=

The change of s into r, as in labos = labor, honos honor, is almost too well known to require mention.**

So much for the final letter of amatur. The tu preceding it is nothing more nor less than the third personal pronoun ta, of which relics are seen in Germ. lieb-t, and in the th or s in Eng. loveth, loves. So that we have the formula ama + tu + r = liebt sich love + s + self. And similarly it might be proved that the Gr. φιλεῖται was formerly φιλείτασι, and that the Skr. kamayate was once kamayatasi.

It is thus proved that there are at least three distinct words integrated in ama-tu-r. Further inquiry will show that there is a fourth, namely, the final vowel a of the stem am-a, which the old grammarians thought they had sufficiently explained when they had christened it "the connecting vowel." That it was originally something more than a mere connecting vowel will appear from the Sanskrit form in the following table : —

This word aya appears in Zend as aye. In Lithuanian it is best preserved in the third conjugation as eyo; while in the second conjugation it has become a. In Latin it is variously contracted, in verbs of the first, second, and fourth conjugations, into a, e, and i. In Prakrit it has become e. The universality of its presence shows that it fulfils an important function; but its primitive signification cannot be detected at a glance. Mr. Clark (Comp. Gram., p. 232) discusses and rejects the only hypothesis which seems to occur to him, namely, that it is a tense-sign. The mere fact that it occurs in all the tenses alike is fatal to such a supposition. We shall find that the true explanation is more difficult and will take us far down into the primitive formation of language.

Sir Graves Haughton (Bengali Gram., pp. 68, 95) has satisfactorily explained the similarly inserted ya of Sanskrit passives as being simply the auxiliary verb ya, "to go." Both in Bengali and in Hindustani the passive is regularly formed by such

This interpretation of the r in amatur rests chiefly upon the authority of Bopp (Vergl. Gramm. II. 688), and Pott (Etym. Forsch. I. 133). It is vigorously but ineffectively combated by Donaldson, who of course does not let slip the opportunity of reviling Professor Key for giving his adherence to it. (New Cratylus, p. 587, fol'g)

66

an auxiliary, e. g. kara yai, “I go to making” = “I am made," with which we may compare the Lat. "amatum iri." Now many facts combine to show that aya is also an auxiliary, being a very old root with the signification "do," "make," or " cause." In Sanskrit, Latin, Greek, and Gothic, it is used in the formation of causatives and denominatives: thus, in Skr. kar-aya-mi, "I cause to make"; in Lat. neca-re, "cause to die," mon-e-re, 66 cause to think," sop-i-re, cause to sleep"; in Gr. xaλ-é-w, "cause to hear" (xλú-w) ; in Goth. drank-ja, "cause to drink," i. e. "drench." Even "in the English word drench, although no part of the original aya is preserved, yet the influence of y is apparent in the change of k to ch" (Clark, Comp. Gram., 271). In Welsh and Irish, aya, represented by ia and aigh, forms causatives and denominatives;* as in Irish beos-aigh-im, identical with Skr. bhush-aya-mi, "I adorn"; and it appears in the Irish adjectival termination ach, as in beos-ach, "beautiful." This last circumstance leads us to regard the ya employed in the formation of denominative abstracts (oop-ía, præsent-ia), and adjectives (div-ya-s, "heavenly," from div, ovpáv-10-s from oupavós, patr-iu-s from pater, Mart-iu-s from Mars), as also a form of aya. Finally, this ubiquitous word appears in Sanskrit as a dative-ending, and is used in some feminine nouns to form the instrumental case, as in jihvaya," with the tongue," from jihva. †

The inference from all this is that aya had a generally causative signification, and was therefore early adopted in the mother-Aryan as a means by which to distinguish verbs from nouns. Its use in the formation of denominatives alone tells the story; for a denominative is simply a verb formed from a noun, and we have now to observe that all verbs are formed from nouns or other parts of speech, sometimes merely by affixing pronouns, sometimes by the additional insertion of a verbmaking word like this aya. The reader must rid himself of the notion, derived from a familiarity with highly developed languages, that between the verb and other parts of speech

*Pictet, Langues Celtiques, pp. 148, 149.

† It appears also in the Greek infinitive-ending μeval, for Skr. manaya, dat. sing. of mana; and in the Old Saxon and Anglo-Saxon infinitive-ending e for ai See Clark, Comp. Gram., 299, and cf. Donaldson, New Crat., p. 664.

there is fixed a deep and impassable gulf. Most languages recognize no such wide and permanent distinction. In Chinese a noun becomes a verb merely by change of tone. In Mandshu the same roots may generally be used as nouns, verbs, adverbs, or even as particles. In Hungarian, lak means "dwellingplace"; followed by a pronoun, it becomes a verb, lak-ih, "he dwells." In the same language lep-em may mean either "my covering" or "I cover." In the whole Tungusic class there is no formal distinction between "my pay" and "I pay." In Turkish, güzel-im may mean "my handsome one," or "I am handsome," according as it is accented.* "There is no way

of distinguishing between a Coptic finite verb and the corresponding noun with pronominal affixes, except that the latter usually has the article, which is wanting in the former." (Garnett, p. 293.) In Tamil, verbal forms may be declined and nouns may be conjugated. In Malagassy they conjugate even adverbs and prepositions. In Basque and several American languages, even conjunctions "may be conjugated through a long array of moods and tenses."

Nevertheless, it is convenient to have a distinction in form. between nouns and verbs; and accordingly, in the motherAryan, we find that three expedients were adopted to secure this desideratum. In the first place, the verb was distinguished from the noun by a different set of pronominal endings. Thus, in Sanskrit, vak, when followed by case-endings, as vak-as, vak-i, etc., means "speech"; but when followed by a different set of affixes, as vak-mi, vak-shi, vak-ti, it means "I speak," etc. In the second place, a noun became a verb by being reduplicated,† as dha-dha-ma, "I do," Skr. dadhami, Gr. Tionμ. The third expedient, which was resorted to at a later period, was the insertion of certain auxiliary forms, called by the Indian grammarians "vikaranas," of which our aya is a specimen. Thus the Latin, from the root voc, makes voc-s, "voice," and

Compare the English perfect and perfect, súrvey and survéy, ábsent and absent, and the still slighter distinction between ádvocăte and ádvocāte.

† After the introduction of new bases, formed by a "vikarana," the old reduplicated forms were usually made to serve for the perfect tense. Curiously enough, in Thibetan we find verbs made by reduplication of the final letter, as nag, “black,” nag-go, "it is black." Max Müller suggests that the variable consonant in Semitic triliteral roots may have arisen from a similar device. Bunsen, Outlines, I. 306 08.

it might have made the corresponding verb by simply affixing the requisite pronoun, in which case voc-t would have been quite identical with Skr. vak-ti. Instead of so doing, however, it first added the auxiliary aya, making voc-a-t," he speaks."

That such was the proper function of aya may be clearly seen from its use in imperatives. As Garnett well observes, a noun may be an imperative, as when the chairman of a turbulent meeting cries out, Silence! Even particles, as Germ. fort! Ital. via! may be similarly used. When the old Aryan languages used their nouns in this sense, they generally affixed to them the "vikarana," or verb-maker. The second person-ending of the Sanskrit imperatives in verbs of the tenth class is simply aya, which still exists in the Gr. Tiμ-a, píx-e, Lat. am-a, hab-e, Goth. tam-ei, hab-ai.

But we have seen this useful verb-maker serving also as a dative-sign for nouns. How are we to reconcile such widely divergent functions? The task will not seem difficult when we recollect that in English, where the loss of inflectional forms frequently makes us return to primitive methods, we are continually raising nouns into verbs by simply prefixing that preposition which we use also as the sign of the dative case. "Love" is a noun, but "to love" is a verb; "to me" is a dative, but "to dwell" is an infinitive. We are strongly inclined to refer this preposition to, Germ. zu, Lat. ad, Skr. adhi, to the venerable root dha, Skr. dha, Gr. On, Lat. da, Germ. thu, Eng. do.* At all events, the kinship between the idea of “doing" and that of "motion towards" is sufficiently obvious to any one familiar with the tricks of language, and is undoubtedly illustrated in the close similarity of the Aryan roots aya, "do," and ya, "go." The English to was originally used, in Gothic, to denote the goal at which an action aims, as du hausjan," in order to hear." †

From all these considerations, we regard it as sufficiently proved that aya is a most ancient word, signifying "do," "put," or "cause," analogous, in short, to dha. Early losing its independent power, it became a useful auxiliary with

* Compare Tooke, Diversions of Purley (ed. Taylor), p. 189, fol'g. And see, for the possibility of a connection between Oŋ and ya, Curtius, Beiträge, p. 329. f Note also the inceptive force of Greek verbs in ιάω, as ιλιγγιάω, "I grow dizzy,” keλawiáw, “I grow black," "xpiów, “I grow pale," etc. Farrar, Greek Syntax, p. 46.

which to form dative (and sometimes instrumental) cases, infinitives, and causative, as well as denominative, verbs. Thus is every syllable of our test-word am-a-tu-r adequately explained; and we find it to have consisted originally of four independent words, amaya+ta + se="love-does-he-himself." *

This example, as well as those previously cited, shows most strikingly the way in which the Aryan languages are built up; and any further inquiry can only serve to illustrate our fundamental proposition that the history of Aryan speech is in the main a history of phonetic integration. The less highly developed Semitic languages can afford us no such striking examples; yet from this more primitive domain we shall select one instance which will show us that it is governed by the same general laws. The earliest device for forming plurals appears to have been the addition of a word signifying "multitude," "crowd," "heap," or "all." In Tamil the plural suffix is gal, which Caldwell derives from dala, "a crowd": the Telugu suffix lu has the same meaning. The Hindustani, which has lost the Sanskrit sign for plurality, has recourse to lab, "people," or sab, "all." Bearing this in mind, let us take a Hebrew word, such as dhov, "good," and observe how it forms the plural. The masculine dhov, "bonus," takes the ending ghim, forming by integration the plural dhov-im. Now this word ghim denotes an aggregate, and is independently used as an adverb and preposition † equivalent in force to ἅμα, apa, óμoû, Lat. cum, simul, Skr. sam, Irish samhuil; though, whether, as Gesenius and Donaldson think, it is etymologically akin to these words, we shall not attempt to decide. So that dhov-im is nothing more nor less than "good-flock." On the other hand, the feminine dhovah, "bona," takes a different suffix, h'eth, with which, by integration, it forms the plural dhovah + h'eth dhovoth. This

=

In the Suabian, Bavarian, and Tyrolese dialects, the verb "do" is (as in English emphatic forms) used universally as an auxiliary with infinitives, to express the various sub-forms of the indicative mood, - only the auxiliary undergoing conjugation. See also the M. H. G. To avoid unnecessary prolixity, I have abstained from any reference to the oblique case in which the pronoun is joined to the verb. The views here set forth are not to be interpreted as conflicting with those expounded by Garnett in his famous paper on "The Nature and Analysis of the Verb," 1849.

Compare the kinship between Fr. troupeau and trop; Ital. truppa and troppo. Lewis, Rom. Lang. 223.

Gesenius, Heb. Gr. p. 4; Donaldson, Maskil le-Sopher, p. 15.