The history of the infinitive makes it clear how little distinction must have been felt at the outset between the noun and the verb. Indeed, the growth of the verb was a slow process. There was a time in the history of Indo-European speech when it had not as yet risen to the consciousness of the speaker, and in the period when the noun did not possess a plural there was as yet also no verb. The attachment of the first and second personal pronouns, or of suffixes resembling them, to certain stems, was the first stage in the development of the latter. Like the Semitic verb, the Indo-European verb seems primarily to have denoted relation only, and to have been attached as an attribute to the subject. The idea of time, however, was soon put into it, and two tenses were created, the one expressing a present or continuous action, the other an aoristic or momentary one. The distinction of sense was symbolized by a distinction of pronunciation, the root-syllable of the aorist being an abbreviated form of that of the present. This abbreviation was due to a change in the position of the accent (which was shifted from the stem-syllable to the termination), and this change again was probably occasioned by the prefixing of the so-called augment to the aorist, which survived into historical times only in Sanskrit, Zend and Greek, and the origin of which is still a mystery. The weight of the first syllable in the aorist further caused the person-endings to be shortened, and so two sets of person-endings, usually termed primary and secondary, sprang into existence. By reduplicating the root-syllable of the present tense a perfect was formed; but originally no distinction was made between present and perfect, and Greek verbs like δίδωμι and ἣκω are memorials of a time when the difference between “I am come” and “I have come” was not yet felt. Reduplication was further adapted to the expression of intensity and desire (in the so-called intensive and desiderative forms). By the side of the aorist stood the imperfect, which differed from the aorist, so far as outward form was concerned, only in possessing the longer and more original stem of the present. Indeed, as Benfey first saw, the aorist itself was primitively an imperfect, and the distinction between aorist and imperfect is not older than the period when the stem-syllables of certain imperfects were shortened through the influence of the accent, and this differentiation of forms appropriated to denote a difference between the sense of the aorist and the imperfect which was beginning to be felt. After the analogy of the imperfect, a pluperfect was created out of the perfect by prefixing the augment (of which the Greek ἐμέμηκον is an illustration); though the pluperfect, too, was originally an imperfect formed from the reduplicated present.

Besides time, mood was also expressed by the primitive Indo-European verb, recourse being had to symbolization for the purpose. The imperative was represented by the bare stem, like the vocative, the accent being drawn back to the first syllable, though other modes of denoting it soon came into vogue. Possibility was symbolized by the attachment of the suffix -ya to the stem, probability by the attachment of -a and -ā, and in this way the optative and conjunctive moods first arose. The creation of a future by the help of the suffix -sya seems to belong to the same period in the history of the verb. This suffix is probably identical with that used to form a large class of adjectives and genitives (like the Greek ἵπποιο for ἱπποσιο); in this case future time will have been regarded as an attribute of the subject, no distinction being drawn, for instance, between “rising sun” and “the sun will rise.” It is possible, however, that the auxiliary verb as, “to be,” enters into the composition of the future; if so, the future will be the product of the second stage in the development of the Indo-European verb when new forms were created by means of composition. The sigmatic or first aorist is in favour of this view, as it certainly belongs to the age of Indo-European unity, and may be a compound of the verbal stem with the auxiliary as.

After the separation of the Indo-European languages, composition was largely employed in the formation of new tenses. Thus in Latin we have perfects like scrip-si and ama-vi, formed by the help of the auxiliaries as (sum) and fuo, while such forms as amaveram (amavi-eram) or amarem (ama-sem) bear their origin on their face. So, too, the future in Latin and Old Celtic (amabo, Irish carub) is based upon the substantive verb fuo, “to be,” and the English preterite in -ed goes back to a suffixed did, the reduplicated perfect of do. New tenses and moods, however, were created by the aid of suffixes as well as by the aid of composition, or rather were formed from nouns whose stems terminated in the suffixes in question. Thus in Greek we have aorists and perfects in -κα, and the characteristics of the two passive aorists, ye and the, are more probably the suffixes of nominal stems than the roots of the two verbs ya, “to go,” and dhâ, “to place,” as Bopp supposed. How late some of these new formations were may be seen in Greek, where the Homeric poems are still ignorant of the weak future passive, the optative future, and the aspirated perfect, and where the strong future passive occurs but once and the desiderative but twice. On the other hand, many of the older tenses were disused and lost. In classical Sanskrit, for instance, of the modal aorist forms the precative and benedictive almost alone remain, while the pluperfect, of which Delbrück has found traces in the Veda, has wholly disappeared.

The passive voice did not exist in the parent Indo-European speech. No need for it had arisen, since such a sentence as “I am pleased” could be as well represented by “This pleases me,” or “I please myself.” It was long before the speaker was able to imagine an action without an object, and when he did so, it was a neuter or substantival rather than a passive verb that he formed. The passive, in fact, grew out of the middle or reflexive, and, except in the two aorists, continued to be represented by the middle in Greek. So, too, in Latin the second person plural is really the middle participle with estis understood, and the whole class of deponent or reflexive verbs proves that the characteristic r which Latin shares with Celtic could have had at the outset no passive force.

Much light has been thrown on the character and construction of the primitive Indo-European sentence by comparative syntax. In contradistinction to Semitic, where the defining word follows that which is defined, the Indo-European languages place that which is defined after that which defines it; and Bergaigne has made it clear that the original order of the sentence was (1) object, (2) verb, and (3) subject. Greater complication of thought and its expression, the connexion of sentences by the aid of conjunctions, and rhetorical inversion caused that dislocation of the original order of the sentence which reaches its culminating point in the involved periods of Latin literature. Our own language still remains true, however, to the syntax of the parent Indo-European when it sets both adjective and genitive before the nouns which they define. In course of time a distinction came to be made between an attribute used as a mere qualificative and an attribute used predicatively, and this distinction was expressed by placing the predicate in opposition to the subject and accordingly after it. The opposition was of itself sufficient to indicate the logical copula or substantive verb; indeed, the word which afterwards commonly stood for the latter at first signified “existence,” and it was only through the wear and tear of time that a phrase like Deus bonus est, “God exists as good,” came to mean simply “God is good.” It is needless to observe that neither of the two articles was known to the parent Indo-European; indeed, the definite article, which is merely a decayed demonstrative pronoun, has not yet been developed in several of the languages of the Indo-European family.

We must now glance briefly at the results of a scientific investigation of English grammar and the modifications they necessitate in our conception of it. The idea that the free use of speech is tied down by the rules of Investigation of English grammar. the grammarian must first be given up; all that the grammarian can do is to formulate the current uses of his time, which are determined by habit and custom, and are accordingly in a perpetual state of flux. We must next get rid of the notion that English grammar should be modelled after that of ancient Rome; until we do so we shall never understand even the elementary principles upon which it is based. We cannot speak of declensions, since English has no genders except in the pronouns of the third person, and no cases except the genitive and a few faint traces of an old dative. Its verbal conjugation is essentially different from that of an inflexional language like Latin, and cannot be compressed into the same categories. In English the syntax has been enlarged at the expense of the accidence; position has taken the place of forms. To speak of an adjective “agreeing” with its substantive is as misleading as to speak of a verb “governing” a case. In fact, the distinction between noun and adjective is inapplicable to English grammar, and should be replaced by a distinction between objective and attributive words. In a phrase like “this is a cannon,” cannon is objective; in a phrase like “a cannon-ball,” it is attributive; and to call it a substantive in the one case and an adjective in the other is only to introduce confusion. With the exception of the nominative, the various forms of the noun are all attributive; there is no difference, for example, between “doing a thing” and “doing badly.” Apart from the personal pronouns, the accusative of the classical languages can be represented only by position; but if we were to say that a noun which follows a verb is in the accusative case we should have to define “king” as an accusative in such sentences as “he became king” or “he is king.” In conversational English “it is me” is as correct as “c’est moi” in French, or “det er mig” in Danish; the literary “it is I” is due to the influence of classical grammar. The combination of noun or pronoun and preposition results in a compound attribute. As for the verb, Sweet has well said that “the really characteristic feature of the English finite verb is its inability to stand alone without a pronominal prefix.” Thus “dream” by itself is a noun; “I dream” is a verb. The place of the pronominal prefix may be taken by a noun, though both poetry and vulgar English frequently insert the pronoun even when the noun precedes. The number of inflected verbal forms is but small, being confined to the third person singular and the special forms of the preterite and past participle, though the latter may with more justice be regarded as belonging to the province of the lexicographer rather than to that of the grammarian. The inflected subjunctive (be, were, save in “God save the King,” &c.) is rapidly disappearing. New inflected forms, however, are coming into existence; at all events, we have as good a right to consider wont, shant, cant new inflected forms as the French aimerai (amare habeo), aimerais (amare habebam). If the ordinary grammars are correct in treating forms like “I am loving,” “I was loving,” “I did love,” as separate tenses, they are strangely inconsistent in omitting to notice the equally important emphatic form “I do love” or the negative form “I do not love” (“I don’t love”), as well as the semi-inflexional “I’ll love,” “he’s loving.” It is true that these latter contracted forms are heard only in conversation and not seen in books; but the grammar of a language, it must be remembered, is made by those who speak it and not by the printers.

Our school grammars are the inheritance we have received from Greece and Rome. The necessities of rhetoric obliged the Sophists to investigate the structure of the Greek language, and to them was accordingly due the first History of formal grammar. analysis of Greek grammar. Protagoras distinguished the three genders and the verbal moods, while Prodicus busied himself with the definition of synonyms. Aristotle, taking the side of Democritus, who had held that the meaning of words is put into them by the speaker, and that there is no necessary connexion between sound and sense, laid down that words “symbolize” objects according to the will of those who use them, and added to the ὄνομα or “noun,” and the ῥῆμα or “verb,” the σύνδεσμος or “particle.” He also introduced the term πτῶσις, “case,” to denote any flexion whatsoever. He further divided nouns into simple and compound, invented for the neuter another name than that given by Protagoras, and starting from the termination of the nominative singular, endeavoured to ascertain the rules for indicating a difference of gender. Aristotle was followed by the Stoics, who separated the ἄρθρον or “article” from the particles, determined a fifth part of speech, πανδέκτης or “adverb,” confined the term “case” to the flexions of the nouns, distinguishing the four principal cases by names, and divided the verb into its tenses, moods and classes. Meanwhile the Alexandrian critics were studying the language of Homer and the Attic writers, and comparing it with the language of their own day, the result being a minute examination of the facts and rules of grammar. Two schools of grammarians sprang up—the Analogists, headed by Aristarchus, who held that a strict law of analogy existed between idea and word, and refused to admit exceptions to the grammatical rules they laid down, and the Anomalists, who denied general rules of any kind, except in so far as they were consecrated by custom. Foremost among the Anomalists was Crates of Mallos, the leader of the Pergamenian school, to whom we owe the first formal Greek grammar and collection of the grammatical facts obtained by the labours of the Alexandrian critics, as well as an attempt to reform Greek orthography. The immediate cause of this grammar seems to have been a comparison of Latin with Greek, Crates having lectured on the subject while ambassador of Attalus at Rome in 159 B.C. The zeal with which the Romans threw themselves into the study of Greek resulted in the school grammar of Dionysius Thrax, a pupil of Aristarchus, which he published at Rome in the time of Pompey and which is still in existence. Latin grammars were soon modelled upon it, and the attempt to translate the technical terms of the Greek grammarians into Latin was productive of numerous blunders which have been perpetuated to our own day. Thus tenues is a mistranslation of the ψιλά, “unaspirated”; genetivus of γενική, the case “of the genus”; accusativus of αἰτιατική, the case “of the object”; infinitivus of ἀπαρέμφατος, “without a secondary meaning” of tense or person. New names were coined to denote forms possessed by Latin and not by Greek; ablative, for instance, was invented by Julius Caesar, who also wrote a treatise De analogia. By the 2nd century of the Christian era the dispute between the Anomalists and the Analogists was finally settled, analogy being recognized as the principle that underlies language, though every rule admits of exceptions. Two eminent grammarians of Alexandria, Apollonius Dyscolus and his son Herodian, summed up the labours and controversies of their predecessors, and upon their works were based the Latin grammar composed by Aelius Donatus in the 4th century, and the eighteen books on grammar compiled by Priscian in the age of Justinian. The grammar of Donatus dominated the schools of the middle ages, and, along with the productions of Priscian, formed the type and source of the Latin and Greek school-grammars of modern Europe.

A few words remain to be said, in conclusion, on the bearing of a scientific study of grammar upon the practical task of teaching and learning foreign languages. The grammar of a language is not to be confined within the rules Learning of grammar of foreign languages. laid down by grammarians, much less is it the creation of grammarians, and consequently the usual mode of making the pupil learn by heart certain fixed rules and paradigms not only gives a false idea of what grammar really is, but also throws obstacles in the way of acquiring it. The unit of speech is the sentence; and it is with the sentence therefore, and not with lists of words and forms, that the pupil should begin. When once a sufficient number of sentences has been, so to speak, assimilated, it will be easy to analyse them into their component parts, to show the relations that these bear to one another, and to indicate the nature and varieties of the latter. In this way the learner will be prevented from regarding grammar as a piece of dead mechanism or a Chinese puzzle, of which the parts must be fitted together in accordance with certain artificial rules, and will realize that it is a living organism which has a history and a reason of its own. The method of nature and science alike is analytic; and if we would learn a foreign language properly we must learn it as we did our mother-tongue, by first mastering the expression of a complete thought and then breaking up this expression into its several elements.

(A. H. S.)

See [Philology], and articles on the various languages. Also Steinthal, Charakteristik der hauptsächlichsten Typen des Sprachbaues (Berlin, 1860); Schleicher, Compendium of the Comparative Grammar of the Indo-European Languages, translated by H. Bendall (London, 1874); Pezzi, Aryan Philology according to the most recent Researches, translated by E. S. Roberts (London, 1879); Sayce, Introduction to the Science of Language (London, 1879); Lersch, Die Sprachphilosophie der Alten (Bonn, 1838-1841); Steinthal, Geschichte der Sprachwissenschaft bei den Griechen und Römern mit besonderer Rücksicht auf die Logik (Berlin, 1863, 2nd ed. 1890); Delbrück, Ablativ localis instrumentalis im Altindischen, Lateinischen, Griechischen, und Deutschen (Berlin, 1864); Jolly, Ein Kapitel vergleichender Syntax (Munich, 1873); Hübschmann, Zur Casuslehre (Munich, 1875); Holzweissig, Wahrheit und Irrthum der localistischen Casustheorie (Leipzig, 1877); Draeger, Historische Syntax der lateinischen Sprache (Leipzig, 1874-1876); Sweet, Words, Logic, and Grammar (London, 1876); P. Giles, Manual of Comp. Philology (1901); C. Abel, Ägypt.-indo-eur. Sprachverwandschaft (1903); Brugmann and Delbrück, Grundriss d. vergl. Gram. d. indogerm. Spr. (1886-1900); Fritz Mauthner, Beiträge zu einer Kritik der Sprache vol. iii. (1902); T. G. Tucker, Introd. to a Nat. Hist. of Language (1908).