Abbildungen der Seite





F. Sommer, in the revised edition of his Handbuch der lateinischen Laut- und Formenlehre, offers no explanation of the development of final -ai in the dative singular of ā-stems (= first declension) to -ae rather than to i; see pp. 146, 327-28. Yet the development of the long diphthong di to i in non-initial (=unaccented) syllables is certain in the dative-ablative plural of a-stems, and an identical development of the long ōi is seen in the same forms of the o-stems, where length is assured by the Sanskrit form: inst. pl. devāiş –Latin dīvīs. The long diphthong in the dative singular of a-stems is proved by the Greek ending in xwpa 'terrae,' bea 'deae.'

It is impossible to separate the consideration of the dative -ae from that of the -ae in the genitive singular and in the nominative plural, where there is a similar development of a diphthong to the value which it had in initial (=accented) syllables. K. Brugmann, in his Grundriss d. vergl. Gramm. d. indog. Sprachen2 1.228 (after H. Osthoff, Zur Geschichte d. Perfekts im Indogermanischen, 197 ff.), explains this as starting from the monosyllabic pronouns quae hae haec, spreading thence to polysyllabic pronouns such as istae, and ultimately to all other words of the class. In this view there are difficulties. The quae hae haec are taken in their accented values, although all other forms of these pronouns (except the nom.-acc. neut. hoc) develop in their unaccented values, as F. Solmsen, Indogerm. Forschungen, 4. 142 f., has observed. Further, there are no monosyllabic genitives or datives ending in -ae, so that the extension could have been justified only in the nominative plural. Secondarily, perhaps, the identity of the genitive singular with the nominative plural in o-stems might have induced a genitive singular of the ā-stems that is identical with the nominative plural, but for the fact that the o-stems did not become absolutely identical in these endings until about 150 B.C.: before that the nominative plural was written quite consistently EI in inscriptions and the genitive singular was written I. For the dative singular, there is no analogy whatever along these lines.

Now the familiar explanation of the normal ending -ae in the genitive singular is that the older -ās, still seen in pater familias 'father of the household or family,' was remade to -ā-i after the of o-stem genitives; and that this dissyllabic -āi in rapid speech was contracted to a diphthong, and later became -ae: cf. Sommer, op. cit. 325-27. The failure of the -ai to eventuate in -ī, as in existimō from *ex-aistumō, is due to its late origin, after the original -ae has traveled part of the way to the monophthong.

Similarly, the -ae which replaced original -ās in the nominative plural is, as Sommer, op. cit. 329, states, the product of a long diphthong which develops secondarily by the analogy of the -oi of o-stems: stem vowel -o- is to plural -oi as stem vowel -a- is to plural -ai. This long vowel also, as being of later development, has escaped the development to was the normal fate of the unaccented short diphthong.


Yet if this be the origin of the plural ending in the a-stems (and I believe that it is, for I cannot agree with Brugmann, Kuhns Zeitschr. f. vergl. Sprachforschung 27.199 ff., that this plural is an original dual in -ai; cf. Sommer, op. cit. 329-30), the fact that it is a long diphthong and not a short one is of importance. For an -ai which developed from a proportion o:oi=a:x, was generated at a time when the short o in unaccented syllables had not weakened, and therefore presumably before the short a in unaccented syllables had weakened. But in early Italic times the long diphthongs became short before consonants, and, if ydiphthongs, lost their y before vowels. Thus our -āi should in continuous speech at once have become either -ai or -ā. So early an -ai would almost certainly have weakened into -ī. We must therefore ascribe to the long vowels of the paradigm an influence which held the length of this diphthong until after original short -ai in the unaccented position had become -ei on its way to -i.

For the paradigm of the a-stems was full of long a's, in precisely the positions where the o-stems had short vowels; we need cite only nom. sing. -ā, acc. sing. -ām, as compared with -os and -om. Thus even this interpretation of the nominative plural -ae needs an analogic retention of length in the diphthong.

But a similar analogical retention of length in the final -ai of the dative singular, where the long diphthong has come down from the earliest stages of Indo-European to which we can go back, is adequate to explain the development of the sounds in this form also. We may note that a final -āi may stand in any one of three positions: before a vowel, where it would become -a; before a consonant, where it would be

shortened to -ai; before a pause, where its development can only be conjectured. But in the dative-ablative plural, the long diphthong is always before a consonant, the -s which is the final sound of the word. Therefore analogies would be less likely to be at work in the dative-ablative plural ending, than in that of the dative singular, where the sound was subject to three regular developments, two of which at least were absolutely unlike.

Before taking up the history of the genitive ending, I must record my belief that H. Ehrlich was right in his Untersuchungen über die Natur der griechischen Betonung, 66 ff., when he championed the old theory that the genitive of the o-stems is in Latin historically a locative. The differentiation in the older Latin inscriptions between -i of the genitive singular and the -ei of the nominative plural he explains as due to a difference in the diphthongs; that the original -ei in the locative-genitive became a monophthong at a much earlier date than did original -ai and -oi, for the latter had first to weaken to -ei before they went on to -1. Sommer's objections, in his Kritische Erläuterungen z. lat. Laut- u. Formenlehre, §87, do not convince me; his strongest point is that there is insufficient formal identity between the two cases, to lead to the use of the locative form in a genitive function. Yet while in form this confusion may be extremely rare, there are ample overlappings in the syntactical field: the genitive in Aen. 1.438 fastigia urbis 'roof-tops of the city' expresses virtually the same idea as the locative phrase in Aen. 1.441 lucus in urbe media 'a grove in the center of the city'; regina deum 'queen of the gods' might have been 'queen among the gods' without surprising us, cf. the phrase primus inter pares; words of fullness may be attended in Latin by either the genitive or the ablative, and genitive and ablative phrases are alike used to express description or quality. Although some of these ablative uses are not locative in origin, still we might note that in Sanskrit the locative case reaches out to express goal, cause, agent, specification, none of which functions properly belong to it; and it is rash to claim that a locative form could not come to be the normal form of the genitive, especially when the argument in its favor is so ably maintained as it is by Ehrlich.

Granting now that the o-stems have a genitive which is historically a locative form, I am tempted to see in the genitive in -ae of ā-stems a similar locative in -ai, restrained from weakening to in the same way as was the ending of the dative singular and that of the nominative plural. But this cannot be demonstrated; for we have indubitably the original genitive in -ās and the remade form in -ãi, so that a third form

is unlikely, especially as -ae can be derived from -ai by shortening of the antevocalic long vowel and contracting of the resulting -ai, even as *rē-i becomes rei and then, at times, a monosyllable, despite the resistance of the paradigmatic stress. Strictly, we should expect the -d-i to maintain itself after a vowel, even as we have normally aciei and diëi, with e unshortened because of the preceding vowel. Certainly, the genitive ending is seldom, if ever, elided before an initial vowel in the writings of Plautus; cf. F. Leo, Plautinische Forschungen,2 334-360. We are forced to the conclusion that -ā-i had come into use so recently that it had only in part developed to a monosyllabic -ai in the time of Plautus, who could then use either form at will, though he preferred the dissyllabic ending. Monosyllabic -ai would be normal after a consonant, but dissyllabic -āi would be normal after a vowel. Levelings took place, as was natural, and the diphthong -ai became -ae at the time when accented -ai suffered this change. A few instances of -ai survive in later poets only as archaisms, for the monosyllabic ending was presently extended to the position after a vowel, to the exclusion of the dissyllabic form. But this ending -āi came into existence too late to yield an -ai which could weaken to -ei and to -i.

My thesis is therefore that the dative ending, though original, maintained its long diphthong by analogy; that the nominative plural next got a long diphthong by an analogical formation; that the genitive singular presently acquired a diphthongal ending (after consonants), as a contraction of a dissyllabic form created by analogy; that all three forms, by the influence of the other forms of the declension, kept the diphthong unchanged until a time when it developed like the ai of accented syllables.

This, by the way, suggests that the accent was less strongly of a stress or energy quality at the time when the retained -ai became -ae than when the original unaccented ai became ei on its way to i. The earlier was the primitive Italic accent on the initial syllable, the later was the historical Latin accent, regulated by the length of the penult; but I cannot pursue this theme at the present time.





Does the history of the liquid and nasal sounds in Semitic disclose any development parallel to that of the nasal and liquid sonants (m, n, k, r) of Indo-European? The problem has never been treated adequately.. Even in the latest works on Hebrew grammar (there has not been published recently a single grammatical study in any other Semitic language that is more than elementary) the question is either entirely ignored or else doubt is expressed whether the assumption of a vocalic function by sounds other than the traditional vowels is in Hebrew at all justifiable.1 And yet what is the value of the lamed in heb'l 'vanity' or the nun in eb'n 'stone'? Or how else can we explain forms like qorobkem 'your approaching' Dt. 20.2 from qorbkem, or the Babylonian mamilkôp 'kingdoms' over against the Tiberian mam/kôp 1 Chr.29.30, 2 Chr. 20.29? We are forced to assume that the liquids resh and lamed served as vowels during the transitory period. For, to take the latter instance, the form mamilkôp can thus be easily derived from the otherwise expected mamlļkop. The latter first became mamlkôp by syncope, then between consonants assumed the function of a vowel thus yielding the form mamĮkôp (samprasāraṇa) and at length a gliding vowel developed between m andĮ (anaptyxis), thus ultimately producing the form mamilkôp.

The present writer has elsewhere expressed the opinion that the troublesome Shwa medium of the Hebrew grammarians may be intimately connected with the vocalic potentialities of the liquids and nasals. Professor Montgomery, according to a personal communication, has arrived at some very important results concerning Hebrew phonology and morphology through a careful observation of the Hebrew

1 Cf. Bauer and Leander, Historische Grammatik der Hebräischen Sprache, Halle 1922, 50d.

'Cf. Bergsträsser, Hebräische Grammatik, Leipzig 1918, par. 23d.

'In a monograph on, The Pronunciation of Hebrew according to Origen's transliterations in the Hexapla, shortly to be published by the Dropsie College.

« ZurückWeiter »