LANGUAGE.] PROVE N g A L 869 would run approximately through Blaye, Coutras (Gironde), Riberac, Nontron (Dordognc), Bellae (Haute Vienne), Boussac (Creuse), Montlugon, Gannat (Allier), Montbrison (Loire). Start ing eastward from Lyons or thereabouts, there appears a notable linguistic fact which is observable in varied proportions in the departments of Ain, Isere, and Savoie, and in Romanic Switzer land. This is, that accented Latin a, in an open syllable, when preceded by a mouillurc or palatalization (whatever the origin of this), becomes e ; on the contrary, when there is no mouillurc, it remains a. Thus we find in the Meditations of Marguerite d Oingt (Lyons, about 1300) ensennier, deleitier, as against desirrar, recontar, regardar. Of these two endings, the former, -ier, is that which is found regularly in French, the second that which is regular in Pr. Pure Pr. would have -ar in both cases (ensenkar, deleitar, desirrar, &c.); Fr. would have -ier (cnseignier, delitier) and -er (desirer). Prof. Ascoli has given the name of Franco-provenqal (franco-provenzale) to the varieties of Romanic in which we find this duality of treatment of Latin a, according as it was or was not preceded by a palatalized sound. Lat. e, I become close c (Ital., e chiuso; Fr. e): habere aver, credet ere, me(n)sem mes, fidem/e, pilum pel. This character is not only common to Italian and Spanish, but also extends over the French domain on its western side as far as Britanny. Certain exceptions noticed in French do not occur in Pr. : thus mercedem, cera, pr (eh) e(n)s um, venenum, which give in Fr. merci, cire, pris, venin, where we should have expected mcrcei, ceire, preis, venein, give regularly in Pr. merce, cera, prcs, vcre. Lat. e preserves, as in Italy, the sound of open e (Ital., eapcrto): pedem, pe, lev at, leva, leper em, lebre. In certain determinate cases, this c from about the 13th century onwards may diphthongize to ic: ego, cu, then ieu, heri, er, ier, for it, fer, ficr. Lat. I is preserved, as in all the Romanic languages : a in i c u ni, ami, r I p a, riba. Lat. I is treated like I long when it precedes (with hiatus) another vowel : pium, pi a, piu, pia, via, via, ligat, Ha. Lat. o, u result in one and the same sound, that of Italian u, Fr. ou (Eng. oo). The same phenomenon takes place in the north of Italy, and in the Romanic of Switzerland. This sound, which is styled by the Donat Proensal the o estreit (close o), is usually symbolized in the early texts by simple o, and is thus confounded in spelling, though not in pronunciation, with the open o (o larc of the Donatz Proensals] a which comes from Lat. o. Lat. u becomes u (i.e., Fr. u), as all over France, and also in North Italy and Catalonia : mil rum, mur ( = miir), durum, dur ( = diir). Lat. au is rigorously preserved over the whole extent of the Pr. domain : auruin, aur, a laud a, alauza, pauper em, paubrc. At present the preservation of Lat. au does not extend much out side the Prov. domain ; it is, however, found in certain parts of the Ladino zone in Switzerland (upper Rhine valley), and in Friuli, and it is to be supposed to have been once general over the whole of that zone. It is attested as late as the 16th century in the Vaudois valleys of Piedmont, and there are also examples of it in old Catalan. Elsewhere the diphthong has regularly become open o (aur urn, It. and Sp. oro, Fr. or, &c.). Atonic Vowels. The atonic vowels (i.e., vowels of the unac cented syllables) which precede the accented syllable present no very characteristic phenomenon ; but it is otherwise with those that follow the accented syllable, the post-tonic vowels. The Pr. is one of the Romanic idioms which, like the French, but unlike the Castilian and the dialects of central and northern Italy, admit of only one syllable after the accent. But the rules are not quite the same as in French. In French the only vowel which can stand after the accented syllable is " e feminine," otherwise called "c mute." In Prov. a and e are the most frequent vowels in this position, but i and o also occur. In French the first of the two post-tonic vowels of a Lat. proparoxytone always disappears ; in Prov. it tends to be preserved, when followed by one of the consonants n, r, I, d: te rminum, tc rmcn, ho minem, o mcn, a ngelum, a ngel, se calem, sc gucl, cre scere, crci sser, te pidum, te bez. Finally, Prov. presents in certain words coming from Lat. pro- paraxytones the trace of forms which (like Ital.) admitted two atonic vowels after the accented syllable : thus we have portvgue and po rgue (po r ticu m), Falre ga, a place name, and/a rgra (f a br ica), perte ya and jic rga (pe rtica), feme na anidfe mna (fe in ina). We have also lagre ma (la cry ma), but a form accented like Fr. larme does not exist. There seems to be no doubt that these forms in which a displacement of the Latin accent is observed were at an earlier period pronounced as proparaxytones ( po rtegue, fa brcga, pe rtega, fe mena, la grcma). Consonants. The boundary usually recognized between Prov. and French is founded upon linguistic characters furnished by the vowels, especially a ; if it had been determined by characters furnished by the consonants, the line of demarcation would have to be drawn farther south, because the consonantal system which is regarded as proper to French really extends in its main features over the northern zone of the Provengal region as defined above. As with the vowels, only a few of the salient facts can here be indicated. C initial, or second consonant of a group, before a (caballum, mercatum), preserves its Lat. sound ( = k) in the greater part of the Prov. region. But in the northern zone it takes- the sound of tch (Eng. eh in chin) as in 0. Fr., and this sound is still pretty well preserved, although there is here and there a Dome, Cantal, Haute Loire, the north of Lozere, of Ardeche, of Drome, of Isere, and of Hautes Alpes, and Castel, Castanet, Cazal, farther to the south. Analogously, g initial, or second consonant of a group, followed by a, becomes j (i.e., dzh = 0. Fr. and Eng. / in jam) in the same zone; Garrica is Jarrija, Jarria hi Dordogne, Correze, Cantal, Haute Loire, Isere, and G arriga farther south. Between two vowels t becomes d : edat, cmpcrador, nadal, amada (seta tern, impcratorem, natal e, a mat a). This was also the case in 0. Fr. until the course of the llth century (honurede, empcredur, lavadurcs, &c., in the Life of fit Alexis). But in the northern zone this d representing a Lat. t fell away as early as in Fr. ; in an 11th-century text from the environs of Valence, we read muraor, coroaa (*muratorem, corrogata), Fr. corvee (P. Meyer, Rccueil d ancicns textes, Provencal section, No. 40). In tbe south, d between two vowels was preserved almost everywhere until about the middle of the 12th century, when it became s (as in Fr. and Eng. zero) : cruzel, azorar, auzir, vezer (crudelem, adorare, audire, vide re). In the 14th and 15th centuries this s, like every z or s soft of whatever origin, was liable to become r (lingual, not uvular): aurir, veren (audire, v id en tern). In Beam and Gascony d. remained; but in the northern zone Lat. d, instead of changing into z, r, disappeared as in Fr. and quite as early. The poem of Boetius, of which the MS. is of the llth century, shows in this respect great hesitation: e.g., (^preserved in chaden, credet, tradar, vcder (cadentem,
- crede dit, *tradare, vide re); d fallen away in crecssen,
fccltat, traazo, vc ilt, fiar (*credess en t, fide li tat em, *tra- dationem, *vidutum, p. pie. of videre, fidare). One of the most general facts in Pr. is the habit of rejecting Lat. final t, of which examples to any number are presented by the verbs. In Fr. this t was formerly retained when it followed a vowel which remained, aimet, cntrct (a mat, intrat), and still remains (in writing at least) when, in Latin, it follows a consonant, aimcnt, fait, v#(amant, facit, *fact, vivit, *vivt); but in Pr. the t is dropped in all cases, even in the most ancient texts : aman, fai, viu. Yet in the northern zone we find the t retained in the 3d per. pi. of verbs, -ant, -ont (Lat. -ant, -unt). //has gone completely (or at least only appears through orthographic tradition, and very intermittently, (h)erba, (h)onor, (h)umil, &c. ), not only in words of Lat. origin, which is the case in 0. Fr., but even in Teutonic words (anta, ardit, arenc, ausberc, elm, Fr. honte, hardi, harcng, haubcrt, hcaume, with h aspirated). By this feature, the northern limits of which are not yet well determined, the Pro- vensal attaches itself to the Romanic of the southern countries. N final, or standing in Lat. between two vowels of which the second is to be dropped, disappears in the whole central part of the Pr. domain: gran gra, ben be, en e, ivn vc, Jin fi, un u (granum, bene, in, venit, fin cm, unum). The forms with n belong to the eastern part (left of the Rhone), the western part (Gascony, but not Beam), and the region of the Pyrenees. It is possible that this loss of n went along with a lengthening of final vowel ; at least, in Bearnese when the n falls away the vowel is doubled: capcraa, bcsii, boo (capellanu m, v i c i n u m, b o n u m), &c. These are the most important characteristics of the consonants, in relation to the extent of space over which they prevail. Others, which appear only within a more limited area, are perhaps more curious on account of their strangeness. It will suffice to mention a few which belong to the district bounded on the west and south by the Atlantic, the Basque provinces, and the Pyrenees, and which extends northward and eastward towards the Garonne and its affluents, as far as the Gironde. (This includes Beam, Bigorre, and Gascony. ) Here the sound v no longer exists, being replaced generally by b ; between two vowels, in Gascony, by u with the sound of Eng. w. Initial r assumes a prosthetic a : arram, arre, Arrobcrt (r a m u m, r em, Robert u m). LI between two vowels becomes r : apcrar, capcran, or (Beam) capcraa, bera, era (apel- lare, cap el Ian urn, bella, el la). On the contrary, at the end of words (viz., in Romanic) II becomes g or t, d; the former change seems to belong rather to Hautes and Basses Pyrenees, Landes, the latter to Gironde, Lot et Garonne, Gers : eg, cd, ct (ille), arrasteg, -ed, -et (r a s t e 1 1 u m), casteg, -cd, -et (c a s t e 1 1 u m), capdcg, -cd, -ct (capi tellum), whence Fr. cadet (in 16th century capdct, originally a Gascon word). For further details upon the consonants in this region of south-west France, see Romania, iii. 435-38, v. 368-69. Flexion. Old Provenal has, like Old French, a declension con sisting of two cases for each number, derived from the Latin nominative and accusative. In certain respects this declension is
more in conformity with etymology in Proven9al than in Old