 The Conditional Mood. 2A*km There is a Conditional in Sanskrit, though it is rarely used. It appears to be formed from the auxiliary Future in -syam, etc., by prefixing the aug- ment and substituting the secondary for the primary personal endings, a-syam, etc., e. g. Future cZasya/mi, * I shall give ;' Conditional acZasyam, ' I would give.' The Latin language seems to have resorted to a similar method in forming the Imperfect, Perfect, and Plu- perfect Conjunctive, for the endings -rem, rim, and es-sem have the same relation to the Future in -ro that the Sanskrit a-syam has to -syami, e. g. : a OQ ■♦3 a # 02 +3 43 •5" CD 1 1 CD i «D 1 > 5 > > > «£ I 02 • i-H 1 O *£ i 1 • i-H 1 • iH •i g 1 CD CD CD i 1 CD i CD 5 > > > > > fc> s » yy> Class 5. star-ya-te str-nu-te' str-no'-ti jj sir or star, 270 VERBS This form of the Passive, which is probably from the root yd = ' go/ is general in Sanskrit and extensively used in Zend, but is rarely met with in the other languages. The Latin verbs morior and fio present remains of it in the i. In fio we have the root (/u) also in a diminished form, as in the instances noticed above. In Gothic, us-ki-ja-na (of which us is a pre- position), Q enatum,' presupposes a Pres. Ind. ki-ja, abbreviated from kin-ja, like s. ja'-ye for jan-ye. That the root yd = ' go 'was the origin of this Passive formative, is rendered the more probable from the fact that in Bengalee and Hindostanee the Passive is ex- pressed by the auxiliary verb ' go/ just as in English it is expressed by the verb c be,' e. g. kord yd% e I go made' = 'I am made.' The same verb is also used as an auxiliary m such Latin Passive forms as amatum iri, etc. Causal Verbs. 2a4L7. The most common form of Causal Verbs is that in which ay a is found inserted between the root and the personal endings. It corresponds exactly with the tenth class, e. g. Mr-aya-mi, ' I cause to make,'* from the root kr or kar. This formative may have sprung from the root % c wish ; ' for the expression ' I wish (you) to make' may easily have come to mean 6 1 cause (you) to make.' The verb ■' have' in English has been similarly appropriated to a special meaning, for 6 1 have (or ' have had ') a house built ' is the same as ' I cause (or * have caused') a house to be built;' and even 'I have built a house' has come to mean ( I have had a house built.' This formative appears as -ja in the first class of Gothic weak verbs, e. g. s. sad-ayk-mi, go. sai-ja, 'I CAUSAL VERBS 271 set,' from s. sad, go. sat, e. sit So lag-ja, * I lay,' from lig-a, ( I lie ; ' nas-ja, ( I make well,' from nas, € to get well ; ' sanqv- ja, ' I cause to sink/ from sinqv-a, * I sink ' (of myself) ; drank-ja, ' 1 cause to drink/ e. drench, from drink-a, 6 I drink.' In the English word drench, although no part of the original aya is pre- served, yet the influence of y is apparent in the change of k to ch. The difference of formation between the causative and non-causative verbs is indicated in English in a few cases by the former having the heavier, the latter the lighter vowel, e. g. set, sit; lay, lie; drench, drink. In Greek, KaXiio, ' I call ' < I cause to hear,' seems to be a Causative of k\v-(o, 'I hear;' the root acAi; has become kolX, and e represents the aya in Sanskrit, as it generally represents aya of the tenth class in verbs in eco. In Latin the long vowels a, e, i of the first, second, and fourth conjugations represent the s. aya of the tenth class, and amongst them are some Causatives, e. g. nec-k-re, ' cause to die;' sed-h-re, ' set,' ' cause to sit;' plor-k-re, ' weep,' ' cause to flow ; ' mon-e-re, 6 cause to think ; ' sop-i-re, f cause to sleep.' The following instances will show the agreement between the Sanskrit and the Latin : 272 VERBS 02 3 a 02 •*? 02 P o3 * «D «3} VJ5 N05 vg >^ v3 1 1 ' • • 8- 1 &• Ph 9< P* -• ••-» -H • II I I I l3 &• Ph a. Ch Oi p^ n * ^» ^> ^ \CJ q VW V35 \c$ VS5 \3 9< 3 & & 1 Ph &• &• <^ a3 *£? < ^ «$ K £ *£ £ 8 £ £ ■ £ £ % ™ QQ £n 02 02 £ Ph After roots in -a, p is inserted in Sanskrit and k (c) in Latin ; e. g. yap-dya-mi, * I cause to go,' from ya, is the same as 1. jac-i-o. The i is reduced in quantity, and the whole assimilated to the third conjugation, as in capio. So also s. bhav-aya-mi, ' I cause to be,' is the same as the 1. fac-i-o, where c is for v, as in vixi (yic-si) from vivo ; jnap-dya-mi, ' I cause to know' DESIDERATIVES 273 (from jna); 1. doc-e-o. In rap-dyd-mi, e I cause to go,' ' move ' (from ra) = 1. rap-i-o, the Latin pre- serves p. In some cases I is inserted in Sanskrit Causatives, and corresponding forms are found in Grreek, e. g. s. pdl-dya-mi, from pa, ' to rule.' So in Grreek, fidWco for ./3a\-y-cD, ' I cause to go,' from fid (in s-/3t]-v, etc.) ; aTsXkco, ' I cause to stand,' for crreX-y-co, from ara (in s-gttj-v, etc.); ld\\o), 'I cause to go,' for lak-y-oo, from ya (2 Aor. of ltj/jll has i-, where the spiritus asper is for y), like the Latin jac-i-o. Desideratives. 2€L8. Desideratives are formed by inserting between the reduplicated form of the root and the personal endings sa in Sanskrit, gks in Grreek, and sci in Latin, of which the following are illustrations : 274 VERBS O i o **» b i 3 8 i I 45* 8 i 3 o i -3 s* h I UJ !* b i I o b i I I i I b I 9 Ul i «3 *a* * i* *.< b b b v ? 4 v 3 ^ a £ ^ ^ E- i O b i -3 •B H CQ r*4 — } • i— c3 e3 • i-i s rG +i +3 i 3 > CQ CQ , ttolttvu-co, ic. If the root begins with a vowel, the whole syllable \ 3 repeated in the reduplication; but if the vowel is hort, it is lengthened in the root syllable, e. g. atat, rom at, 'go ;' azaz, from az, ' eat.' To these correspond Dme nominal forms in Greek, e. g. ayeoy-os, 'guide;' ycoy-svs, ' remover.' But the vowel is shortened in the >ot of oviv-rjfJLt, oTrnrT-zvo), drn-dXkco, and not length- led in dXaX-dfa, £\eX-/£a>. Those which begin with a msonant, if they end in a nasal, do not lengthen the K)t vowel, e.g. s. jahgSim, from gam, 'go;' g£. f Ka\i,§aiv T 2 276 VEKBS ' I shine brightly ;' go. ganga, ' I go ;' a. s. gauge. If the root ends in a liquid, this is either changed to a nasal or displaced by a nasal, e. g. chahchal, from chat, c move;' pamphul, from phal, ' burst;' chahchar, from char, 6 go.' In Greek, 7ri/ui7r\rj/jLL, ' I fill ; ' Trl/jLirprj/jbt, ' I set on fire/ etc. Sometimes, however, the liquid remains unchanged in Greek, e. g. fiapfiaipco^ ' I glimmer ; • fiopfiopvfo, ' I rumble,' &c. A few Intensive forms have been discovered in Zend and in Latin, as gingrire, ' to gaggle.' k) DENOMINATIVES. 2>90. Denominatives, i. e. verbs formed from nouns {de nominibus), have, in Sanskrit, ya, aya, sya, or asya inserted between the nominal stem and the personal ending. If, however, the nominal stem ends in a vowel, it is dropped, e. g. s. kumar-iiysi-si, ' thou playest,' from kumdra, ' boy ; ' sukh-ky&si, ' thou delightest,' from sukha, ' delight;' yoktr-slysi-si, 'thou embracest,' from yd'ktra, 'band;' ksham-aya-si, ' thou endurest,' from kshama', ' endurance.' The examples in Latin are numerous in the first, second, and fourth conjugations, where d, 6, i represent Sanskrit aya, e. g. laud-Si-s, ' thou praisest,' from lauda, 4 praise ; ' can-e-s, ' thou art grey,' from canu-s, ' grey ; ' sit-i-s, ' thou thirstest,' from siti-s, ' thirst.' In nouns of the fourth declension, u shows more tenacity in keep- ing its place in the Denominatives, e. g. fluctu-si-s, ' thou wavest,' from fluctu-s, ' wave.' In Greek also many of those verbs which resemble the 1 Oth class in Sanskrit are Denominatives, i. e. such as end in -aw, -sco, -oco, -a£a>, -/§». In these cases the s. aya is represented by different letters. Examples are Tifjias (for Ti/jLa-e-is), 'thou honourest,' from tl/jlt] ; 7toXs/jlsls DENOMINATIVES 277 (for 7rd\s/A£-€-i$), 'thou warrest,' from ttoXs/jlo-s; BtjXol? (for SrjXo-e-is), 'thou makest manifest,' from &7X0-S, 'manifest;' dyopd^sus (for dyopa-^e-is), ' thou attendest market/ from dyopd, ' market f 8sl7tvi£sls (f or Ssnrvc-^e-t?), ' thou feastest,' from Shttvo-v, ' feast.' Denominatives in era, XX, correspond to Sanskrit forms in ya, having assimilated y to the preceding consonant, e. g. dyysXXoy for dyysX-yoo, from ayyeXo-s; fiaXdacrco for /naXafc-yoe, from fiaXafco-s. This 2/, instead of being assimilated after v or p, is transferred to the preceding syllable in the form of i, e. g. psXaiva) for fjusXav-yco, from jx^Xav ; /caSalpco for Ka9ap-yco, from /caSapo-s. The Denomi- natives in -eu-o) perhaps arose from an interchange of half- vowels, i. e. w (u) for 2/, e. g. SovXsvsc, ' he is a slave,' from SoOXos. In Gothic j (y) represents the formative element, e. g. audag-j-a, ' I call happy,' from audaga, f happy ; ' gauv-j-a, ' I make sad,' from gaura, ' sad ; ' skaft-j-a, \ I create,' from skafti, ' creation.' The stem-vowel of the noun is dropped, except sometimes u, as in Latin and Greek, e. g. thaurs-j-a, ( I thirst,' from thaursu, | dry ;' and iifar-skadv-j-a, 'I overshadow,' from skadu, c shadow.' In some cases the Gothic has 6 for aya, like the Latin a, e.g. jisk-6-s, 'thou fishest,' from fiska, ' fish.' In a few cases n is inserted before the 0, and the stem- vowel preserved in a weakened form, e. g. hdrin-6-s, from hdra, e. whore = gr. icopr), Dor. /ccopa, ' girl,' I wench.' Some Passive Denominatives in Gothic are formed by -n, which seems to be connected with the Sansk. Passive Participle in -na, continued in the Germanic strong verbs, e. g. e. broke-11, etc. Go. Passive Denom. (which may be compared with such Passives as and-bund-n-a 9 278 VERBS ' I am unbound ') are full-n-a, c I am filled,' Act. full-ya, 6 I fill,' from fulla, c full.' Perhaps such forms are the source of the double meanings in such English verbs as ' I fill/ both intransitive and transitive = f I become full ' and ' I make full.' Some Denominatives in Sanskrit, like some Causatives, prefix p to ay a, lengthening the stem-vowel, e. g. artha-pdyd-mi, from artha, ' thing ; ' satya-pdya-mi, from satya, ' truth.' Some Sanskrit Denominatives correspond in mean- ing to the Desideratives, e. g. pati-ya'mi, from pati, 6 a husband : ' putri-ya'mi, from putra, ' child.' So in Greek Savar-tdco, from Sdvaro-9, ' death,' etc. Such Latin forms as ccenaturio, ending in -turio, are verbal derivatives, but equio (equ-i-o) is a Denominative with a Desiderative meaning, from equu-s, f horse.' Denominatives with a Desiderative meaning are formed in Sanskrit also by sya (asyd) 3 e. g. vrsha-syk'-mi, from vrshd, 'bull;' azwa-syk'-mi, from azwa, 'stallion;' madhw-SL&yk'-mi, from madhu, ' honey.' Latin forms in -sso, denoting imitation, correspond to these Sanskrit forms in sya, the y being assimilated to s, e. g. attici-SSO, from atticu-s ; grceci-s&o, from grsecu-s. Latin In- choatives in -asco, -esco, also resemble the Sanskrit forms with -asya, e. g. jlamm-e&CO, ' I begin to flame,' from flamma, ' flame.' 279 XI. DEKIVATION AND COMPOSITION. a) DERIVATIVES. NT or NTA. 251. The Present Participle Active is formed by inserting nt or nta between the verbal stem and the personal ending. The weak forms, however, generally drop the n in Sanskrit, while in the Grreek and Latin the n is preserved throughout. The Grothic has nda, where d was pronounced probably as an aspirate, for the older mute generally becomes an aspirate in the Grothic language. Examples are s. Gren. bh&ra-tSL-s, but Ace. bhdra-ntsi-m ; gr. Gren. cf)spo-i>T-o9, 1. fere-nti-s, go. baira-n&i-ns. In Grreek the i is lost, as is seen from the plural M. cj)spo-vT-£9, N. (j>spo-vT-a. The parallel forms in Latin show that the i is preserved in that language, e. g. fere-nte-s, fere-nti-a. The feminine is formed in Sanskrit by nt + i in the first conjugation, and by t + i in the second, e. g. vdsa-nti F., < inhabiting,' from vas ; daha-nti F., ( burning,' from dah ; sa-ti' F. ? ' being,' from as. In Grreek vtl8 for s. nti 9 like rpiS for s. tri 9 occurs in $spa7r6-VTi$ 9 Gren. $spa7r6-VTi8-os. In these Parti- ciples the Masc. form is used also for the Fern, in Latin. In Gothic the feminine has ndi for ndhi, with the addition of an inorganic n, e. g. visa-ndei F., ' remain- ing,' stem msa-ndein. The Participle of the substantive verb has lost the 280 DERIVATION AND COMPOSITION entire root in both Greek and Latin : Gen. 6vr-o$, 1. enti-s, contain only the formative ovr, enti, and the case-ending os, s. The Epic and Ionic form eovt indi- cates the previous existence of s, as usual in such a position, and t become cr through the influence of i, whilst the i itself then disappears, ovaa is therefore for o-vrt-a. The same element appears in the s. Auxiliary Future, e. g. Ace. S. da-sya-nta-m, ' about to give ; ' in the gr. First Future, Sco-ao-vr-a, and First and Second Aorist, e. g. Xvaa-vT-a, ' having loosed,' \iiro~vT-a, ' having left.' These letters therefore indicate the Participial character of the word in which they are inserted, and have no reference to any peculiarity of tense, for they are applied to Past, Present, and Future alike. WANS. 251 a. The reduplicated preterite in Sanskrit forms an Active Participle by inserting ivans (vans), wat (vat), or ush, according to the different weight of the case, be- tween the stem and the personal ending, e. g. Ace. S. rurud-wk'hs-am, Loc. PI. rurud-w&t-su, Ace. PI. rurud-ii&h-as, from rud, 'weep.' -ush-i forms the feminine, e. g. rurud-ushi. In the Greek perfect, which is also a reduplicated tense, this formative is em- ployed in M. and N. -or, probably for an earlier Pot = s. ivat, but applied alike to both heavy and light cases, e. g. Ace. S. TSTv(f>-OT-a, ' having struck.' The feminine is vi-a for vcrt-a = s. ushi, cr in Greek being usually dropped between two vowels, e. g. T£TV-vL-a. In Latin DERIVATIVES 281 the word securis, ' axe,' for sec-usi-s, means ' cut- ting;' and in Gothic, Nom. PL M. ber-u&j-os means ' those who have borne/ i. e. ' parents.' The Latin adjectives in -bs-us may also have been formed by ivans, to which os corresponds very much as or (6s) does to ans in the comparative forms, e. g. meli-6r- ; so fani-os-us from fama, though the s is not softened in the latter as it is in the former case. it MANA. 252. The Participle for the Present, Perfect, and Future Middle ends in mdna in the first conjugation, and ana in the second. The latter seems to be only a weakened form of mana, just as the plural of the first personal pronoun has a for race of the singular, and for the same reason. Hence also, in those parts of the first conjugation which need a lighter ending than the others, viz. the tenth class and the reduplicated preterite, the form ana is used. The accent is on the last syllable of the ending, where, in the indicative, it would be on the personal ending ; otherwise it is on that part of the stem where the indicative has it. In Greek the perfect has the accent on the last syllable but one of the ending ; elsewhere the accent follows the general rule. The fol- lowing are examples : Present. Perfect. Future. dacl-kna-s tutup-knk-s GREEK. da-sya-mkna-s Si8o-/Z6Z/0-5 f T£TV/JL-fJL€l>OS h(o-a6'[xevo-s Again we see that the participial mana, fisvo, has nothing to do originally with the expression of tense, 282 DERIVATION AND COMPOSITION for, like the form previously noticed, viz. s. nta, gr. vr 9 1. nti, it is used for present, past, and future alike. In Sanskrit, the Passive Participle differs from the middle in the same way as the passive voice generally does from the middle voice; i.e. yd is inserted in place of the class characteristic, e.g. Mid. dad-ana-s (for dada-mana-s), Pass. di-yd-mana-s. In Zend and Greek the same form serves for the passive as is used for the middle, e.g. &a/re-mane-m = gr. fap6~nevo-v>* vaze-mne-m = gr. ixo-fieuo-v. In the first of these instances the vowel a is reduced in Zend as in Greek ; in the second it is dropped, as we shall find to be the case in Latin. In using this form for both voices, the Zend already anticipates the practice in Greek, whilst the Latin almost exclusively appropriates it to the passive voice, i. e. in the second person plural. The Latin forms for the second person plural passive, amd-mini, c being loved,' for ama-mini estis, ' ye are (being) loved,' etc., are clearly instances of this parti- cipial form in the nominative plural masculine, and were probably at first used with the substantive verb as another participle is used in the perfect passive, i.e. ama-mini estis like ama-ti estis. ife-mnu-s is also a passive participle of alere, whereas a middle or active meaning is more suited toV&rtu-mna-s smdVoltu-mna. In the last three instances the vowel a is dropped, as was found to be the case in Zend. The element thus appropriated to the formation of the middle participles also appears in Substantives and Adjectives. In Sanskrit man in strong and man in weak cases forms substantives with an active or passive meaning, e.g. zush-man M., 'fire' (the drier); ve- man M., 'weaving loom' (weaver). 1. fe-raiii-a and e. wo-maxi may be similarly formed from this root (s. ve DEKIVATIVES 283 or ive), denoting c weaver ' — ' spinster ' is still employed of the unmarried females of the family. (See Sec. 256.) har-i-man M., ' time ' (that takes away, hurries) ; dhar-i-man M. 3 ' form ' (what is borne), as 1. /or-ma from fer-ie, and e. be&r-ing from bear. The neuters are more numerous than the masculines: dha-man N., 4 house ' (what is put or made, so e. buHd-ing = what is built); Mr-man N., 'deed' (as 1. fac-tum from facere, and e. deed from do); ro'-man N., 'hair' (what grows). Adjectives in -man are rare : iar-man, * happy.' In Greek there are Abstract Substantives in -fMovrj, e.g. cfrXsy-fiovr/, ' inflammation ;' xap-\iovr], ' pleasure.' Masculine Substantives in -/nov (lengthened in the Nom. S. to /jlcqv) are irvsv-jioov^ ' lungs ' (breather) ; Sal-fjLCQP, 6 god ' (shining one). These have the accent on the stem; but others, with a connecting vowel, have the accent on the last syllable, both as in Sanskrit, e.g. yy-s-fiCDV, ' leader.' The same formative also occurs as /jisv (Norn. /jlt]v\ e. g. TroL-firjv, c shepherd ' (feeder). The long quantity of a in mana is preserved in some instances, as in fcsvS-ficov, Gren. KevB-fjicov-os. The same is the case with /jllv in such words as Ka-ylv-os^ 'oven' (/caico, /cdco); va-fxlvr]^ ' contest' (a contending), connected with s. yudh, ' contend.' Neuter Substantives have assumed the form -/mar (reduced to pa in the Nom. S.),e.g. Gren. S. iroirj-fiaT-os, irpa Y^ar-os*, 6-vo-lJ ' going to give ' and ' giver.' In the 1 st- and 2nd persons of all genders it is joined with the sub- stantive verb to form the Future Tense, but in the 3rd person it is the future tense without the substantive verb. In Greek there is no participle in this form, but there are Nomina agentis in rrjp, rrj, and Top, e. g. Bo-rrjp, ' giver ; ' fiaxn-TTj-s, ' fighter.' In Latin tur-o forms a Future Participle, and tor forms Nomina agen- tis, e.g.da-tur-u-s, 'going to give;' da-tor, 'giver.' The Feminine forms are s. tri, gr. rptS and rpca, 1. trie. The abbreviation from tar to tr is caused by the additional weight at the end, and the addition of B in Greek and c in Latin is what we have seen take place in other instances. Examples of these feminines are s. da-tri', ' the giver ;' gr. \7]pd-rcop, member of a brotherhood), 1. fra-Xer, go. &ra-der, a. s. brd-dher and bro-dher, e. bro-Xhev, is referred by Bopp to the root bhar, ' bear/ denoting the ' bearer' or ' supporter' of the family, s. swds-ar (t lost, but the vowel lengthened as in Latin) (gr. aSs\t]) 9 1. sor-dv (t lost and o lengthened as in Sans., and s between vowels changed to r), go. svis-t&v, a. s. swus-ter and sus-Xer, e. sis-Xer, is referred by Pott and Bopp to the root su, su, ' bear,' ' bring forth.' s. duh-i-Xar, gr. Svy-d-Trjp (7 for ^) (1. filia), go. 288 DERIVATION AND COMPOSITION dauh-ter, a. s. doh-ter, e. daugh-ter, seems connected with the root duh, and to mean ' milker/ i. e. of cows. Bopp gives c suckling ' as the meaning, which is ob- jectionable because it is quite as appropriate to son as daughter, whereas this word is used to distinguish one from the other. 2iS7 B From the same element (tar) arise the neuter -tra and the feminine -tra, the former occurring in many words, the latter in few. They have an instru- mental meaning ; e. g. we'-tra-m, ' eye ' (means of guiding, from ni) ; iro'-tra-m, ' ear ' (means of hear- ing, from zru) ; gra'-tra-m, < limb ' (means of going, from go) ; ddhz-tvk, ' tooth ' (means of biting, from dahz). In Greek the forms of the suffix are -rpo, -rpa, Spo, Spa. The change from a mute to an aspirate frequently occurs in Greek without any apparent cause : apo-rpo-v, 'plough,' from apoco; Ka\v7r-r pa./ covering,' from Ka\v7r-Tco ; ap-^po-v, ' limb,' from apap-icric(D ; ftd-Spa, ' step,' from (3a-lvco. In Latin ara-tYVL-m 3 4 plough,' from ara-re ; fulge-tvk, ' lightning,' from fulge-sco. The aspirate, which is only occasional in Zend and Greek, is regular in Grothic, Anglo-Saxon, and English : go. rnaur-thr (stem maur-£Am), e. mur-ther, from the root mar, 1. mor-ior. go. blds-tra,, ' sacrifice,' which may be inferred from blos-treis, is from blot-an, ' to sacrifice.' Without the formative the Anglo-Saxon blot means ' a sacrifice.' In e. laugh-ter, from laugh, the preceding consonant prevents t from being aspirated ; and the same cause perhaps operated in slaitgh-ter, the gh being pronounced at first. The consonant (ch) is pronounced still in both the corresponding words in German, i. e. lacAen, schlac/^ten. e. wea-ther is con- nected with the s. root wa 9 ( blow.' go. hulis-tr (stem hulis-^ra, the t not being aspirated in consequence of DERIVATIVES 289 the preceding s) ; go. fo-dr, ' sheath,' a. s. /o-dher (stem fo-dra, d = dh), coDnected with the s. root • pa, 'to contain/ and so equal to s. pa'-trSL-m, ' container/ ' vessel.' e. ru-dder, ger. ra-der, is perhaps connected with gr. pcL-%po-v, ' stream,' from psco, though with a meaning resembling the 1. re-mus, ' oar.' The dd is for the aspirate dh. go. hlei-thvSi, ' tent,' a. s. hlce-dre, e. la-dder (d and dd for the aspirate). With the change of r to I, which is of frequent occurrence, this formative becomes in Greek t\o, iXrj, 9\o, SXy, e. g. oxs-rho-v, ' carriage ;' £%£-tA?7, ' handle ;' 3v(T-%\o-v> instruments used in the worship of Bacchus ; yevs-SXr), ' birth.' In Grothic the corresponding form is thla, e. g. ne-thlsi, a. s. nce-dl, e. nee-dle (d instead of the aspirate). TA. 258. The Participle of the Perfect Passive is formed by -ta 9 F. td. It takes the accent, e. g. tyak-tk-m, Ace. S., ' left.' In Grreek Verbal Adjectives (but not participles) are similarly formed, e. g. tto-tos, tto-ttJ, hto-to-v, 'drunk,' with a passive meaning, and the formative accented as in Sanskrit ; but also tto-to-s, ' the act of drinking.' In Latin the suffix is employed in forming the Participle as in Sanskrit, e. g. da-tu-s, da-tSi, da-tu-m, from da-re. In the above instances ta is affixed immediately to the root. Sometimes i is inserted : s. prath-i-tk-s, ' stretched out,' from prath ; Igr. c/ce\-e-TO-9, c dried,' from c/ce\-\qo ; 1. mol-i-tu-s, \ ' ground,' from mol-o. The characteristic of the tenth class (aya) is usually preserved in an abbreviated form, e. g. s. pieL-i-X'k-s, ' oppressed;' gr. (pik-rj-TO-s, 6 beloved,' 1. om-a-tu-*. u 290 DERIVATION AND COMPOSITION Adjectives are also formed from substantives by i-ta 9 e. g. s. phal-i-tti-s, € possessed of fruit ; ' gr. a/jLatj-i-ro-s, 6 possessed of waggons ;' 1. patr-i-tu-s, ' fatherly.' The Latin neuters in -e-tu-m are probably formed from deno- minatives of the second conjugation, e. g. ar&or-e-tu-m, from arbor-e-sco, Perf. -e-vi. Abstract Nouns are formed from adjectives by the feminine ta, e. g. s. zukla-tk, ' whiteness.' go. tho, Nom. tha : niuji-thSL, c newness.' a. s. dhe> dh (also written d) : hcel-dJi, leng-dhe ; gebyr-d, ' birth.' e. heal-fh, weal-th, leng-th, dep-th, bread-th, heigh-th (the last has in recent times dropped the h and become heigh-t). The Latin juven-ta belongs to this class of derivatives. The Sanskrit stem yuvan is in some cases contracted to yun. The same contraction takes place in the Latin comparative jun-ior. In Gothic the abstract jun-da, is formed from this abbreviated stem, having d, as in so many instances, for dh ; whilst the English you-th pre- serves the aspirate, but loses the final vowel and reduces the stem to you for yu. Abstracts in tat-i occur in the Vedas, with which may be compared the Greek t^t, the Latin tat, tut, and the Gothic duth, where the d is again for an aspirate, e. g. v. arishta-tkt-is, ' invulnerability,' from drishta ; gr.TjXiSio-rr]-^ Gen. rjXt^Lo-rrjr-os, * folly,' from qXlSios ; 1. juven-tu-s, Gen. juven-tat-is, ' youthfulness,' from juvenis; sterilitas, Gen. sterili-tkt-is 9 ( barrenness,' from sterilis ; go. ajuk-duth-s, € eternity ; ' ma7iag-d\Lth-i, « abundance ' (2 Cor. viii. 2) ; miJfeiZ-duth-i, ' great- ness.' Similar words in Latin have tud-in, perhaps from tut by changing t to d and adding in : Gen. S. magni-tudin-is, ' greatness.' Abstracts are formed from adjectives and substantives by twd added immediately to the stem, e.g. s. amrta- DERIVATIVES 291 twa-m, ' immortality/ from amf ta. In Gothic the word thiva-dva,, ' serfdom,' occurs (with d for the aspi- rate), from tbiva, 'serf.' In English, -dom resembles the Latin forms in tu-din, from tut, in having added a nasal and changed t to d. In words which are not abstracts, the Gothic preserves the formative as thva, e.g. /r£-o, vrj : asti-vQ-s (for asfivos), ' venerated ;' crtcrj-vr), ' tent ' ('covered in'). The meaning shows that these forms are of participial origin, tsk-vo-v, ' child ' (lit. ' brought forth '), has the accent irregularly thrown back. In Latin there arep^-nu-s, 'full ' (lit. 'filled') ; r^-nu-m, 'dominion' (lit. 'ruled over'). Here again the parti- cipial meaning is very obvious, although the form does not appear among the participles in any Latin conjuga- tion. Many words have deviated more or less from the original meaning, e.g. mc^-nu-s, ' great ' (lit. ' grown'); c%-nu-s, 'worthy' (lit. 'talked of,' or 'pointed to'). A similar deviation appears in corresponding Greek forms, such as azywos, ' venerable ' as well as ' vene- rated.' The formative, which thus appears to have only a fragmentary character in the older languages, extends in Gothic and Anglo-Saxon throughout the strong con- jugations, answering to the strong or 'irregular' verbs in English. The syllable na, Nom. n, however, is joined to the root by means of a connecting vowel a or e> U 2 292 DERIVATION AND COMPOSITION whereas in the languages which we have already noticed it is added immediately to the root, e.g. go. frugr-a-n-s, a.s. graf-e-n, e. grav-e-n; s. bhug-nk-s, 'bent.' In some Gothic Participles used as nouns, however, the syllable is added immediately to the root, e.g. the adjective us-luk-nSi-s, ' open ' (lit. ' unlocked '), and the N. substantive go. bar-n (stem bar-na), a.s. bear-n, ' child' (lit. 'born;' so the Scotch bair-n). 26 0. na is used like ta to form Possessive Adjec- tives from substantives. In this application it also takes the connecting vowel % e.g. s. phal-i-nh-s, 'possessed of fruit,' from phal-a ; maM-na-s, ' covered with dirt/ from mal-a. There are also feminine forms in ni 9 denoting c wife of/ preceded by a, e. g. Indr-k-ni (r changes n to n\ ' wife of Indra.' gr. irsS-i-vo-^ ' flat,' from7TfS-/o^; ovcot£-£-z/o-s (for arKorsa-t-vo-s),' dark/ from a kotos, st. c/coTS?. Some adjectives, like %v\lvo$, XlSivos, have the accent thrown back. An instance of this occurs in the Sanskrit word zfiig-i-na-s, ' horned/ from zfng-a. Feminine forms in vrj, preceded by a>, re- semble the Sanskrit feminines in d-ni, e.g. 'AfcpLcri- (D-vrj) ' daughter of 'A/cptVt-o^. Those in -aiva are for -a-via, e.g. Ssaiva, \vtcaiva. In Latin many words have i before -nu for the connecting vowel, as in Grothic, e.g. stagn~i-nu-s, from stagn-u-m, ' pool ;' bov-i-mi-s, from stem bov (bos, ' ox 3 ). After r the vowel is omitted, as in English, e.g. ebur-nu-s, from ebur, 'ivory;' ver-nu-s, from ver, ' spring. 5 Even in Sanskrit i occurs, e. g. sam-i'-na-s, ' yearly/ from sama/, ' year.' In Latin also a different vowel, viz. a, occurs, but it is perhaps of the same origin, e. g. oppid-k-nu-s, from oppid-u-m, ' town ;' it!om-a-nu-s, from Eom-a. There are also feminines in -na and -nia preceded by 6, e.g. Bell-b-na,, md(r-6-na, Vail- 6-nia. In Grothic na, DERIVATIVES 293 Noin. n, is preceded by ei ( = i), e.g. silubr-ei-n-s, ' of silver fjill-ei-n-s, 'pelliceus ;' liuhad-ei-n-s, * light ;' sunj-ei-n-s, i true.' In English the connecting vowel has become e, and after r is lost, e.g. ivood-e-n, (70M- e-n, leather-n. The later practice has been to use the substantive, without any formative addition, as an adjective, as in ' a silver knife,' ' a gold watch.' A trace of the feminine formative is found in the Anglo- Saxon gyden (gyd-e-n), gyd-e-ne, ' goddess.' A few Abstracts are formed by na, F. rui, e. g. s. yaj-na-s, ' honour ; ' trsh-nk', ' thirst ; ' swdp-nSL-s^ ' sleep;' gr. vir-vo-s, T^X' vr l^ ^ som-nu-s, rop-i-na. TI, NL 261. Feminine Abstracts are formed by ti and ni, which are probably from ta and na, e. g. s. yuk-ti-s 3 6 union ; ' uk-ti-s, ' speech.' Some have a before ti, which is a connecting or class vowel. In these words the root is sometimes accented, e. g. dr-a-ti-s, ' fear ;' ra??i-a-ti-s, 'the god of love;' vah-a-ti-s, 'wind.' gr. Xy~ TL ~ 9 > pr}-Ti-$, (f>d-TL-?, afiTrcD-TL-s. Elsewhere t be- comes <7, except when preceded by cr, which itself has come from a dental, e. g. irla-TL-s from ttiS, tsv^c9 for Qv/c-cn-s. -via has been formed from gi, as -rpia from s. tri. It is seldom added to monosyllabic stems, e. g. $v-(TLa> hotcL^a-ata, cnTra-ata. These latter re- semble in appearance such words as aSavaa-la from dSdvar-os, which are not participial but nominal forma- tions. Some in Grreek, as in Sanskrit, have a connecting vowel before ai, and the same accentuation, e. g. vs/x- e-O't-?, svp-€- ' bread' (lit ' feeder '); /u-ni-s, ' rope ' (lit. < binder '), etc. TU, NIL 262. As the interrogative pronoun appears in three forms, viz. ka, k% ku, so the formatives which we are now considering appear as ta, na ; ti, ni ; and hi, nu. The Sanskrit Infinitive ends in -turn, of which m is the sign of the accusative case, e. g. cZd/ -tu-m, ' to give ;' sthd r -tVL-m, ' to stand ;' d£-tu-m, ' to eat,' from ad. In compounds m is dropped, e.g. tyak-tJi-ka r aias, ' desirous to leave.' The Sanskrit -twd is an instrumental case of tu, formed by adding d, and is employed like Latin gerunds, e. g. tan drsh-tws,', ' after seeing him ' (lit. ' with seeing him'); ity-uk-twk f 9 ' after so speaking' (lit. 'with so speaking '). The Dative case of abstract nouns is sometimes used in the sense of the usual Infinitive or accusative case, e. g. jfam-anaya, c to go ; ' cZd?^-anaya, 6 to show.' The abstracts in ana, which appears in the above examples, are also employed in the Locative Singular in the same sense as the dative, instead of the infinitive, e. g. anwesh-ane, ( to seek.' The same form of infinitive 296 DERIVATION AND COMPOSITION becomes general in the Germanic languages : go. an, gib-em, e to give ;' a. s. gif -an, old e. giv-en, modern German geb-en. A similar formation appears in Greek. The oldest form of the Present Infinitive is -fisvai or 8-/j,svai, which is a Dative (that is, a Locative) form of an abstract in -/nsva or e-^isva. Another form is -jjlsv or s-julev, which results from dropping ai of fjusvac or s-fisvau Again, there are forms of the infinitive in -vat, where the first syllable of/ievai is dropped, unless -vcube the Dative (Locative) of va, as fjusvac is of fjusva, and thus be of dis- tinct origin. This appears the more probable, because forms in -vat occur as early as those in /jlsvcu. The common classical form is -ziv for ssv, and this from s-fju-sv, e. g. slTT-e-fievaL, siTr-e-fiev, slrr-elv, firj-vai. &63. An Aorist form in the Vedas with the meaning of the Infinitive ends in se ( = sai), e.g. me'-she (s changed to sh by the preceding vowel), ' to throw.' This strikingly resembles the Greek First Aorist Infini- tive in -aau, e. g. \v-aai, c to loose ; ' Tvy^rat (rv7r-crai\ 'to strike;' Selgcu (Setfc-aca), f to show.' Both seem identical with the Latin -se after consonants, re after vowels, e. g. esse, s to be ;' dic-e-re, ( to say.' s assimi- lates a preceding t in posse, from pot-se, and is itself assimilated to a preceding I and r in vel-le and fer-re, unless these are for vel-e-re (gr. ffovX-o-fiai) and fer-e-re (fyep-o-fjuai), in which case the only assimilation is that of r to I in velle. The Perfect Infinitive in Latin in archaic forms is also -se, e. g. consum-se, admis-se (for admit-se, from admit-to). As the Latin -perfect gene- rally corresponds in origin to the Greek aorist, these forms agree exactly with the aorists fjv-ai (for (f>rjv-aai, the a being dropped after nasals) and Tv^rai (for tv tt- aai). The more common Latin forms in -sse are of later origin. DERIVATIVES 297 The Passive Infinitive in Latin was probably at first -sese, changed by the laws of euphony to -rere, and afterwards to -ri-re and -ri-er. Hence we meet with the older forms ama-rier, mone-rier, dici-er, molli- rier. The last syllable -er is wanting in the ordinary forms used in the classical works, and in conjugating the Latin verb ; therefore ama-ri, mone-vi, dici, molli- ri are regarded as Passive Infinitives, though in fact they exactly agree in form with the Active Inf. ama-re, etc., with the exception that the final -e was changed to -i, for euphonic reasons, when followed by re {er). There is also the further difference in the third conju- gation, that -re of the Active dice-re is dropped. The so-called Passive Infinitives, therefore, are not Passive in form. The older termination er for re, and that for se, is the same reflexive pronoun as is employed in the form of r in the rest of the passive conjugation. This Veda Infinitive in -se occurs in sentences where its usage is exactly parallel with that of the Dative of Abstract Substantives in -a, and thus illustrates the meaning and force of the Infinitive, e. g. ve'mi twa pushann rhj-ase, ve'mi stb't-ave, ' I come to glorify thee, Pushann; I come for praising (thee).' rhj-ase Inf., and sto't-ave Dat. S. There is also a Veda Infinitive in the form of the accusative singular following the verb zak, ' to -be able.' The English Infinitive without 'to,' after 'can,' resembles this construction, though the Infinitive has not the form *of a case, e. g. apalupan (for apalupam, Ace. of apalupa) net zaknuvan, ' they could not (to) destroy.' This verb zah is even used in the Passive form itself, yadi iak-ya-te, 'if it can,' lit. 'if it (is) can(ned).' A double Passive occurs even in Latin, e. g. ut comprimi 298 DERIVATION AND COMPOSITION nequitur, c how incapable he is of being restrained' (Plaut. Rud. iv. 4, 20) ; forma in tenebris nosci non quita est, ' the form could not be distinguished in the dark ' (Ter. Hec. iv. 1, 58). The Future Infinitive Passive in Latin has likewise the auxiliary in the passive form : amatum iri. 2>@4- In later languages the expression of the Pas- sive seems to have been felt to be difficult. The me- thods resorted to by Ulfilas, in his Grothic translation of the Scriptures, are various and singular ; but in none of them is there a really passive form of the finite verb. The Passive Perfect Participle in th (originally -to) is employed, but not as a past tense. The relations of time are expressed in the substantive verb connected with the participle, e. g. Mark xiv. 5, maht vesi . . ,/ra- bugjan, * was able to be sold.' The word able, how- ever, must be supposed to be Passive, as if mayed could be formed from may like viade from make, and thus the maht vesi might not only express the past tense of rjhvvdTo, but also the passive voice of TrpaSrjvac ; for the Grothic word by which this latter is rendered is active, frabugjan, 'to sell,' instead of 'to be sold.' skidds (th-s) is also used in a similar way to express the pas- sive of the accompanying infinitive : e. g. Luke ix. 44, fjLsWst, TrapaSlSoaSai, ' is going to be given up,' is ren- dered skulds ist abgiban, 'is necessitated (Pass.) to give up' (Act.) for ' to be given up.' The Active In- finitive is also used for the passive when no passive form accompanies it, and the only method bf indicating the voice is in the agent being expressed in the dative or instrumental case ; e. g. jrpbs to SeaSfjvai uvtolSj 4 in order to be seen by them,' is in Grothic ' in order to see by them,' du saihvan im. This use of the active for the passive infinitive, without the least indication of the DEKIVATIYES 299 difference, occurs extensively in the modern German language, e. g. es ist zu sehen, for ' it is to be seen.' In English it is rare, e. g. ' it is yet to do ' for ' it is yet to be done.' The preposition to, which is generally put before the infinitive in the Germanic languages, properly governs the dative case. The Gothic, however, from the habit of dropping the final vowel, has lost the case-sign. The old Saxon and Anglo-Saxon have e (for ai) as a reduced form of the dative ending ay a. The n is also doubled without any apparent reason. Possibly it was intended to regulate the pronunciation of the preceding vowel, as in English the consonant is doubled after a short vowel when a suffix is added beginning with a vowel, e. g. hitt-ing for hit-ing, merely to prevent the first i from being pronounced long in the participle. A similar reason may have caused the forms o. s. and a. s. farann-e for faran-e. As the Dative case, among other things, also expresses the goal at which an action aims, so the Germanic- Infinitive, consisting of the preposition to with a Dative case, was at first confined to this meaning, and after- wards extended by analogy to others. Thus, in Gothic, siv, du meljan, ( to write,' i. e. writing). This mere action without any reference to aim or purpose is often expressed by the Injbiitive which depends upon another verb, e. g. ' he began to go,' i. e. he began the act of going. So, in Lu.iv. 10, < will enjoin 300 DEKIYATION AND COMPOSITION upon his angels (the act) of taking care of thee ; ' go. du gafastan thik, rod BtafpvXd^ai as. In Gothic this Infinitive is also used without the preposition, sometimes in rendering a Greek infinitive, e. g. galeithan, airaX^&lvy ' to go out ; ' and sometimes in rendering a Greek noun, e. g. Luke iv. 36, varth afslauthnan allans, iysvaro ^Jafx^os' sirl iravras, s there came amazement upon all.' Verbs denoting an act of sensation often take two objects, the first expressed by an objective case, and the second by an infinitive without the preposition. The infinitive in this case denotes only the act, as in the dependent infinitive noticed above, e. g. 6 1 saw him go,' i. e. going, where him and (the act of) going are two objects seen and combined in one idea. The Greek has the participial form where the Gothic has this infinitive: John vi. 62, sav ovv Sscopr/rs top vlov rod av&pooirov avaftaivovra, jabai nu gasaihvith sunu mans ussteigan (avafiaivovTa, ussteigan^ ' ascend up '). Where the govern- ing verb does not denote an act of sensation, the nature of the governed infinitive is sometimes not quite so obvious. Yet in such cases as Matt. viii. 18, haihait galeithan siponjons, Q he commanded go the disciples,' i. e. the disciples to go, it is clear that both the act of going and the disciples were objects of command. So also Lu. xix. 14, ni vileim thana thiudanon, 'we refuse him to rule,' him and ruling arealike the objects of refusal. 265. In Greek we meet with the forms s-fievcu, -B-fjbsv, (e-ev) slv (Ion. 7}v\ -£-v (Dor.), and vai. It is not difficult to trace all the others to the first, -s-fievai, which appears in the oldest documents ; for -s-fisv merely drops the final diphthong, (s-ev) -eiv further drops fi, and regularly contracts the two vowels to st, the Ionic shows another and less usual contraction to 77, DERIVATIVES 30 1 the Doric drops s from e-ev. Again, the original form after vowels is -fjusvai, which, by dropping /is, be- comes vai. There is nothing in these changes very different from the usual course of abbreviation to which language is subject. The loss of at all at once from fjLsvat is the least likely ; but in Homer, where the full form occurs, very often the diphthong is elided in e/jl/jlsv for sfifisvai, and the elision of it, however rarely it occurs, indicates the possibility of its being dispensed with altogether. The derivation of -/jlsv from fisvav is also less improbable than that there should have been different sources of the infinitive present of slfit in the existing forms of s/jb/jLevai, sfifjusv. If fievai be referred to the Sanskrit -mane, dative singular of -man, it would make the Greek form ai fuller than the Sanskrit, which is very unlikely to be the case ; but if it be referred to -mdndya, dative singular of -mdna, it would make the Greek a + i for a + y, with the loss of the final a, a more regular repre- sentative of the Sanskrit form. The first part jjlsv answers to s. man, as the participles in -fisvo-s to the s. part, mdna-s. Like other participial terminations, it was probably employed to form abstract nouns, and hence the Greek infinitive is a case of an abstract noun like other infinitives. Bopp refers to the Middle forms, s. me, se, and gr. fiat, gcli, as showing that the s. e may be represented in Greek by ai ; but in this case there was a consonant between the vowels, which will account for the otherwise unusual preservation of the original vowels a l in Greek. (Sec. 2X2, p. 183.) 266. The Passive Infinitive has the form aSai, which Bopp explains as consisting of the reflexive pronoun Si-8-LO-s, ' perishable ; ' a/jL(})d-8-LO-s, ' public ' (to be seen); sfcra-8-io-s, ' broad' (to be spread out). Con- sonant stems also occur, e. g. irdWa (for irak-ia = s. yd Fern.), 'ball' (to be hurled). In Latin exim-iu-s, 4 eminent ' (to be selected). Denominative Adjectives are also formed by ya y e. g. s. cZk'-ya-s, ' heavenly,' from div ; hfd-ya-s, 6 hearty,' ( affectionate,' from hrcl ; z. ydir-ya, ' yearly,' from ycire ; gr. irdrp-io-s^ ' paternal,' from irarrjp ; reXs-io-s, ' perfect ' (for rsksa-i Vs), from riXos ; ovpav-LO-s, e heavenly,' from ovpavos. In Latin they are less numerous than in the above languages. But there are Appellatives as well as adjectives, e. g. patr-ivL-s, ' paternal,' from pater ; Mar-iu-s from Mars, Non-ia, Fern, from nonu-s. It seems more natural to refer the names of conn- DERIVATIVES 305 tries to adjectives of this kind than to substantives, e.g. Gallia, Germania, to Gall-iu-s, German-iu-s, from Gallu-s, Germanu-s, i. e. Gallia (terra), etc., < the land of the Gauls,' etc. Thus in more recent times we have Eng-land,Deutsch-land, named from the people as a whole, and not from an individual. The corresponding Adjectives and Appellatives in the Gothic language end in Masc. -ja, Fern, -jd, whilst some add n and form -jan, e. g. alth-ja,, ' old,' from althi ; leik-jsi, ' physician ' (leech), from leik ; fisk-jsin, ' fisher,' from fisks (stem fiska). 268. tav-ya forms Future Passive Participles. It takes the accent, and is preceded by guna, e. g. s. yok- tavya-s, f to be joined,' from yuj ; da-tsivya-s, 'to be given,' from da. gr. -ts o-s, e. g. So-reo-s, 6 to be given,' for 8o-tsFo-$ from 8o-tsFlo-s ; 1. -tivu-s, e. g. da-tivu-s, where tivw is for tiviu. The meaning is somewhat altered, and even in cap-Hvu-s, though the passive is expressed, it is' referred to the present, not the future, time, i. e. * taken,' not ' to be taken.' 269. ani-ya also forms Future Passive Parti- ciples : s. ydj-Sini'ya-s, 6 to be joined,' from yuj. z. -nya (the i in Sanskrit being perhaps a later development), e. g. 2/ai-nya, ' to be adored.' The Gothic has the same form -nja, e. g. ana-siu-njsi, ' visible ' (to be seen). 270. s. eya seems to be from e + ya, of which the first part probably is only introduced for euphonic reasons. It generally retains the accent on the one or the other syllable. It is used similarly with the simple form ya, e. g. s. das-ey&-s } ' a slave's son,' from dasa, ' 6 slave;' gdir-eysi-m, ' mountain produce,' from giri, ( mountain.' gr. -s*o, and abbreviated to -so : \sqvt-6L0-s x 306 DERIVATION AND COMPOSITION and Xeovr-eo-s, ' of a lion.' 1. eju, and abbreviated to eu : Pomp-eju-s ; ciner-eu-s, ' ashy.' 271. The stems of some Nouns, though not many, consist of the root only. In Sanskrit there are feminine Abstracts of this sort, e. g. s. bhi, 'fear;' mud, 'joy.' In Greek there are Appellatives so formed, e. g. ott (cty), 'eye;' <£Xo7 (<£\6£), 'flame;' but 0-7x7 (°"™£)> 'hatred/ and aiK (al%), 'impetuous motion,' 'spring' (tide), are Abstracts. Latin Appellatives, e. g. due (dux), 'leader.' Bare roots are also used at the end of Compounds, and generally in the sense of the present participle governing the preceding noun, e. g. s. dharma-vid, 'knowing duty;' du:kha-hdn, 'destroying pain;' gr. tysvart-o-Tvy ($r£vcricrTV%), ' hating lies ;' /copv$-aifc (/copvS- d'Cf;), ' shaking the helm ; ' 1. ju-dic (judex), ' uttering law ; ' au-eup (auceps), ' catching birds.' The vowel a is sometimes lengthened, e. g. s. vach, ' speech ' (r. vach) ; pari-vra'j, ' wandering about,' 'beggar' (r. vraj) ; gr. wir (coyfr), 'eye' (r. 07r); 1. reg (rex), ' king ' (r. reg). After a short root vowel t is added, as in some other cases, e. g. s. pari-zru-X, ' flowing round ; ' 1. com-i-X (comes), 'attendant' (r. i). 272. The suffix -a, which is the same, in form at least, as the demonstrative pronoun, is employed both as a primary and as a secondary suffix to form Mascu- line Abstracts. In Gothic these abstracts have acquired the neuter gender, as is seen from their not having s in the Nom. Sing., e. g. anda-beit (stem anda-beita), ' blame ;' af-let, ' forgiveness ' (letting off). One neuter DERIVATIVES 307 occurs in Sanskrit, i. e. bhay-a-m, ' fear/ from bhi, but M. jay-a-s, ' victory/ from ji. The same suffix, with the accent, -a, also forms Adjec- tives resembling the present participle in meaning, as well as Appellatives which generally were at first Nomina agentis, e.g. tras-a-s, ( trembling;' mush-a-s, s mouse' (lit. ( stealer'). In Greek -o, e. g. $>av-6-s, 'shining;' t/oo^-o-9, ' runner.' The meaning is sometimes passive, and the accent sometimes on the root both in Sanskrit and Greek. In the latter language 8 is sometimes added, as we have seen t to be in many cases after a short vowel, e. g. hopic-a-s (stem Sop/c-a-8-), ' gazelle' (' gazer ') ; rvir-d-s (st. TV7r-a-S-), c hammer ' (' striker '). These forms occur especially at the end of compounds, e. g. s. arin-dam-a-s (< taming '), ' tamer of foes ; ' gr. £7T7ro-Sa/x.-o-s k , 6 tamer of horses ; ' 1. nau-frag-u-s, c shipwreck.' The e. wreck as well as break is of the same root as the Latin frag in frang-ere, ' break.' Some words of this kind in Latin have the feminine form -a = s. & applied to both masculine and feminine genders, as in parri-cid-a, ' parricide,' from ccecZ-ere, and sometimes restricted to the Masc. as in coeli-col-a, * dwelling in heaven,' from cohere. Even scrib-a, 1 writer,' tf secretary,' though not a compound, has the feminine form for the masculine. Some other Mascu- line Appellatives ending in -a are really Greek words which- have dropped the final s of the Nom. Sing., like poe-ta, gr. iroirj-rri-s. On the other hand, these compounds in Greek have the masculine form for both masculine and feminine. The Gothic has a few instances of all these formations. Masculine, both compound and simple, e. g. daura- vard-a, ' door-keeper' (e.ivard); thiv-a(J$om. thiu-s\ * lad,'meaning the ( strong,' ' muscular,' from thu, e. thew, X 2 308 DERIVATION AND COMPOSITION s. tu, 'grow,' ' become strong;' but thivais in English degraded to * thief.' Neut. ga-thrask-a, ' threshing- floor.' Fern, daura-vard-6 (Nom. -vard-a), ' porteress.' Adjectives, laus-a, ' loose;' af -let-Si, Met off.' A passive meaning belongs to these forms when com- pounded with the prefixes su, 6 easy/ and dus, ' hard,' in Sanskrit, and with the corresponding ones si, Sv9 in Greek, e. g. s. su-kdr-SL-s, ' easy to be done ; ' dush- kdr-Bi-s, ' hard to be done.' This explanation of these forms, which I have given in deference to Bopp's autho- rity, seems to me somewhat arbitrary and unnecessary. The meaning is perhaps no more passive than in such Germanic forms as ger. leicht zu thun, e. easy to do, which some grammarians also represent as active forms used in a passive sense, whereas the true explanation is, by an ellipsis, easy (for any one) to do ; so also hard (for any one) to do. The above Sanskrit and Greek forms may also be taken in an active sense. As these derivatives have originally the sense of the present participle active, e. g. s. bhay-a-m, ' fear ' (lit. ' fearing'), so, when compounded with su or dus, they retain a similar meaning, e. g. dush-kdr-Si-s = ' hard doing,' not * hard being done;' gr. sv-cfrop-o-s = c easy bearing,' not easy being borne.' The ease or difficulty in each case refers to the agent, not to the thing done or borne. As a secondary suffix, a generally retains the accent, and is preceded by ' vriddhi.' It has a feminine in -%, and forms masculine substantives denoting descent, as well as neuters denoting fruit, etc., e. g. manav-k-s, * man ' (descendant of Manu) ; azwatth-k-m, ' fruit of the azwattha tree ; ' sdmudr-Si-m, € sea salt ' (' sea pro- duce '), from samudrd. In Greek the feminine patro- nymics in -i have the usual 8 affixed, e. g. 'Iva^-/-p, Gen. 'lvax-lS-09, * daughter of Inachus ; ' firjX-o-v, ' apple,' DERIVATIVES 309 from fjL7)\t-$-; ft)(f)-o-z/, 'egg.' In Latin pom-VL-m, ' apple/ from pomu-s; ov-u-m, 'egg ' (' bird's produce'), from avi-s. Neuter Abstracts are also thus formed, s. yduvan-k-m, * youth,' from yuvan ; and Neuter Collectives, s. kdpot- SL-wi, ' a flock of pigeons,' from kapo'ta. Adjectives and Appellatives occur, e. g. s. dyas-a, M. N., ayas-i F., ' of iron,' from ayas ; 1. decor-u-s, 6 proper,' from decus. The feminine d r with the accent is also used to form Abstracts : s. bhid-k', ' a splitting ; ' gr. -77, ' g race ; ' L perhaps such as cced-es (csed-i-), c cutting ; ' go. vunn-i, ' suf- fering' (wound). It is also used, with the accent placed variously, in forming Masculine Nomina agentis and Appella- tives, e. g. chhid-i-s, 6 splitter ; ' ah-is, f serpent ' (mover, creeper); z. az-i-s, 'serpent;' gr. rpox-t-s, 6 runner;' sx~ L ~ 9 > 'serpent;' sometimes S again is added: kott-i-s (fco7r-i,8-), 'knife;' 1. angu-i-s, 'serpent:' go. junga-laud-i, ' young man ' (e. lad). 310 DERIVATION AND COMPOSITION u. 274. The suffix u, without the accent, is employed to form Adjectives resembling in meaning the present participle of desiderative verbs, and governing the accusative case. With the accent it forms Adjectives without the desiderative meaning, e.g. s. didrksh-u: pitarau, ' desirous of seeing parents ; ' tan-vi, ' thin ' (outstretched) ; swad-XL, gr. rj8-v, L sua-Y-is, Q sweet ; ' go. thaurs-u-s, f dry.' In 1. i is added to the suffix, and suavis is for suad-u-i-s. Appellatives are also formed with an accented or unaccented u: bhid-ii, 6 thunderbolt ' (splitter); gr. vs/c-v, € corpse ' (perishing) ; 1. curr-xi-s, c carriage ' (runner) ; go. /d£-u, 6 foot ' (goer). AN. 275. Appellatives are formed by an (an) without accent, e. g. s. sne'h-BJL, ' friend ' (lover) ; ro'j-an, 4 king ' (ruler). In Greek this affix assumes several forms, az>, s *>, oi>, rjv, cov, e. g. rak-ap, 4 patient f app-ev> 4 male ; ' av-ep6-$, ' visible;' (f>\sy-vpo-9, 6 burn- ing;' Kafjbir-vXo'9, 'bent;' 1. ten-er (st. ten-ero), 6 tender;' ag-ili-s, 'active;' teg-ulu-m, 'roof (cover- ing)- Some secondary derivatives are in these forms, e. g. s. azm-ara, ' stony ; ' zri-la, 6 fortunate ; ' medh-ira and medh-ila, ' intelligent ; ' gr. $3ovs-p6-9, 6 envious ;' X a f JLr r^o-9 y ' on the ground ; ' 1. carna-li-s, ' fleshly.' WA (VA). 279. The suffix -wa {ya), Fem. wd (yd), generally without the accent, forms Appellatives, e.g. s. dz- wa-s, * horse ' (runner) ; z. az-pa, where w has become p ; gr. 'itt-tto-s = Ik-ko-s for Ik-Fo-s ; 1. eq-UU-s ; a. s, 314 DERIVATION AND COMPOSITION eoh, in which the formative is again dropped. Adjec- tives are also formed in the same way, e. g. s. rish-wa, c offending ; ' gr. perhaps such forms as Zpofi-ev-s, 6 runner ; ' 1. tor-vu-s, '■ stern ' (piercing) ; go. las-ivs (st. las-iva-), ' weak ; ' e. laz-y, the formative being represented only by y. WAN (VAN). 280. The suffix wan or wan (van or van), without the accent, forms : 1. Adjectives with a participial mean- ing, e.g. s. vaja-dAl -v&n, s giving food.' 2. Nomina agentis, e. g. s. ydj-waii, ' sacrificer.' 3. Appellatives, e. g. s. ruh-waxi, '■ tree ' (grower) ; z. zar-w&n, ' time ' (destroyer). This suffix appears also with an additional t in vant, mant (vat, mat in weak cases). In Latin there is a change of v or m to I, and a further addition of o in -lento. In Grreek the corresponding form would be Fsvt or f st, of which, however, the digamma is generally lost, and svt, st remain. The digamma is preserved in some instances in an inscription found in the island of Corfu in 1845, and published in 1846 by Professor Franz. (See Zeitschrift fur Vergleichende Sprachforschung, i. 118, 1 19.) Among the instances where the digamma is preserved is one word containing the formative in question, i. e. aTovofsacrav for GTovo-fevT-iav* Other instances of this formative are: s. dziva- vant, 'hav- ing horses;' marut-vant, ' having Maruts' (Indra) ; gr. vkrj-evr- (v\r)sis), ' having wood,' ' woody ; ' irvpo- evr- (irvposci), ' having fire ; ' 1. puvu-lent-us, ' having matter ' (pus) ; opu-lent- (opulens), ' having wealth.' DERIVATIVES 315 NIT. 281. Adjectives and Substantives are formed by -nu with the accent, e. g. s. tras-nH-s, ' trembling ; ' 6k-nu-s, ' sun ' (shiner); z. te/-nu-s, ' burning;' jahf-mi-s, ' mouth ' (speaker) ; gr. \1y-vv-9, ' smoke ' (s. dah, ' burn ') ; 1. lig-mi-m (fire) 6 wood.' MI. 282. This is perhaps only a weakened form of ma, Fern. md. It rarely occurs, but is found in a few Appel- latives with the accent, e. g. s. 6M-mi-s, ' ground ; ' 1. te-mu-s, go. hai-m-s (st. hai-mi-) ? ' village,' e. ho-me. The two last denote resting or sleeping place, from the root s. zi, gr. /est. KA. The suffix ha with the accent is rarely used in imme- diate connection with the root, e. g. s. zush-ka-s (for sus-ka-s), ' dry ; ' z. hush-lsa ; 1. sic-cu-s (for sus-cu-s). A vowel is usually interposed between the root and ka, which then has not the accent, e. g. s. ndr£-aka-s, 6 dancer ; ' jdlp-aka,, ' talkative ;' mu'sA-ika-s, f mouse ' (stealer); M'm-uka, 'wanton;' vdvad-ufka, < talka- tive ; ' gr. $v\- a KO-9, ' guard ;' $sv-aK-9 {$iv-a%), ' de- ceiver ; ' fcrjp-vK- (#>7P-&£)> ' herald ; ' yvv-aiK- for 71/1/- aK-t- (7W-4 ' woman'); 1. mecZ-icu-s, ' physician;' am-icu-s, ' friend;' ed-kc- (edax, 6 devouring ') ; vel- OC- (velox, ' swift,' (? = original a) ; cacZ-ucu-s, c fall- ing.' In Greek 0, and in Latin u (for 0), corresponding to s. a in &a, are often dropped ; hence the Nom. Sing, ends in £ for k$, and x for es. 316 DERIVATION AND COMPOSITION The 'English -ing, which is employed in forming Ab- stract Nouns, and has gradually taken the place of the active participle instead of -nd for ndh = 1. gr. s. nt, is probably formed from this suffix by inserting the nasal ; e. g. king, for kin-ing, ' powerful/ is an adjective, unless it be an appellative from the root jan, meaning ' producer.' The same form in heal-ing is used both as an Abstract and as a Participle. The suffix ka also forms secondary derivatives, with i or u inserted after consonant stems, e. g. ma6po-9, 6 seed-bearer,' is perhaps for a7rspfJi(aT)-o-(j)opo-9. 289. Both Zend and Greek furnish instances of the Nominative case being employed in the first part of a compound, e. g. z. daivd-c?d£a, < made of God,' 6 being = s. as ; gr. Steoy-SoTo-s, ' given of God.' The Genitive occurs in the first part of compounds in both Greek and Gothic, e. g. gr. veoas-oucoi, like 322 DERIVATION AND COMPOSITION ger. scMjf-s-hauser, ' ship-sheds ;' ovdevos-copa, ' fit for nothiog;' go. "hSLJirg-s-vaddjus, 'city wall.' 290. A classification of compounds is made by Sanskrit grammarians which it may be useful here to introduce. They are arranged in six classes. 1. Copulative Compounds. 291. Two or more Substantives are united together, with a common termination. Their union serves the same purpose in regard to meaning as connecting them together by a conjunction. Some of them have a plural (or dual) termination, and others a neuter singular, e. g. s. surya-chandramdssLU, ' sun-(and)-moon ; ' pitard- matdriiXL, ' father - (and) - mother ; ' agni - vayu - ravi- fohyas, ' fire-air-(and)-sun.' There is no limit to the number of words which may be thus combined together. Those compounds which have a neuter singular ending consist of words denoting inferior objects, e. g. hasta- padam, ' hands-(and)-feet ; ' gr. vvy§r)n,zpov, 'night- (and)-day;' ^arpa^o-fivo in f3aTpaxo/j,vo-/j,axia, 'the war of frog-(and)-mouse.' 1. suovitaurilia has a neuter plural ending, and consists of three substantives thus combined together, su-ovi-tauri-lia, which is also ab- breviated to solitaurilia, € the solemnities during which were sacrificed a pig-sheep-(and)-bull. 5 Adjectives are also thus combined, though less fre- quently, e. g. vrtta-pina^ ' round-(and)-thick ; ' gr. \svfco-fis\as, ' white-(and)-black.' 2. Possessive Compounds. 292. Possessive Compounds express the possession of what is denoted by the several parts of the compound. COMPOUNDS 323 They are sometimes appellatives, but most generally adjectives. The first member may belong to any part of speech except verb, conjunction, or interjection. The last must be a substantive, which undergoes no change except in the expression for gender, to adapt it as an adjective for all three. The accent is in Sanskrit upon the first member, as it would be in a separate state. In Greek it follows the general rule, being placed on the third quantity (short syllable) from the end. The first part is most frequently an adjective or parti- ciple, e. g. s. chafru-ldchana-s, * with beautiful eyes ; ' gr. \evKO-7TTspo-s, ' with white wings ; ' 1. atri-color, e of a black colour ; ' go. hrainja-hairts, ' with a pure heart;' e. pure-heart-ed. The form of the passive participle is used in English, as if derived from a verb. The first part is a substantive in s. bh'la-putra-s, 6 with a child as son ; ' gr. /3ov-/cs A^ , : : ^ mil m