Romance

Romance LANGUAGES, the name generally adopted for the modern languages descended from the old Roman or Latin tongue, acted upon by inner decay or growth, by dialectic variety, and by outward influence, more or less marked, of all the foreign nations with which it came into contact.

During the middle ages the old Roman Empire or the Latinspeaking world was called Romania, its inhabitants Romani (adj. Romanicus), and its speech Romancium, Vulgar Romancio, Italian Romanzo, from Romanice loqui=to speak Romance; in Old French nominative romanz, objective roman(t), Modern French roman, " a novel," originally a composition in the vulgar tongue. In English some moderns use Romanic (like Germanic, Teutonic) instead of Romance; some say Neo-Latin, which is frequently used by Romance-speaking scholars. By successive changes Latin, a synthetical language, rich in inflexions, was transformed into several cognate analytical tongues of few inflexions, most of the old forms being replaced by separate form-words. As the literary language of the ancient Roman civilization died out, seemingly extinguished by the barbarism of the middle ages, all the forms of the old classical language being confounded in the most hopeless chaos, suddenly new, vigorous and beautiful tongues sprang forth, ruled by the most regular laws, related to, yet different from, Latin. How was this wonderful change brought about? How can chaos produce regularity? The explanation of this mystery has been given by Diez, the great founder of Romance philology. The Romance languages did not spring from literary classical Latin, but from popular Latin, which, like every living speech, had its own laws, not subject to the changing literary fashions, but only to the slow process of phonetic change and dialectic variety. It is interesting to observe that much that is handed down to us in the oldest Latin literature (notably in the vocabulary) reappears in the most recent phase of Latin - the Romance languages. Thus, a verb nivere, " to snow," is known to Pacuvius, but does not again appear until the time of Venantius Fortunatus, and then with a change of conjugation - nivere, while it has now a new term of life in French and Rhaeto-Romanic dialects. It is obvious that there was no break of continuity in the vulgar language, for if in the later imperial ages a verb had been formed from nix, nivis, it must have been nivare, or niviare (Fr. neiger). Here especially the words of Horace come true: " Multa renascentur, quae jam cecidere, cadentque Quae nunc sunt in honore vocabula, si volet usus, Quem penes arbitrium est et jus et norma loquendi." The present article, embracing all the Romance languages, aims at tracing on the one hand their common origin and their common development, on the other hand at pointing out the peculiarities of the individual languages and the possible explanations of the growth of these peculiarities. Their common development is mainly dealt with under Latin Language. The relation of the early vulgar Latin to the literary language, the spread of Latin following the spread of Roman rule, the prevalence of Latin over Oscan, Umbrian, Etruscan, and late Iberian and Gallic - all these matters concern rather the history of Latin than of the Romance languages. But we may say broadly that the language spoken throughout the Roman Empire at the time of Augustus was fairly uniform, and that naturally differentiations took place (varying according to regions) which were not, however, strongly marked, and which even tended to be obliterated in later times.

The main causes of these variations were twofold. (1) The process of Romanizing the various districts took place at epochs far remote one from the other, and between the earliest and the latest of these epochs Latin itself was modified.' (2) We have the reaction on Latin of the languages of the pre-Roman populations.

Applying this first point of view, we should find that the oldest form of Latin (oldest, that is, for our present purposes) was introduced into Sardinia (238 B.C.); next comes Spain (197 B.C.), Illyria (167 B.C.), South Gaul (120 B.C.), North Gaul (50 B.C.), Raetia (15 B.C.), Dacia (A.D. 107). And we can actually trace some of the results of these differences in date, chiefly perhaps in the vocabulary and morphology of the Romance languages. When, for example, we find the dative illui (Ital., Fr., Rum. lui) missing in the Iberian peninsula, we may infer that it was unknown to the Latin introduced there, and conversely that Latin still used the ancient cova (Sp. cueva, " Cava ") and not the more recent cava (Ital. cava), also demagis or gumia, which we only know from Lucilius, Sp. demas, gomia. We may be justified in assigning to these historic causes the beginnings of the divergence from the original uniformity. Neither active intercourse, nor the dislocations of tribes and populations brought about by the exigencies of military or colonizing enterprise, ever effected a complete fusion of these divergences. To this we must add, as a second element, ethnic considerations.

To begin with, we seem to find in Italy itself, among the Italic population in country districts, the survival of isolated forms which had been discarded by the literary language with its levelling tendencies, and in consequence also by what may be called " Average Latin " (Durchschnittslatein). In early Latin. d becomes r before labials, e.g. ar me advenias occurs in Plautus; arvorsus, arger from *arfger are the ancient forms. Only arbiter has survived as a word of the official language and because in general feeling the noun was consciously connected with the verb baetere, though it was soon discarded. Arger, under the influence of aggerere, aggestus, became agger, and arvorsus was displaced by advorsus. In Abruz. we have arbendd, " to repose," beside Sicil. abbintari which suppose *arventare beside adventari; Abruz. armuri, " to put out the fire," represents Lat. *armoriri instead of admoriri; arbukkd is found beside Ital. abboccare. All these forms are only attested in Italy, and they might by reason of their prefix be classed as Umbrian, since in Umbrian ar for ad is even commoner, cf. the place-name Arestaffele in Molise, which in Latin would be ad Stabula, save that the limitation to the cases that are in line with the Latin rule prove precisely that this is not a case of Umbrian influence, but of a preservation of ancient and popular forms. Beyond the limits of Italy arger has been preserved, e.g. Sp. arcen, and not only Ital. argine; further armissarius, " stallion," in the Lex Salica and in Rum. armesarii; perhaps Sp. almuerzo, " breakfast," for *armuerzo beside Lat. admorsus. In the second place we have, especially in Italy, clearly UmbroOscan forms. Contrary to Latin use, these two dialects, the most important in ancient Italy, have f between vowels from an early bh, dh, as against Latin b, d; and Umbrian, Paelignan, &c., e, o, from an early ei, ou, as against Latin i, u. Thus crefrat (in the glosses), as against Latin cribrat, is both by right of its vowel and consonant, an Umbrian form. And with this we must compare Ital. bifolco beside Lat. bubulcus; Ital. taffiari, " to feast," beside tabulari; tafano, " horsefly," beside Lat. tabanus; bufalo, beside Lat. bubalus. Further, Neap. Ottufro, " October," morfende, " eyeteeth," Lat. mordente, &c. There is a special interest in cases like the French mandrin beside Ital. manfano. What has come down to us is manphur, which is not Greek, its ph notwithstanding, but which owing to its f we must take to be Osco-Umbrian; while the corresponding Latin form would be *mandar. The Latin supplies the French, the Osco-Umbrian the Italian form. As to the other instance, Varro points to vella beside villa as rustic, and to this we must add Ital. stegola, Sardin. isteva, Sp. and Port. esteva (*steva for stiva), " plough tail "; Ital. elce, Sardin. elige, Fr. yeuse, " holly " (*elex for ilex), or Ital. pommice, Fr. ponce, Sp. pomez, " pumicestone " (*pomice for pumice). It must not be overlooked that the last word denotes an object found chiefly in Sicily and near Naples, that is, in the ancient seat of the Oscans. It will be clear that we are dealing chiefly with words connected with agriculture, and it is remarkable that those of our second category spread all over the empire, while those of the first were entirely, or almost entirely, limited to Italy.

As a parallel we may cite the vocabulary of North and South Gaul, which yields a number of Gallic elements, and one may safely infer that in the first few centuries after the Roman conquest these elements were more numerous than at a later stage, and there is in fact a definite justification for this inference. The socalled Endlichers glossary of the 5th century is a compilation, by a native of South Gaul, of Gallic words which were clearly at that time still current in the south of France. 2 And in this we have not only dunum, " montem," cambiare, " pro re dare" (Fr. changer); caio, " breialo sive bigardio " (Fr. quai); nanto " valle," Savoy. nd, " stream," but also avallo, " poma," which was lost in later times but is preserved in its derivative amelanche, " medlar." Another Gallic word recorded by ancient tradition - tegia, "hut " - still exists to-day with this meaning in the Venetian and Raetic Alps, and moreover plays an important part in toponomy - Fr. Arthies from Gall. are Tegias, " at the huts," N. Ital. Tezze; but in the oldest Gallo-Romance it may have been in use as an appellative, and thence have passed into Basque - e.g. Basq. tegi, " hut." The permeation of the Latin vocabulary by Gallic elements dates from the time of the contact of Gauls and Roman forces. Many of these elements - e.g. bracae, camisia - were widely used at so early a stage as to have penetrated into Rumania (Rum. imbrdcci, " put on," cameasa, " chemise"); others again have scarcely, if at all, passed beyond their ancient limits, even those that Roman literature has preserved for us. It is true that Martial says " Barbara de pictis veni bascauda Britannis Sed me jam mavult dicere Roma sibi," but only in France has bachoue been preserved up to the present, while so far no traces of bascauda have been established for Italy.

Glancing over the Gallic contributions to the Gallo-Romance vocabulary, we see at once that they belong to a considerable extent to the sphere of agriculture, and that among the implements mentioned it is chiefly vehicles of all kinds which have Gallic names. The record of Roman times supplies us with benna, carpentum, carrum, caruca, agredum, petorritum, rheda, but carrum alone gained a firm footing; caruca in the form of charrue, " plough," survives in France, and benna (Fr. banne, Ital. benna) in its ancient home. Under this heading we may perhaps add taratrum, " gimlet," in Isidore, Fr. taribre, Engad. tareder, Sp. taladro, Port. trado; Fr. jante, " felloe of a wheel " (Bret. Kammed), Fr. taranche, Gall. tarinca. With caruca we may class soc, " plough-share," and O. Fr. raie, Mod. Fr. rayon, " furrow," Gallic *rica (cf. Cymr. rhych). A further group is formed by cervoise, " beer," from Gall. cerevisia, O. Fr. braiz, Mod. Fr. brai, " malt," brasser, " to brew," Gall. brace; lie, " yeast." Among the names of plants Gallic betulla has survived wherever the tree is common. Within narrower bounds we find Fr. if, " yew," Gall. *ivum (cf. Ir. eo); probably also *cassanus, " oak," Fr. chene, Prov. casser; Fr. verne, " alder " (cf. Ir. fern and the Gall. place-name Vernodubrum, " alderwater "); beloce, " sloe," bulluca, and S. Fr. aranhon, " sloe " (Ir. airne). Pliny mentions marga, " marl," as being in use among the Gauls as manure for soil, from the diminutive *margila, Fr. marne. An agricultural measure was called arepennis, Fr. arpent. Fields were separated by a hedge - Prov. gorce (cf. O. Fr. gort, " fence"); a hedged-round piece of land is called in French lande, Ir. land. Another method of demarcation was by means of hurdles, Fr. claie, Piedm. cia (cf. Ir. cliath); or of barricades, Fr. combre (whence the verbs encombrer, dncombrer), which corresponds to a Gallic *comboros. Inside the hurdles the sheep and cows were kept whose milk yielded mbgues, " whey " (Ir. medg). The wood needed for the erection of fences was cut with the " wood-knife," Gall. vidubium, Fr. vouge. We may notice further the group broga, " enclosure," " preserve," Prov. brogo and the diminutive brogilo, Fr. breuil. In north Italy we find fruda, " torrent " (cf. Cymr. frwth), which is a parallel to na mentioned above; also Comasc. dren, " blackberry," Ir. dren, " thorn," and (over a large part of north Italy) bar, "" bunch," " tuft," O. Ir. barn. To single out a few words, there is Prov. ban, " horn," Cymr. ban; Piedm. vinverra, from a word that has come down to us as Latin, but is really Gallic: viverra, Cymr. gwywer, Gaelic feoragh, " weasel," and in the RhaetoRom. dialect in Switzerland carmun, from a Gallic carmon, which is cognate with O.H.G. harmo, Mod. H.G' hermelin, " ermine." In this way we might amplify examples, and it should not escape notice that we have to deal chiefly with substantives, with few adjectives and hardly any verbs.' In precisely the same way the Spanish vocabulary must have been seamed with traces of Iberian elements. But the process of elimination took place more rapidly and thoroughly in this case, so that the number of Iberian or Celtic-Iberian words that have resisted time and change is small. On a Latin inscription from Spain we find Paramus, " plain," and paramo occurs to this day in this sense. As the Iberian does not know the sound p, the word cannot be Iberian, and must be Celtic.

In Isidore we find baia, " bay," which should be read baia, as Sp. and Port. bahia prove - doubtless an Iberian word, since Fr. baie and Ital. baia are forms quite recently borrowed from Spanish. This baia is perhaps somehow connected with the place-name Bayona. Again, the lapides lausiae of the Lex Metalli Vipascensis are Celtic rather than Iberian (cf. Sp. losa, Port. lousa, as well as Prov. lausa, Piedm. losa). Considering our ignorance of Iberian, and the pronounced colouring of Basque by Spanish words, it is not often easy to decide on which side the indebtedness lies when we meet with a word in Spanish and Basque whose etymology is still uncertain.

Much discussion centres round the question as to how far the pre-Romanic nations influenced the phonology of the Romans in the process of their assimilation. Opinions are strongly divergent. While G. I. Ascoli has repeatedly assumed influences of this kind on a large scale, the present writer is very sceptical.' It may be well to give the essential points.

Plautus uses distennite and dispennite instead of distendite and dispendite - forms he imported from his native Umbria. And like the Umbrians, the Oscans too pronounced nn instead of nd. Later we find this same change throughout the whole of south and central Italy, and even in Rome, whereas it is not observed in Tuscany, north Italy and other Romanic countries. We may therefore confidently assume that this is due to a reaction of the Oscan-Umbrian dialects. Similarly it is in accordance with Umbrian pronunciation to convert breathed plosives into voiced after nasals, e.g. iuenga = Lat. juvenca; and similarly we have cinque in central and south Italy beside Tusc. cinque (quinque). But even in this particular the change affects not only the regions of ancient Umbria, but also those of the Oscans and Messapians, though again it must be admitted that we do not know what the pronunciation of the ancient Messapians was. And finally, we find the Latin d represented in Umbrian between vowels by a sound which has a separate sign in the national alphabets and which in Latin is reproduced as -rs. And since the Paelignan alphabet too has a sign for a modified d, one may perhaps assume that in these districts d had a specialized sound as th, or r; and this view agrees with the fact that in the dialects of central and southern Italy d was pronounced sometimes like r, sometimes like th. And probably this sums up all we can say with certainty.

It has always been maintained that French u (pronounced as German u), derived from u, is due to the influence of Gallic. The u (with modern sound) is identified with the whole area of the French language except part of the Walloon, part of French Switzerland, and Piedmont, Genoa, Lombardy, the Grisons, Tirol and the northern part of the Emilia; but not Friuli, Venetia and Istria. On the other hand, the ancient u became i in Cymric, to which ii must be regarded as an intermediary step, that may therefore have existed in Gallic. But in the first place we must observe that Greek writers always render the Gallic u by ov, never by v; that the Romans too write ii, never y; and further, that over a large part of the area ii came in comparatively recently. Secondly, in Gallic inscriptions the combination CT is frequently replaced by XT, so that the Irish pronunciation cht (Ir. nocht, " night ") is as old as Ancient Gallic. And since the preliminary stage of the Fr. fait from factum, nuit from node, is likewise cht, it is natural to suppose a relation between these facts, and all the more because the Iberian Peninsula on the one hand, and a large part of the western and central area of upper Italy on the other, show an identical process; but in Venetian, central and southern Italy ct became tt. Thirdly, nasalized vowels are in evidence chiefly in the ancient seats of the Celts - in northern and 'southern France, in Piedmont, Genoa, Lombardy and partly in Raetia, also in Portugal, but not as far as southern Emilia. At this point again evidence from the Gallic fails completely. Finally, an attempt has been made to trace back the general characteristics of the French and the Gallo-Romanic dialects of Italy to the peculiarities of the Gallic accent. It is assumed that there was a decided stress-accent, which brought about an over-emphasis of the stressed syllable at the expense of the unaccented ones, with the result of a marked weakening of the unaccented vowels, and particularly of those following the stressed syllable. Here again we can only ' Cf. R. Thurneysen, Keltoromanisches (Halle, 1885); W. MeyerLiibke, Einfi Krung in die romanische Sprachwissenschaft, p. 38 ff. a G. I. Ascoli, Una Lettera glottologica (1880); Archivio glottologico italiano, x. 260; Sprachwissenschaftliche Briefe (1887; cf. H. Schuchardt, Zeitschrift fur rom. Phil. iv. 140 and elsewhere); and Meyer-Lubke, loc. cit. 205 ff.

say that Gallic itself affords no evidence for this assumption, and that, on the contrary, this peculiar accentuation maybe due to other reasons, unknown to us. To turn to morphology, the method of enumerating - as we find it, for example, in Fr. quatre-vingts, &c. - would seem to be Gallic, since it is common to all the Celts.

But even if we admit certain regional variations, all these were overlaid by an " Average Latin " which presents a number of essential features uniformly over the whole area, and which differed from the literary language. These characteristics (in historical sequence) are as follows: (I) Loss of final m in polysyllabic words (which we find exemplified in the very oldest inscriptions); (2) loss of the h- sound, a loss which outside the towns was of great antiquity (cf. anser), and at the beginning of imperial times was fairly common; (3) loss of n before s coupled with the lengthening of the vowel, for which Varro is evidence in his alternations of mensa and ,mesa; (4) the assimilation of rs to ss - e.g. sursum from sursum (Ital. suso, O. Fr. sus, Mod. Fr. dessus). Toward the end of the Republic v is lost before u - e.g. vius instead of vivus, rius instead of rivus (Ital. Sp. rio), anticus instead of antiquus (Ital. antico). In the first century A.D. b became v between vowels, thus merging itself into the latter sound, so that in examining the Romance languages it is impossible to decide whether the original was v or b. And this change spreads in sentences to the initial b (as in the inscription manduca vibe lude e beni at me), which leads in some cases to some uncertainty in the use of v and b. And lastly, we have the case of cl and tl - e.g. veclus (Ital. vecchio, Fr. vieil, Sp. viejo, Port. velho, Rum. viechiu) instead of vetulus; the reduction of di before vowels, of j, g before e, i, and of z to a single sound j, or rather dj, in consequence of which we have diurnum (Ital. giorno, Fr. jour); juvenis (Ital. giovane, Fr. jeune); gener (Ital. genero, Fr. gendre); zelosus (Ital. geloso, Fr. jaloux), all represented by the same initial.

To turn to vowels, we must first notice that, according to Varro, ae was pronounced e in the country, but that in the cities the diphthong was maintained at first, while the simple sound was only admitted during the course of the 1st century A.D. If this is an instance of an early spreading of a rustic pronunciation, we have in another case a victory for that of the bourgeoisie and aristocracy. 0 for au belongs to Umbrian, Volscian and vulgar Latin, which explains why Appius Claudius Pulcher changed his name to Clodius when he deserted the patricians and went over to the plebeians. And there is other evidence of this change of sound. But in the inscriptions of the Empire o for au is very rare, save in proper names, and the Romance languages have partly preserved the au to this. day with little or no change (cf. Rum. auzi, Prov. auzir, Port. ouvir from audire), or only changed it to o at a later stage (cf. Fr. chose, where ch could only have arisen before a, not o), so that one may assume that the " Average Latin " always preserved the au. Then, without entering into detail, we must mention the prothesis of i before st, sp, sc, a phenomenon which arose, judging from the inscriptions, in the znd century A.D. We find it at the beginning of the sentence, and also within it after consonants, but not after vowels; e.g. ilia spata, but illas ispatas; istare, istd, but to istds, &c.

Most important of all are the modifications that affect the accented vowels, which give a new look to the language as a whole. In Old Latin and even towards the end of the Republican age, vowels varied solely according to their quantity, e.g. a was longer than a, e longer than e, but the vowel sound was the same, or at any rate the difference in quality between long and short must have been quite insignificant, seeing that Cicero and Quintilian wished the word divisio to be avoided in speech from motives of decorum, because of the likeness in sound to vissio. Quantity was not influenced by the number of the consonants following: actus was pronounced with a, factus with a, &c. In the course of the 1st century approximately quality was differentiated in addition to quantity in all vowels except a - short vowels being pronounced with an open, long ones with a close, sound. The written language expresses this change by writing ae for e, i for e, e for i, u for o, o for ic. In addition there are statements of the grammarians, though they mention only the double pronunciation of e and o, not that of i and u. It was probably in the course of the 4th century that the further change took place, by which all vowels were lengthened before a single consonant, and shortened before two or more, e.g. 'sitis became sitis, while tectum became rectum. But the older qualitative variations were maintained so that even now sitis and vitis, or tectum and lectum did not contain the same vowel-sound, the former having a cl o se, the latter an open, vowel. (Cf. Ital. sete, vite, Fr. soif, vis, Sp. sed, vid; or Ital. ktto and letto, Fr. toit and lit.) . It is at the end of the 4th century that Augustine says: " Afrae aures de corruptione vocalium vel productione non judicant," and the uncertain practice of the poets in the matter of quantity points to the breaking down of the old conditions. .This was not the end of the process of development; but the most important stages were already accomplished. In this, too, we are concerned with changes affecting the whole Romance region. The final step was taken when open i and close e, open u and close o, were reduced to one sound which may be called close e (or o). This step was not taken by the eastern regions, excepting as to e, and Sardinia remained completely unaffected (v. infra). The vowel-system that developed in course of time is thus as;follows: e ei i u uo o .1/?? ?

e e i 2t Q 9 ,a beforei const. -before i const.

"before 2 consts. -before 2 consts. 4/ U In the department of flexion we find less radical changes. The genitive was the first case to disappear. In general its functions were usurped by the preposition de. But for the possessive sense the dative was adopted, cf. HIC Requiescunt Membra Ad Duos Fratres, in an inscription from Gaul. The accusative serves for the case after prepositions under all circumstances, and therefore even in places where the older language used the ablative, e.g. magister cum suos discentes in a Pompeian inscription. Nouns of the third declension with monosyllabic nominative, e. g. lens, stirps, ars, &c., form a dissyllabic nominative, e.g. lentis, stirpis, &c. The dividing line between masculine and neuter, at all times doubtful, is frequently broken down, especially in the singular, e.g. Lubitum instead of cubitus, and there are converse cases. The absorption of the fourth declension by the second is almost complete. In the declension of the pronouns the genitives ipsuius, illuius, dat. ipsui, illui, fern. illaeius, illaei, are found in several inscriptions, but do not belong to the common language, since, as we have already said, they are not at home in the Iberian peninsula. On the other hand, all the Romance languages show that *eo took the place of ego. The use of ille as personal pronoun, and also of ipse, and of both these forms as articles, dates from ancient times. We find a parallel to the weakening of these demonstratives in the amalgamation of the pronominal combinations to be found as early as Plautus with ecce, eccum, which results in new forms, e.g. ecceille (0. Fr. cil, Mod. Fr. celui) or eccuille (Ital. quegli, Sp. aquel); ecceiste (Fr. ce- (t)); eccuiste (Ital. questo, Sp. aqueste). In the verb-system, a characteristic change is the disappearance of the future and passive forms, the explanation of the phenomenon in both cases being psychological rather than formal. Popular language is not familiar with the future, and replaces it by the present - or, more strictly speaking, the vulgar person deals only with the present or the past. The case of the passive is similar. The transposition of active into passive is too complicated a process for the simple mind. The object of the action remains the object; when the subject of the action is not known, they resorted to the indefinite third person plural, e.g. vendunt casam is the popular mode of expressing domus venditur. And further, the perfect amatus sum was replaced by amatus fui, since fui was a perfect and could now take over the function of a present. For the moment, all other tenses and moods of the verb were preserved, only of the infinite forms, the gerundive, perfect infinitive and the two supines disappeared. Of the gerund nothing remained but the ablative. In compensation, however, we soon find a form habeo cantatum springing up beside cantavi in use as perfect, e.g. litteras scriptas habeo meant in the first instance, " I possess written letters," with nothing implied as to who wrote the letters; `but later this usage is limited to cases where the owner is also ' the originator of the state of things expressed in the participle, and thus it attains to the force of a perfect.

There is little change in the formation of individual verb-forms. It is natural that the infinitives esse, velle, posse, being exceptional, should have been brought into line with all the rest. This was done by simply adding -re on to esse (Ital. essere, Fr. etre), while the other two were constructed from the forms of the verb whose ending was .accented, or from the perfect, e.g. volebam, potebam, volui, potui, gave rise to *volere, *potere, on the analogy of docebam, docui, monebam, monui, nocebam, nocui, &c.; with infinitives, docere, monere, nocere, &c. (cf. Ital. volere, potere, Fr. vouloir, pouvoir, Sp. and Port. poder, Rum. urea, putea). In other infinitives there is much confusion, especially as between -ere, and -ere verbs, noticed by the Latin grammarians themselves; we have evidence, too, that at an early stage the present forms in -io, -iam led to a confusion of the -ire and -ere conjugation, e.g. Plautus has morire (Ital. morire, Fr. mourir, `Sp. morir, Rum. muri); Lucretius has cupire; Cato has fodire, &c. For the rest we may note as important that perfect-forms without u-, such as -asti, -astis, -arunt, infected the first person singular, e.g. -ai instead of -avi. A new type in -idi arose on the model of vendidi, and then affected other verbs in -ndere, e.g, descendidi (in Gellius), prendidi (in the grammarian Probus) and in general verbs of the third conjugation. But its spread was slow, so that it can scarcely be said to have been common to all the languages.

In the formation of words the popular language probably had far greater freedom than the written language. We find not only a marked preference for diminutives in -ulus and -ellus, but many other types are established, or new ones created. And as the chief ones we must mention the post-verbalia (nouns constructed out of verbs). Thus pugnare, being itself derived from pugnum, then produces pugna (on the pattern of planta, plantare), and these formations soon became extremely common, and not only in a- verbs, but also in ere-verbs, cf. in particular dolus, " grief " (not to be confused with the ancient dolus, " craft "), C.I.L. x. 4510 (Rum. dor, Ital. duolo, Fr. deuil, Sp. duelo). As examples of other types we have -ura beside -or, which we can trace back to ardura, a contamination of ardor and arsura, which extended to fervura; also to strictura beside strictus; directura beside directus, when the old participles had separated both in form and in meaning from the verbal-system and had become adjectives, whose t was felt to be part of the stem. Another feature of the verb is the gradual retreat of old simple formations in favour of derivatives from the participle, e.g. cantare, adjutare, ausare, &c., in place of canere, adjuvare, audere; then for denominatives -icare and the Gr. -izare (Ital. -eggiare, Fr. -oyer, Sp. -ear) which, coming in with Christianity, was soon added on to Latin stems, e.g. (in Fulgentius) citherizantium aut tibizantium. Among points of syntax we may single out the replacing of infinitival sentences (following verbs of feeling, seeing, hearing, wishing) by clauses with ut, quod or quia, whence Ital. che, Fr. que. The latter particle spread most rapidly, and soon took precedence over the other conjunctions, not only in the cases just mentioned, but in introducing object-, subject-and final-clauses.

It is in the vocabulary that it is most difficult to define the relations of the common and the literary language. So much of the Latin vocabulary as appears over the whole Romance area comes of course from the everyday language which was used from the mouth of the Ebro to that of the Danube, but it is by no means all. It is more interesting to inquire whether anything can be reconstructed from Romance, and, if so, how much? The existence of a form aiutare, for example, mentioned above (Ital. ajutare, Fr. aider, Sp. ayudar, Rum, aiuta) and appearing in all the Romance languages, is indisputable. Between Fr. grolle (" crow "), Lyon. grola, Gascon. agraulo, Tirol. grolo, and (with change of gender) Apul. raulu, Rum. graur, the connexion, both in form and meaning, is so close that one is led to assume a common basis for all these words. This basis is *graulus, -a, and it is safe to assume that such a word goes back to Latin, though remembering that it was not found in the western regions. Rum. afld, Sic. asciari, Sp. hallar, Port. achar, Gris. aflcir, Dalm. aflucir, " to find," all point to afflare, and in this case, too, the change in meaning may be safely ascribed to Latin, only in this case Gaul is not included. Rum. aripei, Fr. aube, Prov. aubo, Sp. alabe," paddle-board," in Rum. meaning also " wing," and in Sp. also " the wickerwork on both sides of a vehicle," in Port. " the wing of a parapet," point to a form *alapa, which meant " wing " and which must have belonged to the vulgar language, even though no trace of it survives in Italy. Many other points could be enumerated, but problems are involved which have as yet hardly been taken up.1 In dealing with the division of this common language into a number of individual languages there are still further points of view to be considered. Before we can touch upon these, we must first take a general survey of these languages. There are altogether nine - Rumanian, Dalmatian, Sardinian, Italian, Raeto-Romanic, French, Provençal, Spanish and Portuguese. Of these nine languages, Dalmatian is now extinct, and even what we learn of it from the ancients is very meagre. On the one hand, Ragusa and the plains of Dalmatia never attained the degree of independence in literature which would have brought about a floruit in the language such as Provençal has to show. Neither, on the other hand, was its political independence stable enough, nor was it sufficiently remote to escape intercourse with the rest of the world, like the Raeto-Romanic dialects. The hordes of Sla y s pressing forward from the inner regions of the hinterland soon put an end to the Romanic civilization, first in the country and then it the towns. And when the Venetians, who were, both in point of culture and of commerce and of politics, on a higher level, regained their power over the Dalmatians by occasional conquests, chiefly over the cities, the result was of course all in favour of the Venetian dialect. On the island of Veglia alone there were still living about the middle of the 19th century a few people who still spoke Old Dalmatian. The last of these is now dead. Our approximate notions of this language are gleaned from the speech of these natives of Veglia, from a few more ancient notes, place-names, proper names and from the Romance elements in the Servo-Croatian dialect of Ragusa.' We may begin by reducing these nine languages to seven groups - Dacian, Dalmatian,. Sardinian, Italic, Raetic, Gallic and Iberian. The most striking peculiarity of the first three of these groups is the absence of Germanic words in the vocabulary. In other words, they were withdrawn from the influence of the general " Average-Latin " before the beginning of the more decided permeation of Latin by Germanic elements. There are other signs of their antiquity. In Central Sardinian c before e, i, and 1 Cf. G. Grober, Archiv. f. lat. Lexicographie, i. 204 ff.

Cf. M. G. Bartoli, " Das Dalmatische " (1906), (Schriften der Balkan-Kommission der K. Akademie der Wissenschaften, linguistische Abteilung, Bd. iv. and v.).

in Dalmatian c before e are always preserved as velars, and in south Sardinian and in Rumanian the palatalization is more recent, and secondary. The preservation of the tenues between vowels as breathed fortes is peculiar to Rumano-Dalmatian, but as north Sardinian used breathed lenes in their place, while the dialect of Nuoro, in Sardinia, preserved the fortes, we have every ground for assuming that central and south Sardinia also possessed either fortes or lenes in earlier times. Moreover, south Italy, Sicily and a large part of central Italy as far as the Apennines replace the old Latin tenues either with breathed fortes or breathed lenes, in marked contrast to the regions of the Po, to Gallic and the Iberian group. All these phenomena may perhaps be explained in conjunction with two historical events. By the abandonment of the province of Dacia (in A.D. 270), Rumanian lost its close touch with the languages of nearest affinity; and the division of the empire under Diocletian and Constantine necessarily entailed a linguistic division. At that epoch the linguistic conditions were roughly as follows: The principal changes in the vowel-system, especially the development of qualitative beside quantitative variations, had been accomplished, but there was still a difference between i and e, i and O. The old future had disappeared, and no tendency to produce a substitute had as yet appeared. The Latin pluperfect subjunctive still maintained its old usage, probably also the imperfect subjunctive and the future perfect. In declensions the type membrum, -a, had begun to spread; but corpus, -ora, was still in existence. Sardinia seems to have been, perhaps owing to its isolation, the first to have detached itself from this group. For it was not content with differentiating e and i, but it also retains -s, whereas the East-Rumanian and an Italian group suppressed -s, and in consequence also the difference between the nominative and the accusative singular. This and the levelling of neuters in -us and masculines in -u made it possible for the types membra and corpora to spread at the expense of the type loci, - a possibility of which South Italian and Rumanian made the fullest use.

On the given basis the various languages carried on their various developments, influenced partly by contiguity of other idioms, partly by causes unknown to us. Among neighbouring idioms, Greek had by right of its degree of civilization and its political power great influence in giving Rumanian and South Italian a similar direction, and that at a time when every trace of a geographical connexion between these two language-groups had long vanished. Thus, the replacing of the construction " I will come " by " I will that I come " took its rise in Greece and was passed on to Rumania and Apulia. The rise of the new future voiu c¢ntd, " I will sing," in Rumanian is probably due to Greek influence. In Latin itself both ille caballus and caballus ille are found, the position depending on the accentual conditions of the sentence. Then the loss of s made room for the form caball[u] ille with a victory for the inverted order. In Rumania alone this was the actual process, under the influence of the surrounding speech - Illyrian or Bulgarian, or perhaps independently of them, in this latter case serving as prototype to these languages. Dalmatian and South Italian, on the other hand, were so closely connected with the languages that preserved s and therefore prefixed the article that in this particular they separated from Rumanian. This is not the place to show how the Rumanian vocabulary and the structure of words was permeated markedly by elements from Slav, less markedly by elements from Turkish, Mod. Greek and Hungarian, which gave the language an alien appearance in point of vocabulary.

In its consonants, and, as far as one can judge, in its morphology, Dalmatian has preserved the stamp of antiquity. But in its vowelsystem there are marked changes, especially in the substitution of diphthongs for close vowels, e.g. changing a to e, u through the stage to oi, i to ei, o to au, e to ai. Diphthongs such as they appear also in Istrian and Abruzzian, so that we must presuppose some sort of connexion.

It may be that Sardinian took another course of development because (A.D. 458) the island was rent from Rome and incorporated in the African empire of Genseric, king of the Vandals. Therefore the sympathies of Sardinia were alienated from Italy, and turned on the one hand towards Africa (and unfortunately we have no information as to the " latinity " of this region), on the other towards the Iberian peninsula. These conditions lasted for a while, but later we find Genoa and Pisa fighting at intervals for supremacy in Sardinia, their organization being in many points identical with that of the island. On the whole, this new combination has not materially affected the language, especially in Logodoro. The vowel system (of great antiquity), as well as the velar pronunciation of c before e, i, remained unchanged, neither did they get as far as to adopt the future-forms current on the mainland; on the contrary, the Sardinians arrived independently and later at their usage of depo cantare or haia a cantar. But the use of ipse as an article in Sardinia, Mallorca, and in the earliest times also in the CatalanianGascon area, clearly proves the linguistic connexion which for a time covered this area, and we may also see some connexion in the fact that the lenes became voiced between vowels. On the whole, and in spite of everything, Sardinian is the most archaic of the Romance languages. Owing to its retaining s, it has failed to extend the membra-tempora types of formation, indeed it has almost rejected them entirely. It has retained the imperfect subjunctive to this day, and as a corollary it has lost the pluperfect of that mood. And though every Romance language has a number of Latin words that are not common to the rest, yet in this language the number of these a ra Xeyo A eva is greater than in others, and it is noteworthy that these have here survived such common expressions as domo, " house," mannu, " great," with other examples.

The East-Rumanian group (coupled with Sardinia) finds its counterpart in the great group based upon the Latinity of Gaul, the Iberian peninsula, and north Italy. This group contains a considerable number of fundamental peculiarities in phonology, morphology and vocabulary which prima facie lead us to assume a fairly long period of contact.

The chief of these peculiarities is the final change of the vowelsystem, i.e. the loss of the distinction between e and i, between o and ii; then the change of breathed plosives and fricatives between vowels into voiced plosives and fricatives respectively; the use of the pluperfect subjunctive instead of the lost imperfect subjunctive (Ital. cantasse, Fr. que je chantasse, Sp. cantase, Port. cantasse), the formation of a new future from the infinitive of the verb and the present, or (as the case may be) the imperfect or perfect of habere, e.g. Ital. cantero, canterei, Fr. je chanterai, chanterais, Sp. cantare, cantaria. If it is safe to assume that this latter formation had its origin in places where we find it most firmly rooted, we are led to assign it to the north of France. For it is only there that both elements in the formation are inseparably connected from the beginning of our record. In the old Provencal the two constituent parts are still separable; in the oldest Spanish and Portuguese their position is not fixed (i.e. the auxiliary may follow or precede the verb). In north Italy we frequently find the form avro cantare instead of cantaro, obviously because this formation is not properly acclimatized. But at any rate it is clear that the change of function from cantare habeo to cantabo belongs, to the time when the three great groups were still in close contact, and the evidence of the Latin texts falls into line with this view, showing this construction well established from the second half of the 4th century.' In the vocabulary we must note, among other things, the introduction of Germanic words, e.g. elmo, Fr. heaume, Sp. yelmo, " helmet "; harpa, " harp," Ital. arpa, Fr. harpe, Sp. and Port. arpa; medus,. " raced," which is found in Antimus and Isidore, but disappears later (cf. O. Fr. mies, " meed "); waidanian, Ital. guadagnare, Fr. gagner, Sp. guadanar, and many more.

The further steps in the process of differentiation were conditioned by the breaking up of the Roman empire by the great migrations. The establishment of the rule of the Franks in north Gaul, of the Visigoths in south Gaul and the Iberian peninsula, loosened old ties, created new nations and in consequence new and independent groups of languages.

The Iberian group was marked primarily by a striking simplicity in its flexions. The three-case system was given up at an early stage, even in prehistoric times, and has left no traces whatever. Owing to the preservation of -s the type membra was doomed to perish, and thus we find, from the beginning of our record and therefore presumably soon after the great cleavage took place, the prevalence in nouns of the following simple rule: sing. -e, -o, -a; plur. -es, -os, -as. The loss of the dative may have some connexion with the fact that the form illui for the 3rd personal pronoun had not yet established itself; and the desire for uniformity may have ousted the nominative of o- stems. There are analogies in the conjugation. The pluperfect indicative was preserved, and even (largely) with a Latin significance, but in the region of flexion much simplification took place, e.g. uniformity of accentuation in the three conjugations, marked reduction of the s- perfect and uperfect forms and a great reduction in the number of uparticiples.

The vocabulary is characterized by certain archaisms, and still more by the fact that a series of common ideas are rendered by new words limited in use to the Iberian peninsula. Thus we have querer (quaerere) instead of velle; quedar (quietare) instead of manere; callar (deriv. uncertain) for tacere; hablar (fabulare), " to speak "; llegar (plicare), " to arrive "; dejar (?) instead of lac are, &c. Further, we may mention the preference of tenere to habere even for the formation of perfect-forms, of which examples are to be found in Orosius, and of magic to plus for expressing comparisons, for which also we may find examples in Latin authors or the Iberian peninsula. The influence of the Goths or Suevi and Vandals on the vocabulary is inconsiderable, and when we trace it it is not easy to explain; e.g. Galician laverca, " lark," is clearly from a western Gothic *lawerka, but it is difficult to see why the name for this bird should have been supplied by the Germanic. To sum up, one may say that the Latin of Iberia was a self-contained language, at first showing little modification by influences from Iberian, or later by those from Germanic; further, that its development was slow, and that it aimed at simplicity.

At the present day there are three great groups, running almost 1 See Thielmann, in Archiv f. lat. Lexikogr. ii. 48 seq.

parallel from N.E. to S.W., e.g. Catalanian on the coast of the Mediterranean, akin to Provencal, Spanish in the centre, GalicianPortuguese on the Atlantic. From the historical point of view one part might be called Gothic-Romance, the other Suevo-Romance. But the national and linguistic history of the times and countries we are dealing with is still very obscure. The difference between the two idioms is chiefly one of phonetics, while in their morphology and vocabulary they do not greatly differ. Spanish may be described as a language which favours vowels at the expense of consonants, and which therefore shows, more than other Romance languages, a weakening even of initial consonants. It changes voiced stops first to fricatives, then to mere noises or " burrs " which finally disappear altogether, and s before a consonant or finally, becomes h (through a middle stage s) and is finally lost. The preferential treatment of vowels, however, entailed not a single change except that e was changed to the diphthong ie, o to ue; all else were preserved, e.g. diet (decem), tiempo (tempus), bueno (bonus), fuerte (fortis); but haver (habere), lid (lite), corona (corona), humo (fumus). The weakness of the initial sound is shown in enero (januaruis), hazer (facere), llamar (clamare with a transitional *clyamar), llaga (plaga), &c. The written language has no sign for voiced plosives between vowels, but -atho or -ao is spread over nearly the whole region.

In contrast to Spanish, Portuguese has a strong pronunciation of initial sounds, and so does not go beyond janeiro, fazer, and changes cl (with transitional form cly, ky), and also p1 (via ply, py) to ch, e.g. chamar, chaga. On the other hand, it has a careless articulation of vowels and consonants, and consequently no diphthongs. The unaccented vowels are weakened, as finals almost to vanishing point. It shows further a fusion of nasals with the preceding vowel, so as to form a nasal vowel, and this new nasality takes the colour of the preceding vowel, e.g. vina becomes vinho, but una becomes uma, otherwise before a vowel the nasal finally disappears; cheio and cheia, from plenus, plena. Similarly l was lost between vowels, e.g. ceo (caelum); before consonants it became I, or u, e.g. outro (alteru), caldo (calidu). Voiced plosives have a weak pronunciation between vowels, and these are sometimes made fricatives. In relation to the somewhat careless articulation we note a marked reaction on accented vowels by the final vowel (e.g. novo has a close vowel, nova an open one), and also by the following consonants: velarizes, s palatalizes preceding sounds, hence estas pronounced istas, " thou art," with reduced i, but devedor (debitor), " debtor," with reduced e. Lastly, the division between words is not sharp - the interaction of initial sounds and finals being very striking. Devedor has a plosive d, a devedor has a fricative; istas has a breathed -s, but istas nos ceus, " thou art in heaven," has a voiced -s; seja, " be," has a reduced a; o nome is pronounced u nome, but seja o nome is pronounced sej o nome, with an open o from a+o, &c.

The separation of Gaul took place likewise in the second half of the 5th century, when the Visigoths had settled down in the south, the Burgundians in the east, and the Franks in the north. The type of language that was evolved here is distinct from Spanish primarily and principally in the loss of final vowels except a, or, when the formation of the word was incompatible with this loss, in a weakening to e. On the other hand, the declension is strongly conservative. Nowhere are the old case-endings so clearly preserved as in this region, e.g. reis, " king," but la rei fille (regi filia), " the king's daughter "; veit le rei, " videt regem "; dunet le rei, " donat (dat) regi "; these are the modes of expression, and they last till far into the literary period. But at an early stage there was a breach between the Franks of the north and the Burgundians of the east on the one hand, and the Visigoths of the south on the other. For while the latter (the Visigoths) retained the old system of accented vowels, the former changed e to a diphthong ie, ó became uo, ue, and moreover e and i became ei, o and u became ou; a was changed to a, assuming that these vowels were long in accordance with the later Latin pronunciation, e.g.

Lat. debere nepote pede mola pratu North Fr. deveir nevout piet muele pret South Fr. dever nebot pe mola prat The northern group, moreover, weakened the consonants still further. D and g, secondary consonants from t and c, disappear like the primary ones, and thus pratellus becomes preau, S. Fr. pradel; advocatus becomes avoue, S. Fr. avogat; a secondary p (from b) becomes v, as we see by the form which replaces nepos 'above. If we are right in ascribing this to the effort to stress the accented vowel at the expense of the other constituents of the word, we may take this to be connected with the weakening of a where final, and between two accented syllables, e.g. N. Fr. aime from amat, as against S. Fr. ama; or in one case armeure (Mod. Fr. armure), in the other armadura, from armatura. Parallel to the preservation of -s on the one hand, and the close following of the old flexions on the other, we find the type membra preserved at first, though not spreading, whereas the tempora-type is abandoned. In the verb the variety in Latin perfect forms is still fairly well preserved, though there is a distinct extension of the u-perfect and the dedi-perfect. As we might expect, the vocabulary seems to be strongly coloured by Germanic elements of Frankish, Burgundian and Gothic origin.

The Raetic dialects, in their prehistoric phase, are less clear than others. Their contact, at an age nearing the Carolingian, with the French of the south-east in Valais seems to have caused a similar process of growth, especially as they change e and o into the diphthongs ei and ou, leaving at the same time the consonants more intact. At an early stage the inroads of the migrating nations cut off Raetia from the Po valley, and the pressure of the German tribes severed its union with the Romance-speaking nations of the west. Thus isolated it was free to follow its own course. This language also preserved at first the three cases and the type membra, the latter being developed later freely in use as a collective plural. But its further development was checked by the Lombards and Venetians.

But the most difficult problems are those that arise in Italy. Though one may say generally that the dialects of the region of the Po, and those of Liguria, belong to the types of north and western Romance, that is to say that the breathed plosives between vowels became voiced, yet they approach the typically Italian groups by their loss of -s. This means that when the whole Italian peninsula was separated from Gaul as well as from Iberia (after the close of the 5th century) and became again one homogeneous whole, the forms without s found their way into the north of Italy only slowly, so that s has remained in the west, i.e. in Piedmont, in monosyllabic words to this day, e.g. as, " thou hast," ses, " thou art "; the same rule prevailed in older times in the east, in Venice, and there the s was also preserved (in questions) in polysyllabic words, e.g. venis-tu, " comest thou? "; and the old form maintained itself in Milanese in the single form sistu, " art thou ? " To the loss of s we trace the extinction of declensions, but as its action began to take effect later, the membra-type gained little footing, the tempora-type none at all. In the vocabulary the Lombard elements are numerous, extending, like the supremacy of the Lombards, over the whole peninsula. It may be that s was lost under the influence of central Italy acting on the north. If so, we may surmise that a similar influence has changed cl, p1, and fl to chi, pi, fi (chiamare, pianta, fiamma). For it is precisely this point that differentiates both the Raetic dialects and Provencal from the contiguous Italian dialects, and the change certainly took place only after the latter were completely detached. On the other hand the Italian vocabulary has been strongly influenced by the north, especially in Tuscany.

The rise and development of the Romance languages, in its large outline, appeals to the imagination as a vast historical phenomenon closely bound up with the fate of nations. One other element must not be overlooked on which we have touched more than once in the above sketch, for it bears so directly on the Romance vocabulary as to deserve the tribute of a general survey: this is the Germanic.

When mercenaries of Germanic origin pervaded the Roman armies, Germanic words found their way first into the language of the camp, and thence into the vulgar language generally. And at that stage perhaps many words may actually have been imported which were, partly at any rate, lost again later. Roman and Greek authors admit a considerable number of Germanic words, including terms belonging to warfare, e.g. bandum, " standard," used by Procopius, which still continues in the form of O. Fr. ban, Ital. bandiera, Sp. bandera, &c. Brutis, " bride," " daughter-in-law," which occurs frequently in inscriptions, may date from the period of camp life, but for the rest it is retained only in Fr. bru, and in Friuli and Dalmatia. On the other hand, companio is clearly a Latinization of Gothic ga-hlaifa, the meaning of which carries us back to the same sphere. Other old words express ideas of culture, or names of animals which the Romans learned to know in the German-speaking north, e.g. ganta, " wild goose " (in Pliny), O. Fr. gante, Prov. ganta; or taxo, "badger," Ital. tassone, Fr. taisson, Sp. tejon. But the impression made was not pronounced until the age of the Germanic invasions, and then we find a great variety in the various Romance countries. In Italy we have two invasions to consider - by the Goths, and by the Lombards. But the destruction of the rule of the Lombards by Charlemagne, and the introduction of Frankish elements consequent upon it, should not be considered under the same head, since these Franks may themselves have been a Romance-speaking tribe. Goths as well as Lombards have left a trail as noticeable in the language as elsewhere. Thus we find in several instances some uncertainty as between b and p as an initial sound in Italian words borrowed from Germanic, e.g. banca and panca, balla and palla, the forms with b being Gothic, those with p Lombardic. Or again recare, " to bring up," goes back to Gothic rikan, " heap up," " collect "; ricco, " rich," to Lomb. rihhi, &c. Whereas the vocabulary shows impartially an impress of both nationalities, the Lombards have left their stamp unmistakably on the proper names. Speaking generally, Italy as well as the other Romance countries follows the rule that medieval names of persons are either " Christian " (in the strict sense) and therefore of Hebrew or Graeco-Roman origin, or on the other hand Germanic. Roman names that are not also Christian seem to have survived only in south Italy in any great number, while on the contrary the Germanic are not represented at all in Dalmatia. One of the characteristics of Gothic is the change of e to i, so that it has names ending in -mir. Of these we find no trace whatever in Italy, on the contrary we find Gundimar, Ildimar, &c. Then we have abbreviated forms in izzo, e.g. Gaudizzo, Albizzo, &c., which are distinctly Lombardic; but not Gothic ones in -ila. There is no parallel to all this in the Iberian peninsula. As we have already said, the Gothic contribution to the vocabulary is very slight. But on the other hand in the th century the great majority of proper names is Gothic, e.g. Alfonsus (Hadufonsus), Gundomirus, Recimirus, &c.; or Recila, Fafila, or Elvira, O. Port. Gelvira, Goth. *Gailavira, and scores of others, all proving the great influence of Gothic.

And lastly, France possesses the largest number of Germanic elements in its vocabulary, Gothic in the south, Frankish in the north (though it is often impossible to ascertain to which class they belong). But beside these there are many Old High German words, and again Anglo-Saxon and northern ones, more particularly those connected with shipping and the sea. These Germanic elements cover nearly all branches of human activity. Thus bat, Fr. bettir, " to build," from *bastyan, " to bind together with bast," " to plait "; hourder," to cover with boards," from hurdi," hurdle "; mason, " the mason," in Isidore makjo (Frankish rather than Gothic) refer to house-building; guacher from waskyan, broder from *brusdan, point to the occupations of women, and danser from dinsan and O. Fr. treschier, " to dance," from treskan, " to thresh," to their amusements. Women's work is probably denoted further in rouir, rotjan, and E. Frank. naisier, natjan, " to net "; the same remark applies to the dyeing of cloths (Fr. touaille, Engl. " towel," from thwahila), and ribbons (bande from binda) with guede, " woad," and other colouring matters, whence we have, e.g., brun, bleu, blond, blanc. But while the vocabulary has had its accessions drawn from various races, the proper names show the same rules as in Italian, i.e. Frankish gains the sole supremacy. We find, it must be admitted, some Gothic names in -mir in the south early in the middle ages, but they were not maintained as late as the Romance period, such was the influence of the victorious northern race.

Even after political and literary independence had enabled the individual Romance languages to grow as separate units on their own basis, they retained their interconnexion and were open to mutual influence. But this influence is only partial, i.e. it affects nothing but the vocabulary, and has a certain relation to various tendencies in the developments of civilization. And under this head the most important point is the really enormous influence which France (both south and north) has exercised on all the Romance countries, just as she has on the Germanic - an influence which has hitherto not been duly recognized. The first traces go back to the invasions of Charlemagne already mentioned. To instance only one, we have schiavino, " justice, alderman," which cannot be derived directly from the Germanic, as is shown by the v. The second important period is the age of chivalry and the literary tendencies centring round it. A word like budriere, " baldric," is derived from Fr. baudrier, not directly from Germanic Balderich; Ital. banda goes back to O. Fr. bande, and this again to binda; Ital. giallo is not from galbinus but from O. Fr. jalne (Mod. Fr. jaune), derived from that word, &c. But it seems that in one of the prehistoric periods the Tuscan vocabulary was strongly affected by that of the Gallo-Romanic. Whereas in the Iberian peninsula, in Sardinia, in south Italy, Rumania and Rhaetia dies survives, in O. Fr. di has been almost completely ousted by jour, but in Tuscan and the Italian literary language we find giorno and di side by side. Thus trouver, Prov. trobar, spreading from France into Italy, drove the old aflare more and more back towards the south. The most recent layer was introduced during the reign of the house of Anjou chiefly in south Italy and Sicily, and kept its hold to the present day in spite of the Sicilian Vespers, e.g. Sic. vuccieri, " butcher," from Fr. boucher. The Iberian peninsula can likewise bear witness as to French influence, e.g. O. Sp. fonta, " shame," is not from Goth, *haunitha, but from Fr. honte; O. Port. saluer not from Lat. salutare, but O. Fr. saluer. On the whole, Portuguese seems to possess more of these Gallicisms than Spanish, history supplying a simple explanation.

Italy too yielded its contributions, especially in the 15th and 16th centuries, many military terms (noble and ignoble), e.g. French carogne and canaille; poignard, " dagger," from Ital. pugnale, instead of O. Fr. poigniel; but also panache, " plume," from pennacchio, and many others that have become common property. But the influence of the Iberian peninsula on the contrary was not so strong as to be more than sporadic; the Sicilian and Neapolitan vocabularies alone are more closely akin to Spanish, and this is easily explained on the ground of their political and commercial relations.

As to the Romance languages beyond Europe we have but little to say. There is a distinction to be made between Creole and genuine Romance. Belonging to the latter we have the French of Canada, the Spanish of Central and South America, the Portuguese of the Brazils. Speaking generally we may say that the particular languages retained the form of the language in the 16th and 17th centuries, that is to say that of the time of the immigration, and that they developed along the lines already established.

Thus in Mexican Spanish the loss of d, g, between vowels, of s before consonants and as a final, has been carried further than in the mother-country. There are no proved traces of any noticeable influence from the languages of the natives.

Literature. - The real founder of scientific Romance philology and linguistics is Friedrich Diez, in his Grammatik der romanischen Sprachen (3 vols., Bonn, 1836-42), and Etymologisches Worterbuch der romanischen Sprachen (2 vols., 1852). All questions concerning Romance philology and the historic grammar of the different Romance languages are treated in G. Grober's Grundriss der romanischen Philologie (2nd ed., Strassburg, 1906), and in W. Meyer-Liibke's Grammatik der romanischen Sprachen (4 vols., Leipzig, 1890-1900); Einfiihrung in die romanische Sprachwissenschaft (2nd ed., Heidelberg, 1909). The principal magazines devoted to the subject are Zeitschrift fiir romanische Philologie (ed. Grober; since 1877); Zeitschrift fiir neufranzosische Sprache and Literatur (ed. Behrens; since 1879); Romanische Forschungen (ed. Vollmoller; since 1885); Archiv fiir das Studium der neueren Sprachen (since 1846); Romania (ed. G. Paris and P. Meyer; since 1812); Archivio glottologico italiano (ed. G. I. Ascoli; since 1873). The great development of Romanic philology after Diez is due principally to A. Tobler, G. Grober, W. Forster and H. Suchier in Germany; A. Mussafia (d. 1905), H. Schuchardt in Austria; G. Paris (d. 1905), P. Meyer in France; G. I. Ascoli (d. 1907), and F. d'Ovidio in Italy.

(W. M.-L.)