870 French, having been less influenced by analogy. The following are the types of this declension, taking them in the order of the Lat. declensions. 1. Words in -a coming from Lat. 1st decl., increased by certain words coming from Lat. neuter plurals treated in Prov. as feminine singulars ; one form only for each number : sing, causa, pi. causas. 2. Words of the Lat. 2d decl., with a few from the 4th ; two forms for each number : sing, subject cavals (cabal Ins), object caval (cab all urn); pi. sub ject caval (c a b a 1 1 i), object cavals (c a b a 1 1 o s). 3. Words of Lat. 3d decl. Here there are three Lat. types to be considered. _ The first type presents the same theme and the same accentuation in all the cases, e.g., c a n i s. The second presents the same accentuation in the nominative singular and in the other cases, but the theme differs: co mes, co mitem. In the third type the accentuation changes : pecca tor, peccato rem. The first type is naturally confounded with nouns of the 2d decl. : sing., subj. cans or cas, obj. can or ca. The second and third types are sometimes followed in their original variety; thus corns answers to co mes, and co mtc toco mitem. But it has often happened that already in vulgar Latin the theme of the nominative singular had been refashioned after the theme of the oblique cases. They said in the nom. sing, heredis, parent! s, principis, for heres, parens, princeps. Consequently the difference both of theme and of accentuation which existed in Lat. between nominative and accu sative has disappeared in Pr. This reconstruction of the nomina tive singular after the theme of the other cases takes place in all Lat. words in -as (except abbas), in those in -io, in the greater part of those in -or, at least in all those which have an abstract meaning. Thus we obtain bontatz (bonitatis for bonitas) and bontat (bonitatem), ciutatz (civitatis for civitas) and ciutat (c i v i t a t e m), amors (a m o r i s for a m o r) and amor (a m o r e m). All present participles in the subject case singular are formed in this way upon refashioned Latin nominatives : amans (a mantis foramans) amant (amantem). It is to be remarked that in regard to feminine nouns Pr. is more etymological than Fr. In the latter feminine nouns have generally only one form for each number; bont6 for the subj. as well as for the obj. case, and not bontes and bonte; in Pr. on the contrary bontatz and bontat. Still, in a large number of nouns the original difference of accentuation between the nominative singular and the other cases has been maintained, whence there result two very distinct forms for the subj. and obj. cases. Of these words it is impossible to give a full list here ; we confine ourselves to the exhibition of a few types, remarking that these words are above all such as designate persons: a bas aba t, pa stre jxisto r, sor soro r, cantai re cantado r (cantator, -or em), emperai re cmperado r, bar baro , compa nh companho; lai re lairo (latro, -on em). To this class belong various proper names : E ble Eblo , Gui Guio; Uc Ugo: A few have even come from the 2d decl., thus Pci re Pciw, Pans Ponso , Ca rle Carlo , as if the Latin types had been Petro, -onem, Ponso, - o n e in, Carlo, -onem. We may mention also geogra phical adjectives, such as Bret Breto , Bergo nhz Bergonho , Gasc Gasco; &c. The plural of the 3d decl. is like that of the second: subj. aba t, soro r, cantado r, emperado r, baro , companho , lairo", obj. aba tz, soro rs, cantado rs, cmperado rs, baro s, companho s, lairo s, as if the Lat. nominative pi. had been abbati, sorori, cantator i, &c. It is barely possible that such forms actually existed in vulgar Latin ; no trace of them, however, is found in the texts, save in the glosses of Cassel (8th c.), sapienti for sapientes, and in a great many ancient charters p a r e n t o r u m, which implies a nominative p a r e n t i. The words of the 4th and 5th^ declensions present no points requiring mention here. This declension of two cases is a notable character of the whole Romanic of Gaul, north as well as south, i.e., Freuch as well as Provengal. It must be noted, however, that in the south-west it existed only in a very restricted fashion. In the old texts of Gascony it is no longer general in the 13th century. In Beam it appears to have been completely unknown, the nouns and adjs. having only one form, usually that of the obj. case. In Catalan poetry its application is often laid down in the 13th century, but as the charters and documents free from literary influence show no trace of it, its introduction into the poetry of this country may be assumed to be an artificial fact. In the region where it is best observed, i.e., in the centre and north of the Prove^al territory, it tends to disappear from ordinary use already in the 13th century. The poet grammarian Raimon Vidal of Besalu, who flourished about the middle of the century, points out in various troubadours transgressions of the rules of declension, and recognizes that in PROVENQAL [LANGUAGE. modern Pr. the words in the ending -ai re (answering to Lat -a t o r) are as frequent as those in -adou- (repr. -atorem). But there is a slight difference of meaning between these two suffixes. Adjectives, generally speaking, agree in flexion with the nouns.
- there is one fact particular to adjectives and past participles
which 13 observed with more or less regularity in certain 12th and 13th century texts. There is a tendency to mark more clearly than in the substantives the flexion of the subj. pi., chiefly when the adj. or participle is employed predicatively. This is marked by the addi tion of an i, placed, according to the district, either after the final consonant, or else after the last vowel so as to form a diphthong with it. The following are examples from an ancient translation of the New Testament (MS. in library of the 1 alais Saint-Pierre, Lyons, end of 13th century): "Die a vos que no siatz consirosi" (no solliciti sitis, Mat. vi. 25); "que siatz visti d els" (ut videamini ab eis, Mat. vi. 1); "e davant los reis els princeps seretz menadi" (et ad presides et ad reges ducemini, Mat. x. 18). In charters of the 12th and 13th centuries we find in the subj. case pi., and especially in this predicative use, payaiy, certiftaih, acossailhaih, representing pagati, certificati, adconsiliati. It is in the verbs that the individuality of the different Romanic idioms manifests itself most distinctly. At a very early date the etymological data were crossed, in various directions and divers manners according to the country, by analogical tendencies. The local varieties became little by little so numerous in the Romanic conjugation that it is not easy to discover any very characteristic features observed over a territory so vast as that of which the limits have been indicated at the commencement of this article. The following are, however, a few. The infinitives are in -ar, -er, -re, -ir, corresponding to the Lat. -are, -ere, -ere, -ire, respectively ; as in the whole Romanic domain, the conjugation in -ar is the most numerous. The table of verbs, which forms part of the Pr. grammar called the Donate Proensals (13th century) contains 473 verbs in -ar, 101 in -er and -re, 115 in -ir. In the -ar conjugation we remark one verb from another conjugation : far (cf. It. fare) from facere. The con jugations in -er and -re encroach each upon the territory of the other. The three Lat. verbs c a d e r e, c a p e r e, s a p e r e have become -er verbs (caze r, cabe r, sabe r) as in Fr. chcoir, -ccvoir, savoir ; and several other verbs waver between the two : credc r, crcer, and crei re (c r e d e r e), qucre ~r and que rre (q u aj r e r c) . This fluctuation is most frequent in the case of verbs which belonged originally to the -ere conjugation: ardc r and a rdre, plazc r and plai re, taze r and tai re (ardere, p lac ere, tacere). Next to the -ar conjugation, that in -ir is the one which has preserved most formative power. As in the other Romanic languages, it has welcomed a large number of German verbs, and has attracted several verbs which etymologically ought to have belonged to the conjugations in -er and -re : emplir (i m p 1 e r e), jauzir (g a u d G r e), cosir (con suere), crcbir (e rip ere), fugir (fug ere), sequir (* seq uer e = sequi). Except in the -ar conjugation, the ending of the infinitive does not determine in a regular manner the mode of forming the different tenses. The present participles are divided into two series : those in -an (obj. sing.) for the first conj., those in -en for the others. In this the Pr. distinguishes itself very clearly from the French, in which all present participles have -ant. There is also in Pr. a participial form or verbal adjective which is not met with in any other Romanic language, except Romanian, where more over it is employed in a different sense ; this is a form in -do r, -doi ra, which supposes a Lat. type -tor ins, or-turius; the sense is that of a future participle, active for the intransitive verbs, passive for the transitive: cndcvenido r, -doi ra, "that is to happen"; fazedo r, -doi ra, " that is to be done" ; punido r, -doi ra, "to be punished." In conjugation properly so called, we may remark the almost complete disappearance of the Lat. preterite in -dvi, of which traces are found only in texts written in the neighbourhood of the French-speaking region, and in Beam. In return, a preterite which seems to have been suggested by the Latin dedi, has increased and become the type of the tense almost everywhere in the -ar conjugation, and in many verbs in -er and -re : amei , amc st, ame t, ame m, ame tz, ame ron. In Fr. there is a form like this, or at least having the same origin, only in a small number of verbs, none of which belong to the first con jugation, and in these only in the 3rd. pers. sing, and pi. (perdie, perdiereni ; cntendie, entcndiercnt, &c.) It is well known that re duplicated preterites had greatly multiplied in vulgar Latin : there have been recovered such forms as a s c c n d i d e r a t, o s t e n d e d i t, p a n d i d e r u n t, a d t e n d e d i t, i n c e n d i d e r a t, &c. (see Schuchardt, VoJcalisimis dcs Vulcjarlatcins, i. 35, iii. 10 ; cf. Romania, ii. 477). But, in order to explain the Pr. form -ei, -cst, -et (with open e), we must suppose a termination not in -idior - e d i, but in - e d i. In the western region the 3d pers. sing, is generally in -ec, probably by analogy with preterites like lee, crcc, dec, sec, formed after the Lat. type in -u i. Another notable peculi arity, of which Old Fr. shows only rare traces, in texts of a very remote period, is the" preservation of a preterite in -ara or -era, de rived from the Lat. pluperfect, ama ra or ame ra, "I loved." The former comes directly from Lat. a m a r a m, the latter has been in fluenced by the ordinary preterite in -ci. This preterite is used with the sense of a simple past, not of a pluperfect, and conse quently is an exact doublet of the ordinary preterite, which explains
how it was at length eliminated almost everywhere by the latter, of