21. Between the 1st and the 7th century of our era, the Classic Latin quantity died out: it had apparently disappeared from unstressed vowels as early as the 4th century, from stressed by the 6th. It left its traces, however, as we have seen, upon accentuation (§ [16]), and also upon vowel quality, the originally long and short remaining differentiated in sound, if they were accented. Of the unaccented vowels, only i shows sure signs of such a differentiation, and even for i the distinction is evident only in a final syllable: vēnī vēnĭt ＞ vẹnị vẹnįt.
ACCENTED VOWELS.
22. The vowels of Vulgar Latin are a, ẹ, ę, ị, į, ọ, ǫ, ụ, ų, with the diphthongs áu and úi; the old æ and œ had become identical in sound with ę and ẹ. As early as the 3d century of our era, į was changed, in nearly all the Empire, to ẹ, and thus became identical with the vowel coming from original ē. A little later, perhaps, ų, in the greater part of the Empire, became ọ, thus coinciding with the vowel that was originally ō. Ypsilon, in words taken from the Greek, was identified, in early borrowings, with Latin u; in later ones, with Latin i: βύρσα ＞ Pr. borsa, γῠρος ＞ Pr. girs. Omicron, which apparently had the close sound in Greek, generally (but not always) retained it in recently borrowed words in Vulgar Latin: τόρνος ＞ tọrnus (cf. Pr. tọrn), but κόλαφος ＞ cŏlăphus = cọlapus or cǫlapus (cf. Pr. cǫlp).
The development of the Vulgar Latin vowels in Provençal will now be examined in detail:—
a
23. Cl. L. ā, ă ＞ V. L. a ＞ Pr. ą: ărbŏrem ＞ ąrbre, grātum ＞ grąt, măre > mąr.
1. The ending -arius shows an irregular development in French and Provençal, the Provençal forms being mainly such as would come from -ĕrius; as in parlier, parleira. In the earliest stage we find apparently -ęr´ and -ęr´a; then -ęr´ and -ęir´a; next -ęr, -ięr and -ęira, -ięira; finally, with a reciprocal influence of the two genders, -ęr, -ięr, -ęir, and -ęra, -ięra, -ęira, -ięira: caballarium ＞ c(h)avaler -ier, -eir, ✱man(u)aria ＞ manera -iera -eira -ieira. The peculiar treatment of this suffix has not been satisfactorily explained. See E. R. Zimmermann, Die Geschichte des lateinischen Suffixes -arius in den romanischen Sprachen, 1895; E. Staaff, Le suffixe -arius dans les langues romanes, Upsala, 1896, reviewed by Marchot in Zs., XXI, 296, by Körting in Zeitschrift für französische Sprache, XXII, 55; Meyer-Lübke, Gram., I, 222, § 237; Zimmermann in Zs., XXVI, 591; Thomas in Rom., XXXI, 481 and in Bausteine zur romanischen Philologie, 641. The likeliest theory is that of Thomas: that -arius was associated with the Germanic ending -ari and participated in the umlaut which affected the latter; cf. Phon., pp. 34-36.
2. In Gascony and Languedoc ei is used for ai ＜ habeo. The ei perhaps developed first as a future ending (amar -ei) by analogy of the preterit ending -ei (amei): see Morphology, §§ [152], 1, [162], (4), [175], (4), where this latter ending is discussed also. For a different explanation, see Meyer-Lübke, Gram., I, 222, § 237.
3. A few apparent irregularities are to be traced to the vocabulary of Vulgar Latin. For instance, Pr. sereisa represents, not Cl. L. cĕrăsus, but V. L. cĕrĕsĕa: see Meyer-Lübke, Einf., § 103. Uebre is from ✱ŏpĕrit, or apĕrit modified by ✱cŏpĕrit = cōperit. Voig is from ✱vŏcĭtum = vacuum: Einf., § 114.
4. Such forms as fontaina = fontana ＜ fontāna, etc., and tres = tras ＜ trans, etc., are French or belong to the borderland between French and Provençal.