l 6 t*-e>-<<^- N\V\ w .' 1 \ \ o t\ [Reprinted from the Publications of the Modern Language Association of America , xxi, 3] THE HISTORY OF AI AND El IN FRENCH BEFORE THE DENTAL, LABIAL, AND PALATAL NASALS. In the following pages the history of the pronunciation in French of accented ai and ei before n, m and n is to be investigated in detail. The subject naturally falls into two divisions. On the one hand we have ai and ei before n or m , and on the other we have the words in which these diphthongs are followed by n. Both divisions are closely allied and the development of the one is often identical with that of the other. Yet for the better control of the material it will be advisable to separate the history of ain or aim and ein or eim from that of aign and eign. The various grammars differ in the outline of the history of the sounds in question. Concentrating our attention first upon ai and ei followed by the dental nasal, and granting that the most direct road from the oldest stage ain to modern § is through ein, what would seem to be the simplest explanation may be found in Nyrop’s Grammaire historique de la langue frangaise , I, §§ 217 and 222. According to him, ain in the 11th century was pronounced din , and ein was ein. ju the 12th century the two sounds coincide with the value cf gwi, and remain so until about the 16th century, when the modem pronunciation f develops. When both syllables had the same value, one could be written for the other, and hence the well known Old French confusion in the orthography. This vie\W>f the question is shared by Meyer-Lubke, Grammatik derl^XLnischen Spraehen, I, § 89. There is, however, evidencS^B^the history of these syllables was not so simple. manuscripts present an orthography quite the value 17 638 JOHN E. MATZKE. of ein for graphic ain , and rimes can be cited which show that din remained stationary while ein became din. In consequence Behrens in the Schwan-Behrens Altfranzosische Grammatik, §§ 257 and 258, admits the history outlined by Nyrop only for the literary language, while dialectically and in a region which he describes as lacking accurate delinea¬ tion he accepts din as the common value of both ain and ein. This pronunciation of din , which Behrens believes was dia¬ lectic, is looked upon by Suchier, Altfranzosische Grammatik , § 45, as the regular sound of these syllables in the literary language; ‘ ei ist zu ai geworden etwa in der Mitte des xii Jahrhunderts. Seit dem werden ein und am promiscue geschrieben, und es ist ganz gleichgiltig, ob ein Schreiber jenes oder dieses bevorzugt.’ It is evident that the question is sufficiently encumbered to merit a detailed investigation. The arguments available must be sought in rimes and to a less degree in the orthog¬ raphy. For this purpose a long and representative list of texts 1 has been studied, extending in chronological order 1 1 add here a rough chronological list of the texts that have been most directly utilized for the present study and I include one or two titles that are not cited in the discussion, so that others interested in the same prob¬ lem may be spared the trouble of searching through the same texts again. If the arguments presented here should not be found convincing, new evi¬ dence will have to be sought in different sources. Karls des Grossen Reise nach Jerusalem und Constantinopel, nrsg. von Kosch- witz, Heilbronn, 1883 (Altfrz. Bibl., n) ( Voy. Chari.). Li Cumpoz Philipe de Thaun, hrsg. von E. Mall, Strassburg, 1873. Le Bestiaire de Philippe de Thatin, Texte critique par E. Walberg, Lund, 1900. Les Voyages Merveilleux de Saint Brandan, Legende .... publi^e par Fr. Michel, Paris, 1878, and by Sufi^ier, Rom. Stud., i, p.o. 567-587. ( Brandan.) Le Couronnement de Louis, puJ^^^K E. Langlois, Paris, 1888. ( Soc. d. Anc. Textes.) Le Roman de ThebesJid Constans, Paris, 1890. ( Soc. d. Anc. Textes.) AI AND El IN FRENCH BEFORE NASALS. 639 from the beginning to the 16th century and arranged in groups according to the dialects. I now present this Le Roman de Tristan par Beroul, public par E. Muret, Paris, 1903. (Soc. d. Anc. Textes.) Eneas , public par J. S. de Grave, Halle, 1891. ( Bibl. Norm., iv.) Maistre Wace’s Roman de Ron et des Dues de Normandie, hrsg. von H. Andresen, Heilbronn, 1877-1879. (R. Ron.) Benoit de Sainte-More et le Reman de Troie par A. Jolv, Paris, 1871. ( R. Troie.) Le Roman de Troie par Benoit de Sainte-Maure, public par L. Constans, vol. I, Paris, 1904. (Soc. d. Anc. Textes.) Chronique des Dues de Normandie par Benoit, publi^e par Fr. Michel, Paris, 1836-1844. (Coll. d. Doc. Ined. s. VHist. d. France.) Adgar’s Marienlegenden, hrsg. von C. Neuhaus, Heilbronn, 1886. (Altfrz. Bibl., ix.) Aiol et Mirabel und Elie de Saint Gille ; Zwei altfranzosische Heldenge- dichte, hrsg, von W. Forster, Heilbronn, 1876-1882. Amis et Amiles und Jourdains de Blaivies ; Zwei altfranzosische Heldenge- dichte, hrsg. von C. Hofmann, Erlangen, 1852. Aucassin und Nicolete, hrsg. von H. Suchier, Paderborn, 1899. De Saint Laurent, poeme anglonormand du xn e siecle, publie par W. Soderhjelm, Paris, 1888. Sainte Catherine —Dve Verse starofrancouzsk6 Legendy o Sv. Katerine vydal U. Jarnlk, Prague, 1894. La Vie de Saint Grilles par Guillaume de Berneville, publi^e par G. Paris et A. Bos, Paris, 1881. (Soc. d. Anc. Textes.) Ille et Galeron von Walter von Arras, hrsg. von W. Forster, Halle, 1891. (Rom. Bibl., vii. ) Christian von Troyes, Siimtliche Werke, hrsg. von W. Forster, Halle, 1884- 1899. Le Roman de Tristan par Thomas, public par J. B4dier, Paris, 1901. (Soc. d. Anc. Textes.) Les Chansons de Gace Brule, publics par G. Huet, Paris, 1902. (Soc. d. Anc. Textes.) Die Lais der Marie de France, hrsg. von K. Warnke, Halle, 1900. (Bibl. Norm., m.) Die Fabeln der Marie de France, hrsg. von K. Warnke, Halle, 1898. (Bibl. Norm., vi.) Der Roman du Mont Saint-Michel von Guillaume de S. Paier, hrsg. von P. Kedlich, Marburg, 1894. (Ausg. u. Abh., xcn.) Estienne von Fougiere's Livre des Manures, hrsg. von J. Kremer, Marburg, 1887. (Ausg. u. Abh., xxxix.) 640 JOHN E. MATZKE. material, which, though not exhaustive, is yet sufficiently complete to warrant the drawing of definite conclusions; and I hope that it may serve to throw light upon one of the La Vie de Saint Thomas le Martyr par Gamier de PontrSainte-Maxence, pub¬ liee par C. Hippeau, Paris, 1889. The Metrical Chronicle of Jordan Fantosme, edited by R. Howlett, London, 1886. Hue de Rotelande’ s Ipomedon, brsg. von E. Kolbing und E. Koschwitz, Breslau, 1889. Estoire de la Guerre Sainte par Ambroise, publiee par G. Paris, Paris, 1897. ( Coll. d. Doc. Ined. s. V Hist. d. France.) L’Escoufle, publie par H. Michelant et P. Meyer, Paris, 1894. ( Soc. d. Anc. Textes.) Robert le Diable, publie par E. Loseth, Paris, 1903. (Soc. d. Anc. Textes.) Le Roman de la Rose ou de Guillaume de Dole, public par G. Servois, Paris, 1893. (Soc. d. Anc. Textes.) Char dry’s Josaphaz, Set Dormanz und Petit Piet, krsg. von J. Koch, Heil- bronn, 1879. (Altfrz. Bibl., i.) The Song of Dermot and the Earl, published by G. H. Orpen, Oxford, 1892. . Li Romans de Carite et de Miserere du Renclus de Moiliens, publie par A.-G. van Hamel, Paris, 1885. ( Bibl. d. V Ec. d. Hautes Etudes.) Maistre Elie’s Ueberarbeitung der dltesten franzbsischen Uebertragung von Ovid’s Ars Amatoria, hrsg. von Kuhne und Stengel, Marburg, 1886. (Ausg. u. Abh., xlvii.) La Clef d’Amors, hrsg. von A. Doutrepont, Halle, 1890. (Bibl. Norm., v.) Li Chevaliers as deus Espees, hrsg. von W. Forster, Halle, 1877. Li Romans de Durmart le Galois, hrsg. von E. Stengel fur den litterarischen Verein in Stuttgart, 1873. Le Roman de Renart, hrsg. von E. Martin, Strassburg, 1881-1887. Guillaume de Palerne, public par H. Michelant, Paris, 1876. (Soc. d. Anc. Textes.) Le Bestiaire . ... des Guillaume le Clerc, hrsg. von R. Reinsch, Leipzig, 1892. (Altfrz. Bibl, xiv.) Le Besant de Dieu von Guillaume le Clerc de Normandie, hrsg. von E. Martin, Halle, 1869. Raoul de Houdenc, Le Songe d’ Enfer, Le Songe de Paradis, Li Romans des Eles, publies par A. Scheler, Trouveres Beiges, Louvain, 1879, vol. ii. La Vie de Saint Gregoire par Frere Angier, publiee par P. Meyer, Romania, xii, pp. 145-208. Trois Versions rimees de V Evangile de Nicodeme , publiees par G. Paris et A. Bos, Paris, 1885. (Soc. d. Anc. Textes.) AI AND El IN FRENCH BEFORE NASALS. 641 very vexing problems of French Historical Grammar. We shall consider in the first place the history of ai and ei before n or m. U Hisloire d# Guillaume le Marechal, publide par P. Meyer, 3 vols,, Paris, 1891-1901. ( Soc. d. VHist. de France), vol. I. La Bible Guiot de Provins in Fabliaux et Contes des Pontes Frangais, publics par Barbazon et Meon, Paris, 1808, vol. n, pp. 307-393. Le Pit des Hues de Paris, ibid. , pp. 238-275. Les Cneries de Paris, ibid. , pp. 276-286. Les Moustiers de Paris, ibid., pp. 287-292. Le Roman de Galerent, publie par A. Boucherie, Paris, 1888. (Soc. pour V etude des lang. rom.) TFistasse le Moine, hrsg. von W. Forster und J. Trost, Halle, 1891. (Rom. Bibl., iv.) Lyoner Yzopet, hrsg. von W. Forster, Heilbronn, 1882. (Altfrz. Bibl., v.) Le Roman de la Rose, public par Fr. Michel, Paris, 1872. Floris et Liriope des Robert de Blois, hrsg. von W. von Zingerle, Leipzig, 1891. (Altfrz. Bibl., xii. ) Jean Bodel, Le Jeu de Saint Nicolas. Adam de la Halle, Le Jeu de la Feuillie, both published by Monmerqud et Michel, Thedtre Frangais au moyen dge, Paris, 1885. Adam de la Halle, Le Jeu de Robin et Marion in Bartsch-Horning, La Langue et la Litterature Frangaise, Paris, 1887, col. 523-548. Trouveres Beiges, publics par A. Scheler, Louvain, 1866-1879. Rutebeuf, Euvres Completes, publics par A. Jubinal, Paris, 1874-1875. (Bibl. Elzevirienne .) Richars li Biaus, hrsg. von W. Forster, Wien, 1874. Les (Euvres poetiques de Philippe de Remi, sire de Beaumanoir, publiees par H. Suchier, Paris, 1884-1885. (Soc. d. Anc. Textes.) La Manekine, vol. i. Li Dis dou Vrai Aniel, hrsg. von A. Tobler, Leipzig, 1884. Octavian, hrsg. von K. Vollmoller, Heilbronn, 1883. (Altfrz. Bibl., ill.) (Euwes Completes d'Eustache Deschamps, publiees par Queuxde Saint-Hilaire et G. Raynaud, Paris, 1878-1903. (Soc. d. Anc. Textes ), vol. i and ii. Miracles de Nostre Dame par personnages, publics par G. Paris et U. Robert, Paris, 1876-1883. (Soc. d. Anc. Textes), vols. I, ii and m. (Euvres poetiques de Christine de Pisan, publiees par M. Roy, Paris, 1886- 1896. (Soc. d. Anc. Textes.) Meliador par Froissart, public par A. Longnon, Paris, 1895-1899. (Soc. d. Anc. Textes.) Poesies Completes de Charles J Orleans, publiees par Ch. d’H^ricault, Paris, 1896. 642 JOHN E. MATZKE. I. AIN—EIN. The two syllables ain and ein are not found forming assonance or riming together in the Alexis, the Roland, the Reimpredigt, the French translation of Marbod’s Lapidary, the Comput and Bestiaire of Philippe de Thaon. They are kept distinct even as late as Guillaume de Berneville’s Vie de Saint Grilles. 1 In these texts the pronunciation was din and ein respectively; cp. mains, Rol. 3965, in assonance with -an, ceinte, ibid. 984, in ei assonance, and peine 1787, aleine 1789, feindre 1792 in assonance with sanglente, temples, entendent, etc.; cp. also Engelmann, TJeber die Entstehung der Nasalvokale, Halle, 1882, pp. 20 ff. The earliest 2 definite evidence of a confusion of the two syllables in rime occurs in the Brandan . 3 Here ain and ein CEuvres Poetiques de Guillaume Alexis, prieur de Bucy, publiees par A. Piaget et E. Picot, Paris, 1896-1899. ( Soc. d. Anc. Textes.) 1! Arnant rendu Cordelier, poeme attribue a Martial d? Auvergne, public par A. de Montaiglon, Paris, 1881. (Soc. d. Anc . Textes.) Le Mistere du Viel Testament, publie par le baron James de Rothschild, Paris, 1878-1891. (Soc. d. Anc. Textes.) Vol. i. Die WerJce Maistre Frangois Villons, hrsg. von W. von Wurzbach, Roman- ische Forscliungen, xvi, pp. 405-584. (Euvres Completes de Clement Marot, publiees par B. Saint-Marc, Paris, Gamier, without date. 1 Cp. G. Paris, ed. p. xxvii. 2 In his Altfrz. Gram., p. 72, Suchier cites as earliest instance of the mingling of ain and ein, desteint: refraint, Bestiaire 2865. However, this rime would be so unique for Philippe de Thaon, that Walberg in his edition of the text, pp. xlviii and 146, rejects the reading and adopts restreinte: desteinte instead. 3 To this categorical statement the following note must be added. The Voyage de Charlemagne in its assonances shows the same pronunciation of the syllables in question as the Roland, cp. Koschwitz, Rom. Stud., n, pp. 38 ff. At the same time this poem contains one laisse, 11. 783-802, which AI AND El IN FRENCH BEFORE NASALS. 643 rime freely; cp. quarenteine : semaine 133, cha [e ]ines : semaines 866, funtaines : pleines 998, 1586, meindres : greindres 1004, and from thi^ poem on, mixture of the two is constant in Old French texts. Suchier suggests, Altfrz. Gram., p. 72, that the process was somewhat slower on the continent than it was in England. Wace in the Roman de Ron has 48 pure ain and 14 pure ein rimes, and only one instance of fusion, Saint Oain: secrestain (andcenum : -anum), 1. 347. However Pohl, Rom. Forsch., ii, pp. 581-2, has shown that while this observation is undoubtedly exact, it is very probably true that the inference that Wace con¬ sciously separates the two syllables is not justified, for in the same author’s Brut mixture of the two is much more frequent, and, at any rate, in other continental texts of the same period, such as Eneas, the poems of Marie de France, and Benoit de Sainte-More, the Livre des Manures, fusion of the two syllables is quite the rule. In view of these facts it is not at all surprising that in the prose texts of this appears to contradict the accuracy of this assertion. Here the following sequence of assonances is found : Charlemaignes: compaignes : deplaindre: France: regne: grande: enfraindre : remaignet: Charlemaigne: plaines: pleines: descendre : ente : aimet: regne : France : Charlemaignes : plaigne. It will be impossible to avoid the conclusion that for this poem en and an form a correct assonance (cp. Rom. Stud., ii, p. 49), though the mixture of the two is not very frequent. That regne should be found in the same laisse need not surprise. It had become rane in pronunciation, and there are numerous other texts giving evidence of a similar pronunciation of the word. However, difficulty is created by the appearance of pleines, 1. 793, in the series. Since the separation of ain and ein seems so clearly demanded by the assonances of this text, Koscliwitz, l. c., p. 40, suggests that the lines 783-802 should be divided into three laisses as follows : (1) 783-792, di:a, (2) 793-795, ei: e, (3) 796-802, ai:a. Considering the probable age of the poem and its dialect, this explanation is presumably correct, but whatever the final decision may be, the consideration of the present problem need not concern itself with these lines. Everything depends upon the age of the poem, and if it should be younger than the Roland, these assonances would only corroborate what we can observe elsewhere. 644 JOHN E. MATZKE. period ain and ein are constantly confused in orthography; cp. for the MonUbourg Psalter, Harseim, Pom. Stud., iv, pp. 277 and 283 ; for the Cambridge Psalter, Schumann, Vokalismus und Konsonantismus des Cambridger Psalters, Franz. Stud., iv, fasc. 4, pp. 17 and 29, and for the Quatre Livres des Pois, Schlosser, Pie Lautverhaltnisse der Quatre Pivres des Pois, Leipzig, 1887, pp. 13 and 31. The next point to be discussed is the pronunciation of this syllable written indifferently ain or ein. To facilitate the control of the available material, we shall divide the examples according to the letter or letters which may follow after the nasal consonant. 1). ain — ein. The union of these two syllables in rime leaves no question that their pronunciation was identical, but it gives no clue as to the nature of the vowel or diphthong that was pronounced. Inasmuch as this was either din or ein we may look for imperfect rimes with an or en as capable of throwing light on the problem. A few examples of ain : an occur in the texts which I have examined, and we may add those of ains : ans as having the same value. Cp. Brandan: pan (pannum), Brandan 480, Brandan: an, ibid. 824, Brandan : vilain, ibid. 163, Brandan : main, ibid. ‘658, Brandans : mains, ibid. 203, pan (panem) : ahan, Mist. Adam 434, but pain : Evain, ibid. 786, Trajan : roman, Angier, Gregoire 2539, but Traien : paien, ibid. 2715, der- vans : Johans, Angier, Dialogues, 72, r°b, cited by Miss Pope, Langue de Frere Angier, Paris, 1903, p. 12, esturman : certan, Eneas 205. For the correct appreciation of these rimes it should be noted that of the words in question, vilain, main, pan, roman, derrans and certan, involve the Latin vowel a, and that Brandan is a proper name in which the tonic syllable may have retained its Latin value. 1 When 1 Cp. Suchier, Zs. f. rom. Phil, ix, p. 89, note. AI AND El IN FRENCH BEFORE NASALS. 645 we take into account, furthermore, the fact that the texts from which these few rimes are taken cover a period of a hundred years, and that ains , which certainly had the same value of vowel as ain, occurs in the same texts and in others of the same period and dialect in rime with ens and iens (cp. below, p. 646), it becomes evident that these examples represent rime licences. They are either pure Latinisms 1 or remnants of the earlier practice exemplified by the asso¬ nances cited above from the Roland , but they have no argumentative value for the determination of the pronuncia¬ tion of ain in this dialect and during this period. Rimes of ain or ein with en on the other hand are even rarer. It must be borne in mind that gn in the final sylla¬ ble of the word exists in Old French only in a few learned names and in words of the categories of bene and paganum. Of the former, as far as I know, only a single instance occurs in the texts examined, Jerusalem : Alein, B. Chron. 36894. When compared with Alain: main, ibid. 36940, main :frein 16366 and Jerusalem : huem 31752, it becomes evident that both ain and ein in this text cannot have been ain and this is also the conclusion of Pohl, Rom. Forsch., ii, p. 554. Mixture with ien points in the same direction. The actual rimes that can be cited here are not very numerous, because the inflected forms of the syllables in question are more frequent in rime than the uninflected. Their discussion may, therefore, be deferred to the succeeding paragraph. 2). ains — eins, ainz — einz. Several rimes show mixture with ens or enz; cp. defens: mains (minus), Mist. Adam 148, meins : tens, .Gaimar 1811, ateinz : defenz, B. Chron. 22848, genz : seinz, ibid. 32235. Since checked e before a nasal in these texts was certainly g 1 Cp. Miss Pope, l. c., p. 12* 646 JOHN E. MATZKE. it will be difficult to avoid the conclusion that ains and eins tended in the same direction. This inference is emphasized by the existence in the same and similar texts of a certain number of rimes just referred to, beginning with the R. Troie, which show fusion of ains and eins with iens. In the Eneas we can observe only mixture of iens and ens; cp. Troiiens : tens (tempus) 565, 601, 5811, 6319, anciens :tens 4127, Sabiens : cuens 3949, euens : Volcens 5093, and the same is true of Marie de France; cp. anciens : tens, Milun 63. The JR. Ron contains no examples in point, but the R. Troie and B. Chron. continue the tradition of the JEJneas, while they add besides rimes of ainfs) and ein(s) with ien(s). Since the latter was ign(s), as shown through the union with ens — gns, there can be little question about the pronuncia¬ tion of the former. The following are the examples :— ien(s) : en(s) — R. Troie. 1 Troien : sen 5813, 6815, nequeden : Troien 1.8641, Troiens : tens 581, 7179, 20471 , : porpens 19915, Atheniens : tens 8489, Sisiliens : buens 18581, Poflago- niens : suens 20515; B. Chron. suen :bien, 1-1765 (cp. suens: buens , n-3005), bo ens : cristiens, II-24307, tens : Egiptiens, 1-413, paiens : sens, II— 23081, tens : crestiens, n-39017. ien(s) : ein(s )— R. Troie. meins : Troiains 5275 ; B. Chron. Men : Saint Oien, n-19550 (cp. Sainz Oieins : mains (manus), n-25840, main (mane) : Oain, n-7009, 19354). ien(s ): ain(s) — R. Troie. primer ains : Troiains 13903,18745, 25273, ger mains: Indiains lAOdl, Frisains: primer ains 15549 ; B. Chron . cristiens : premerains, 1—925, : parr eins, ii—6577, 7988, Men : Saint Oien 19550 (cp. Saint Oain : main, ii- 19354). Of the same general nature are the rimes Swein : buen B. Chron. 31046, Sueins : buens 38889 and seins (= suens) 1 References to the R. Troie as far as 1. 8292 are given according to the new edition of the poem by Constans, Soc. d.. Anc. Textes, Paris, 1904. AI AND El IN FRENCH BEFORE NASALS. 647 : Aldus 31008 (cp. Alains : seins (sanus), R. Rou 2735), interesting particularly for the orthography, for Suein is the usual bisyllabic Soen , Suen, 1 and seins stands for suens. The Chron. ML S. Mich, presents only a single rime andeins : pens (penso) 3752, interesting also for the orthography, while the Livre d. Man. has no rime in point. But the Beroul Tristan contains Brengain : bien 523, : mien 553, Lan [ci] en :Ivein 1155, TJrien : Dinoalain 3487, suen : Denoa- len 4435. Here Brengain may stand for Brangien and can, therefore, not enter into the argument, except in as much as it shows the value ascribed to the combination ain by the copyist. Frans : mains B. Trist. 3327 is discussed by Muret, edition, p. xliv. On the basis of Gottfried of Strassburg’s rime Isot als blansche mains : Kdedin li frains , which he probably derived from Thomas, and Heinrich of Freiberg’s appellative li frenis of the same knight he concludes ‘ nous devons peut-etre retablir au vers 3327 une epithete tradi- tionelle, distincte de l’adjectif franc.’ Finally, though the text does not belong strictly to the same dialect, and yet pointing in the same direction as far as the pronunciation of the syllables in question is concerned, there should be cited from the R. Th&bes the rimes ren : germain 6807, and demen ( demain ) : ben 8249, 8271, ben wen (vanum) 8487. Here also ien rimes with uen, as in rens : suens 4343. The same rime of ain : ien is probably also involved in the follow¬ ing series from the Vie de Saint Thomas of Gamier de Pont- Sainte-Maxence where prochain rimes with main ’.plain: sain : soverain on p. 53 and in the form prosceins with biens: miens : riens :jiens on p. 130. It seems to me that it will be difficult to avoid what appears to be the x evident inference demanded by these 1 Cp. also Suain:plain (plenum), La Vie de Seint Edmund le Rei 3703, edited by F. L. Ravenel, Bryn Mawr College Monographs, 1906. 9 648 JOHN E. MATZKE. rimes. The fusion of ien(s ) with en(s) on the one hand and with ain(s) and ein(s) on the other, together with the rimes of ains or eins with ens or enz in texts where ens was pro¬ nounced gns, allows only of one conclusion, viz., that ain(s ) and ein(s) in the dialect represented by these texts tended in the direction of their modern value, and had certainly arrived at least at the stage ein{s). That iens should then in the pen of a copyist speaking the same dialect become iains or ieins need cause no astonishment. This orthog¬ raphy merely represents his effort to make the rime accept¬ able to the eye. How far geographically this orthographic habit extended I am unable to say. Stock, Rom. Stud., 11 , p. 468, cites a similar example (chrestiains : sains) from the Rom. de Mahomet, 1. 1091. Angier writes premerain: Maximiain, Gregoire 257, and Miss Pope, l. c., p. 14, adds arrien : sen, arriens : tens but arrieins : veins from the Dia¬ logues of the same author. There are, no doubt, other instances of this orthography that might be collected, as for example primeraine : meiaine (moyenne) in the Bestiaire de Gervaise, 1 1. 503, but it is evident that the harvest outside of the manuscripts of R. Troie and B. Chron. is limited. Occasionally a Latinism of the type cited above, p. 644, may be found; cp. ancian : an, Gaimar 1682, but ancien: mien, ibid. 4319, Oetavian : pan, Marie de France, Lanval 85, Troian : oan, Eneas 1699, 2109, Troians : chans, R. Troie 2299. Suchier, Altfrz. Gram., p. 75, adds some similar orthographies from the R. Rou and the Ps. Cott., but all such examples are rare and do not affect our problem. 3). aint — eint, ainte — einte. As in the preceding rime-groups the Anglo-Norman and Norman texts under consideration present little beyond l Cp. Romania, i, pp. 426-442. AI AND El IN FRENCH BEFORE NASALS. 649 simple mixture of the two syllables. However, as before, a few isolated rimes show the direction in which the develop¬ ment tended. Suchier, l. c., p. 73, cites nient: veint and nient : desteint from Sanson de Nantuil. We may add, from B. Chron., ateint: destruiement 17391, teint (tenet) : aimt 20779, feint (fingit) : vient 21670, seinte : reinte (redempta), Angier , Gregoire 1467 and ceynt (cinctum) : torment , fivang. de Nicod&me, version C, 767. 4). aindre — eindre. Our texts here show only fusion of the two, but never rime with endre in such words as prendre. Since verbs with this ending are comparatively frequent, it is rather remark¬ able that this should be the case. Only mendre—meindre < minor might seem to contradict this rule. However this con¬ tradiction does not exist in reality, for mendre appears to be the original form of the word and the diphthong in meindre is due to the analogy of meins < minus. As a matter of fact mendre is found frequently at the end of the line, but, as far as my observation goes, it rimes only with words like entendre , B. Chron. 403, 22516, descendre 6243, prendre 10159 and never with remaindre , feindre and the like. This is true of all the Norman texts in this study, the Clef d? Amour in¬ cluded. 1 An instance of what indeed might seem to be mixture of eindre with endre exists in strophes 54 and 111 of the Livre des Manures; cp. raindre (r. raeindre) : pleindre : remeindre : ateindre , str. 54, and defendre : vendre : prendre : raiendre , 1 1 have noted but one exception to this rule, meindres : greindres, Brand. 1004, which invites construction into an argument for the pronunciation of greindres of the type of eint : ent just noted. However the matter is too doubtful to be pressed ; we probably have to do simply with an early sporadic case of the analogy so common later. The old form persisted for a long time. Christine de Pisan rimes chambre : tendre : remembre : mendre , Livre du Dit de Poissy 280, though elsewhere she joins meindre : remaindre, Livre du Due des Vrais Amants , 1155. 650 JOHN E. MATZKE. str. 111. Both raindre and raiendre represent the same Latin verb redimere > raembre, which on account of its unique form was drawn over to the -ndre verbs, like tre- mere > oraindre. But while the stemvowel in both forms, raeindre and raiendre is entirely analogical, it is probably wrong to consider the two as equivalent. The former, pro¬ nounced raeindre , could rime with pleindre, the latter was raiendre , if not raendre, which forms rime with prendre. Another rime pair with similar bearing stands in Guil¬ laume le Marshal, 1. 2883, remendre ( =remaindre ) : at- tendre . Unfortunately this rime loses its argumentative value from the fact that the last six letters of the second word represent a manuscript correction not written by the original scribe in the place of something else that has been erased. 5) . einge. As early as the Brandan this syllable could rime with enge as in prenge. It is not of frequent occurrence, being found only in those peculiar Anglo-Norman present sub¬ junctives in -ge as meinge : prenge, Brandan 119. Since the second word here certainly contained the nasal f, cp. calengent: prengent, ibid. 1472, meinge must have had at least the value of meinge. 6) . aine — eine. The rimes for the most part show only the usual mixture of the two syllables. Instances proving the development of the pronunciation in the direction of fne are rather late. As a matter of fact the combination of g —nasal -f- e is rare in Old French. Its main representatives are the French forms of femina and regnum. Rimes between these two words are well known in Old French texts. Without in any way aiming to be exhaustive I may cite Comput 469, Eneas 3, Wace Brut , though not the R. Rou , cp. Pohl, Rom. Forsch ., II, p. 554, R. Troie 3937, B. Chron. 1621, B. Trist. 287 as AI AND El IN FRENCH BEFORE NASALS. 651 texts where an and en are not mixed in rime, from which it follows that femme : regne means f$me : rgne or r%ne. Now these same texts in a limited number of rime pairs show these words joined to others in -aine or -eine ; cp. peine : regne, Eneas 2523, estreine (strena) : regne, R. Troie 8317, Loherenne : femme, B. Chron. 18052, regne : Loheregne, ibid, 18066, vilaine : reigne, B. Trist. 57. The conclusion that in these texts aine or eine mean at least fme will be difficult to avoid. 1 Another combination of rimes pointing in the same direc¬ tion is that of aine or eine with iene, similar to that of ain(s) or ein(s) with ien(s) already noted. As earliest instance of this fusion might be cited paaine : soltaine, Eneas 2141, though it is of course possible to regard this rime as a Latinism. It is different, however, with plataine : Egiptaine, R. Troie 22995, and B. Chron. vilaine : paene , 1-951, do- maine : paene, n-15812, (cp. anciene : paene, II-57, paene: crestiene, ii—3073, 4383), mundaines : celestienes, ii—20898, estrienne (strena ): Bauveisienne, n-18484, Rentiene (=de Reims ) : plaine, B. Trist. 3727 and maenne ( moyenne ) : enehaenne (== enehaine), Clef d’Amours 3419. These words rime only if aine or eine are pronounced eine or ene. The Livre des Manures has three strophes which make some difficulty; cp. the following rimes: anciennes :paien- nes : sennes (synodos) : fames (feminas), str. 247 ; fame (femina) : fame (fama) : raiemme : jame (gemma), str. 60; enteime: deraime : sorseime : feme, str. 312. It is seen that in strophe 60 fame = fama seems to oppose the value of for fame= femina. Kremer in his edition of the text p. 27 cites the various explanations of this discrepancy that 1 A single rime in Adgar’s Legends , mainent: chantent 4-9, seems to con¬ tradict this conclusion. It is so unique that I regard it as erroneous for hantent : chantent. 652 JOHN E. MATZKE. have been offered, but they all fail to satisfy. It seems to have been overlooked that the line containing fame = fama, 60 b, is too short by one syllable. While this fact tends to throw doubt on the reading of the line, it does not, however, prove that the rime is incorrect. We will have to accept femme as riming in this text both with e and with a. The same phenomenon is not unknown in other texts where an and en are kept distinct. Cp. realme : femme, Gaimar 3601, but mereenne : femme 2507, regne : femme 2531, fame : dame, R. Troie 18154 ; see also Rom. Stud., ii, p. 39, and Rom. Forseh., ii, p. 554. 7). aime — eime. Only very few rimes of this category are available. Since eime < ema does not exist in Old French, dime < ama can only rime with itself. Later, however, when dime had become eime, a few other words having gme became available. This, I believe, is the explanation of esment (estimant) : eleiment, Angier, Gregoire 645, aime : baptesme, Simund de Freine, Vie de Saint Georges 1324 and meime {= mesme) : aime, Rom. de Philosophie 981 of the same author. Suchier, Altfrz. Gram., p. 71, thinks esment stands for eiment and that the disappearing s had called forth a j which formed a diphthong with the preceding vowel. Granting that this was the case, we should even then have evidence in these rimes that dime had become eime. I doubt, however, whether this explanation is exact. That s before l and n in the course of disappearance passed through a sound capable of palatalizing the following consonant seems fairly reasonable, but this question can not enter into the dis¬ cussion here. Examples in support are cited by Wal- berg, Le Bestiaire de Philippe de Thaun, p. lxv. It does not follow, however, that the process was identical before m. Here it is much more likely that the intervening sound AI AND El IN FRENCH BEFORE NASALS. 653 was a voiceless m, such as Wulff 1 has shown to exist under similar circumstances in Andalusian Spanish. When finally all trace of the s had disappeared, the vowel in words like esment , baptesme was g, and ei and ey are only graphic variants utilized by the copyist to represent this sound. Then these words were pronounced gment, baptgme and the rimes in question are entirely parallel to those cited above for eine : gne. This conclusion is emphasized by the following rimes of crime :ieme from the R. Thebes; cp. crement : ement 5077, creme : afeme (adfamat) 7377, entrement: crement 8603, creme : erne 9199. All the evidence presented so far goes to prove that in the northwest of France and in England ai and ei with n had become ein about the middle of the 12th century, and that under favorable circumstances the sound could even approximate gn. The evidence of the orthography points in the same direction. We have to do with a region where for the most part an and en are kept distinct. A scribe who pronounced en -|- cons . as $n would not have written en for either ain or ein, had he pronounced these am, as Brandan enz (ainz) 1010 ; desclem 534, sen (saint) 157, nor would he have introduced ein in mendre if meindre had meant maindre for him as in the rime tendre : meindre, Ipomedon 2651. A survey of the orthographic habits in Anglo-Norman texts can be found in Slimming’s edition of Boeve de Haumtone, 2 pp. 185, 196-7, and 201, and in the manuscripts of conti¬ nental texts similar habits prevail. Gorlich arrives at the same conclusions from the study 3 of the original documents 1 Un chapitre de phonetique andalouse , in Recueil de Memoires philologiques presente a Monsieur Gaston Paris par ses elh'es suedois, 1889, pp. 45 ff. 2 Bibliotheca Normannica, vol. vn. 3 Gorlich, Die Nordwestlichen Dialekte der Langue d? Oil , Franz. Stud. , V, Heft 3, pp. 17-18 and 41. 654 JOHN E. MATZKE. of the Northwest dialects of the second half of the 13th and the 14th centuries. Besides the noncommittal ein he cites e in preschene, prochen, Alen, men, nonen, fren, plen, plene, Magdalene, ae in maen, daraene, preehaens, prochaenne, aei in Alaein, prochaein, ee in dareen, oei in Moeine and oe in damoene, demoene. We may close this division of our subject with the fol¬ lowing passage from the Orthographia Galliea 1 which em¬ phasizes the conclusions at which we have arrived; 6 item diversitas scripture facit aliquarum diccionum quamvis in voce sint consimiles, verbi gracia .... teindre tendre tenir attendre atte [i] ndre .... aymer amer/ which can of course only mean tendre and teindre, attendre and atteindre are pro¬ nounced alike. The rule is found only in mss. C and O, i. e., in the later version of this earliest grammatical treatise, but even thus it gives valuable evidence of the pronunciation of our syllables in the 14th century, and is entirely in harmony with what we have been able to observe so far. II. Leaving the Norman and Anglo-Norman division of Old French texts we may now continue the history of our syllables into the dialect from which the Modern French most directly sprang. One of the earliest texts to be cited here is the Roman de VEscoufle, written about the year 1200. In this poem en and an -f- cons, are kept apart in rime, in spite of the fact that feme rimes with ame, as 1. 125, and gemme with dame, as 1. 5739. But as already pointed out above, p. 652, the same phenomenon can be observed in other texts sepa- 1 Ed. Stiirzinger, Altfrz. Bibl., vol. vm (Heilbronn, 1884), p. 14. AI AND El IN FRENCH BEFORE NASALS. 655 rating en and an. 1 The nasal fn exists in such rimes as suensisens 3189, vit Ven:Julien 4867, enmena : cil Ven a 3647, amena \ dame en a 5969. In the light of these losenge: vos ain ge 2877 can only be interpreted as meaning lozenge : vozgnge. The Roman de Guillaume de Dole , which Gaston Paris was inclined to attribute to the same author, 2 shows the same mixture of feme: dame 1508, 3008, and rimes maint (manet) : esloint 4192. This fusion of ain :oin y which occurs here, as far as I know, for the first time speaks for a pronunciation gn : ogn. Another isolated early example of the same rime, though of uncertain date, can be found in the Rom. de Renard , branch 9, written by the priest of La Croix-en-Brie, moines (monachus) : poines (penas) 505. From this date forward it is possible to cite an uninter¬ rupted series of texts, including Marot, all showing the same fusion of ain and om, which thus present unmistakable evi¬ dence that in the dialect centering in Paris both ain and ein were steadily developing toward their modern value. I copy this list without further comment, and as much as possible in chronological order. The fact that these rimes are few in number in comparison with the ain : ein rimes probably means that they were felt to be irregularities or rime licences, and this will have to be taken into account in the final estimation of their meaning. Gaufrey 3 — hautaine : Karlemaine : humaine : emmaine : regne : souveraine : demouraine : demaine : essoigne : Couloigne : entente : jenne (=jeune) : quarantaine , p. 316. 1 For Picard and Wallonian cp. Haase, Das Verhalten der pikardischen und luallonischen Denkmaler des Mittelalters in Bezug auf a und e vor gedecktem n. Halle, 1880, pp. 41 ff. 2 Cp. Romania , xxxii, pp. 487-488. He there suggests 1185 as the date of the Escoufle and 1200 as that of Guillaume de Dole. The last edition of his Lilt. Frang. au moyen dge places Guillaume de Dole in 1200 and the Escoufle in 1210. 3 Cited by Engelmann, l. c., p. 23. 656 JOHN E. MATZKE. Roman de la Rose—paintes (pinctas) : cointes, I, pp. 20, 30, 46-47, paintes : pointes (punctas) I, p. 31, maintes : ointes, I, pp. 249-250, poine (pena) : moine (monachus) i, p. 100, saine : essoine , i, p. 73. Rutebuef — avaine : vaine : couvaine , I, p. 33, poigne (pugna) : sovraine : moine : essoine , I, p. 153, lainne : avainne: semainne , n, p. 57, demaine : moine, n, pp. 122, 137, moine: emmaine, n, p. 129, Jordain : enjoin , II, p. 276, nonains : sains : certains : mains (minus) ii, p. 42, plaindre ijoindre: poindre, I, p. 216, saintes : jointes : empraintes : maintes, n-253. Eustache Deschamps — moins (minus) : mains (*manti) : mains : vains, i, xxv, conjoint : ^>om£: vaint (vincit) i-lxxi, doint (donet) : pourpoint : point : point : faint (fingit) : vaint: (vincit) ii-ccxl. Christine de Pisan—loings : moins : besoings ifroins (fre- num) i, p. 26, moins : besoings, I, p. 56, mains : mains (minus) i, p. 123, hi, p. 40. Miracles de Nostre Dame—moins : chastellains, iv-178, lointain : soing, xi-9, moine (monachus) : amaine, xvm- 1313, estraine : royne, iv—908, cp. Guillaume Alexis and Villon below. Charles d’ Orleans—plains (plango) : plains (plenus) : moins : mains, 1-192, besoing : loing : baing : poing, n-98, avoine : Touraine :paine : sepmaine, n-157. Guillaume Alexis—fainctes (finctas) : coinctes, Deb. de VHomme et de la Femme 128, traine : demaine : pourmaine : chanoine : gaine : maine, Blason des Faulses Amours, str. 40, primeraine : essoine : Anthoine : royne : villaine : loingtaine, ibid. 61, mains : mains (minus) : plains ; mains: poins : (pugnus) : point (punctum), ibid. str. 80, chanoines : peines : demaines, Martyrologue des Fausses Langues 254. IJAmant rendu Cordelier — mondaines : avoynes : marjo- AI AND El IN FRENCH BEFORE NASALS. 657 laines : certaines 410, saindre (cingere) : estraindre ijoindre : atteindre 1706, baings :aubefoings ipoings : mains 1762. Villon—fain (famem) \foing (fenum), Poesies Diverses 140, bain (ms. boing ) : poing, ibid. 148, Anthoine : Saine : essoine : ydoyne , Pet. Test. 226. Marot — moindre: paindre, 1—104, moindre ijoindre, 1 — 162 , veine : Antoine , 1-239, Antoine : souvienne , 1-239, moins : in- humains , n-313. In the light of these rimes certain other combinations, which would have little argumentative value by themselves, may be advanced as pointing in the same direction. Roman de la Rose—vaine : raine (regnum), i—15, Lohe - regne 1 : regne , ibid., 1-5. Rutebuef—raine (regnum) : chanoine , II-119, regne : resne : saine : plaine : estraine : raisne , 1-127, raignes : raines (ranas), ii-90, vilaine : raine (regnat), h-206, amaine : raine (regnat), II-254, tain (teneo ): soucretain, II-118, 137, 139, vain (venio) : vain , n-139. Guiot de Provins—Aquitaine : Vienne , Bible 334, eitoien : vilein 998, Magdalene : eertene 2230, Egipeiene : Elene 2248. Miracles de Nostre Dame—Estienne : maine, xiv—389. Eustache Deschamps—Requiem : prouchain : moien : bien , i-XLViii, certain : cappitain : tain (teneo), i-clxxiv, Ro- mains : Rains : plains : restrains : tiens : certains , H-CCLIII, plaine : prouchaine , I—xii, paine : aviengne , I—x VI, craime 2 ( crhne ) : aime, i-xxxm. Christine de Pisan—Athenes : certaines , I, p. 250, peine : Polixenne : vaine : prochaine , Debat de Deux Amants 692, ancients : humaines : fontaines : mondaines , Livre du Dit de 1 The ending in this word, which can be found elsewhere and earlier, is evidently due to analogy with regne, as if the word meant the kingdom of Lorraine. * The pronunciation is indicated by the rime baptesme : cresme, Mir. d. Nostre Dame , xx-643. 658 JOHN E. MATZKE. Poissy 660, Magdaleine : peine, Oraison de Nostre Seigneur 215. Guillaume Alexis—penne (penna) : penne (pena) : penne (= Mod. Fr. panne), Le a b c des Doubles 1056. This unique composition, among other peculiarities, is composed entirely in rimes bquivoqu&es, whence the spelling of the second rime word. There can be no doubt about its pro¬ nunciation. 1 L’Amant rendu Cordelier—jenne (=jeune) : mondaine 169 ; cp. the same combination cited from Gaufrey above, p. 655. Mistere du Viel Testament—Damascene : regne 4341. Villon—douzaine : Estienne 2 : paine : sepmaine, Grant Test. 1913, roynes : regnes : Penes : estrenes, ibid. 414, Neapoli - taines : Pruciennes : Egipciennes: Castellaines, ibid. 1524, vil- laine: Helaine, Poesies Diverses 42, Magdelaine : laine, ibid. 53. Marot—veine : Antoine, 1-239, Antoine : souvienne, 1-239, tienne : estraine, 1-245, Magdalaine : Helaine : souveraine : 1 In view of the positive evidence of this rime and those cited above, ayme ( amat) :jlamme : femme : blasme : flame : enflame, Blason des Faulses Amours, str. 105, and dame : fame, Passe Temps de tout Homme et de toute Femme 1083 must be accepted asLatinisms without evidence for the history of our syllables. Similar rimes are cited by Metzke, Her Dialect von lle- de-Franee im 13. und 14. Jahrhundert, Herritf s Archiv , lxy, p. 61. The same is probably true of claime : aime : semme (seminat) : reme (= rame) in Christine de Pisan, Debat de deux Amants, 484. 2 Villon rimes also ien with & as in ancien : Valerien : an, Grant Test. 1552. A similar rime is aneiens : sciens in Guillaume Alexis, Le ab c des Doubles 113. The editors Piaget and Picot in a note to this line cite aneiens : Cana- neans, Mist. Viel Test., m-23050, and crestiens: ceans, Montaiglon, Becueil, 1-53, to emphasize the fact that similar rimes are occasionally found in other authors of this same period. We may add ancienne : Adriane : sienne, Guillaume Alexis, Martyrologue des Fausses Langues 198, Adriane : moyenne, ibid. 209, as showing the same liberty also for the feminine form of this ending. Yet the regular pronunciation of this author was iene, as shown by the rimes sienne: reviengne, Faintes du Monde 266 and preigne : appar- tiengne, Passe Temps de tout Homme 3703. Cp. also Nvrop, Gram. I, § 218. AI AND El IN FRENCH BEFORE NASALS. 659 seraine, 1—337, Heleine : aveine : alaine : plciine, 1—418, Philo - mene : meine , 1—488, peine : Clymene, ii— 175. We must now face the difficult question how this rather definite evidence of these rimes extending over a period covering several centuries can be harmonized with the state¬ ments of the early grammarians 1 in regard to the pronuncia¬ tion of our syllables. The natural interpretation of the rime of ain or ein with oin as meaning en : oen appears to be invalidated by the fact that the early grammarians teach that these syllables arrived in the 16th century with a full pronunciation of their diphthong. To harmonize this dis¬ crepancy it might be argued that the grammarians were influenced in their statements by the written form of the syllables. Yet this point of view is scarcely tenable in view of the rather positive assertions made by the more accurate among them, who even go so far as to make use of phonetic transcriptions, as Meigret and Baif; Cp. Thurot, l. c., p. 342. Or it might be maintained that the popular pro¬ nunciation differed from the literary. The people said en, but good taste demanded ein. Here we are met by the fact that the earlier authorities are practically unanimous in their statements, and that we have in general little evidence of such discrimination on their part. There seems no possi¬ bility of avoiding the acceptance of their testimony as fairly exact, and if this be so, our problem consists in finding the explanation which will coordinate their assertions with what seems to be the evident history of our syllables. In an article on The Pronunciation of the French Nasal Vowels in ain ein in the xvi and xvii Centuries published some years ago in the Publications of the Modern Language Association, ix, pp. 451-461, I endeavored to interpret this whole body 1 Cp. Thurot, De la Prononciation Frarqaise, Paris, 1881, vol. I, pp. 321 and 342. 660 JOHN E. MATZKE. of evidence from this point of view. I now desire to modify the conclusions reached there to a slight degree, on the basis of the material presented in this study. The hidden difficulty of the whole problem probably lies in the value of the n. The i preceding it being originally a glide ’which developed between the nasalized a or e and the dental n , as in vanum > vain , plenum > plein, it is evi¬ dent that as long as this n kept its original value, which was the case when the syllables in question were final or followed by a consonant, or as long as eine was pronounced eine, the conditions which had produced it originally con¬ tinued to be potent, and ein and eine must have tended to retain the diphthongal value of their vowels. Here we have the explanation why the grammarians almost unani¬ mously speak of a diphthong in this connection. It is perfectly possible, to be sure, to pronounce en without an intervening i , and that this was often done is proved by the rime licences which we have observed in the earlier portions of this study. But it must be borne in mind that, when eins rimes with ens, the dental n is still present, and the rime does not represent the modern pronunciation. And similarly the union of eine with oine or ienne means e(i)ne : oene or iene , and not gne : ogne or igne. The fundamental changes, by which the modern value of our syllables was established, took place during the 16th century. At that period the dental n disappeared, the diphthong became a simple vowel, and the nasal vowel in the feminine syllable became an oral vowel, ein developed into g and eine into gne. How this change came about must be a matter of surmise. Probably the diphthong was at first nasalized throughout, ein became em, and its second half then passed rapidly through gen > ggn > g. In the article cited I suggested a shifting of the accent from gen ]> gen for the latter half of the century. This presumption AI AND El IN FRENCH BEFORE NASALS. 661 was based upon the interpretation of the statements of certain of the grammarians, which seem to demand such an inference. However, their language is obscure and it is not impossible that this interpretation is imperfect. A shift of the accent is certainly not necessary, for |en can become § quite as easily as gen, provided the dental n becomes silent. In the case of gine the process was not quite identical. Here the nasal quality of the vowel disappeared, and in conse¬ quence the diphthong gi was readily reduced to a simple vowel, but the dental n remained; eine become gne. If this point of view is accurate, the Old French rimes harmonize perfectly with the evidence of the 16th century grammarians, and it is seen that the growth of our syllables was constantly in a uniform direction toward their Modern French value. III. We may now turn to a group of texts in which the development made evident so far has not taken place, and where, on the contrary, ein has become ain. Texts in assonances showing this development are enu¬ merated by Engelmann, l. c., pp. 22 if. The evidence is contained in the laisses having ai and ei -f nasal in asso¬ nance with a or 6 + nasal + consonant. Since an and en have here undoubtedly the value of an, it follows that ain and ein were pronounced din. Cp. the following assonances: Aie d y Avignon—comence :Elainne : dame , p. 53. Aiol—porpensent : Losane : Bretaigne : renge, 11. 8768 ff., bataille : Chartres : kaine (catena) : arse , 11. 8800 ff. Ogier—demain : grant 2318, France latendre 5970, gentes: Bomaine 8497, lantaine : cravente ipoissance 9034. Benaus de Montauban—Karlesmaine : remaigne : ceigne : ensamble : Flandre: Bomaine 142. 662 JOHN E. MATZKE. Amis et Amiles—-parrain : main : desirrans : approchant: vaillant: gaaing : sain : Romains : serjant 2499, plains (ple- nus) : sains : mains : certain 3080, convent : volant : gent : serjans iforment 1803. Jourdains de Blaivies—Jordains : dolans : vans (ventus) 2192, parrain : certain : Jordain 3034. Very often aigne and eigne are joined to the words cited. To introduce these into the discussion at this point is con¬ trary to the plan of this study, the object being to establish the pronunciation of ain(e) and ein(e) as a basis for the further elucidation of the history of aigne and eigne. It is sufficient therefore for our purpose at this point to note that in the texts cited above both ain and ein were unquestion¬ ably din. The search for similar proof in the rimed texts has proved rather fruitless. I am able, however, to cite the following pairs of ain : an — Richars li Biaus , bans (— banc) : a(i)ns 2007, main : Flamain ( =flamand ) 1607, Floris et Liriope , meshaing : awan 563, Rom. de Renart , vilein : Brian lb, 2981, Octavian , ata\i]nt: garant 2287. While these assonances and rimes are few in number, it is interesting to note the rather circumscribed territory to which the majority of these texts belong. All have distinct Picard characteristics. Aiol and Renaut de Montauban are classified by Grdber, Grundriss, as Francian-Picard. Richars li Biaus is placed by Forster, ed. p. ix, in the neighborhood of the French-Belgian frontier, and Knauer 1 agrees in the main with this localisation. Octavian is a Picard text copied by an Anglo-Norman scribe. In view of these facts the Picard origin of the trait under consideration becomes a pertinent inquiry. The help of the assonances and rimes being exhausted, our only resource can 1 Zur altfranzosischen Lautlehre, Leipzig, 1876. AI AND El IN FRENCH BEFORE NASALS. 663 be the orthography, to be studied primarily in charters and original documents. Here we must move very cautiously, and conclusions can be established only within certain limi¬ tations. Yet where we find written constantly am, ein being almost unknown, the inference is certainly justified, that in the opinion of the scribes the pronunciation was ren¬ dered more accurately by ain than by ein. Such is the case in the texts studied by Raynaud, Le Pialecte Picard dans le Ponthieu , Paris, 1876. In these original documents extend¬ ing over the years 1254—1333 the orthography ein seems unknown ; cp. p. 67. The same is true of the documents of Champagne examined by Forster as the basis for his study of the language of Chrestien de Troies, cp. Cliges, pp. lxi-lxii. The same predominance of ain we find in the manuscripts of the following texts examined for the present purpose: Aucassin et Nicolete, Pis dou vrai Aniel, Pichars li Biaus, Ille et Galeron , Trouvh'es Beiges {Beaudouin de Conde, Jean de Conde, Queues de B&thune, Jacques de Bai- sieux ), Guillaume de Palerne, Jean Bodel {Jeu de Saint Nicolas), Adam de la Halle, {Jeu de la FeuilUe, Robin et Marion ), Floris et Liriope, Robert le Diable, Durmart, Cheva¬ lier as deus Espees, Raoul de Houdenc, Philippe de Beau- manoir, Chronique de Floreffe, l Froissart. If the argument of the orthography is of any value the pronunciation din for both ain and ein should be ascribed to these texts. It is possible to control this conclusion to a certain extent by comparison with the modern dialects. Gilli^ron’s monu¬ mental speech atlas 2 (the first 16 fascicules) contains a number of maps of service in this connection. These tabulate the pronunciation of the words bain, andain, main, essaim, Rain, faim, douzaine, chaine, faine. For all of them the pre- 1 Cp. Peters, Zs. f. rom. Phil., xxi, p. 12. 2 Atlas Linguislique de la France por Gilli^ron et Edmond, Champion, Paris. 10 664 JOHN E. MATZKE. ponderance of pronunciation in the whole langue d’o'il territory is | or gn. However there are two regions where d or an with various shades of vowel are the rule. These are the Northern portion of the department of Manche with sporadic instances in Calvados, and particularly the Northern portion of Somme and the department of Pas de Calais, going now and then over into the contiguous territory. The same pronunciation is cited by Eggert, Zs. f. rom. Phil., xm, pp. 375 and 380, for the patois of Val de Saire, La Hague, Guernsey, and Jersey, and corroborated for the dialect of Guernsey by Lewis, Publications of the Modern Language Association , x, pp. 18-19 and 27-28. Now it is, of course, entirely possible that we have here a modern development of older qn and gne. Yet, if the development of ein into l current in Modern Wallonian is taken into consideration, cp. Gilli6ron’s Atlas and Horning, Zs. f. rom. Phil., ix, pp. 482 and 484, together with that into on as in von (vena) avon (avena) pon (pcena) in the ,same region, where an older ei had become oi under the influence of a preceding labial, the conclusion is made rather probable that the basis of this modern a is an older din. Now it is striking to note the number of texts in the list just cited, which fall into the region described in a general way by the words departments of Pas de Calais, Nord, and Somme, though to be sure not all of them can be definitely assigned to this section. This is the territory of the Picard dialect, and these two facts taken together seem to establish the conclusion that the development of ein to ain, made evident by the overwhelming use of ain as the graphic expression of the sound, is a characteristic of the Picard dialect extending South into the Francian and Champenois. The further question of the geographical expansion of this trait the material at hand does not solve entirely. We can give, however, some indications. According to Forster’s AI AND El IN FRENCH BEFORE NASALS. 665 observations 1 it included the district of Champagne, and in that case the same pronunciation must be assigned to the Roman de Galeran , whose home probably lies in the de¬ partment of Aisne. But its spread was stopped by the Burgundian dialects, where, as is well known, ein became oin, while ain remained, thus showing that the two syllables had not coincided. This is the condition in the Lyon Yzopet, 2 though this text contains one rime, rainne (rana) : poinne (pcena) 165, of a -j- n with e -j- n. The evidence collected by Gorlich, Lev Burgundische Dialekt im xiii und xiv Jahr- hundert , 3 corroborates this statement. In the documents examined by him a + n is represented by ain and ein, very rarely by oin, p. 18, while e n appears regularly as om, ein, ain , pp. 62 ff. Gorlich is inclined to look upon ain in the latter case either as a stage in the development of ein to oin , or as a later secondary alteration of om, but does not at all associate it with ain from a -J- n. Further North in Lorraine both a -{- n and e -j- n are written ain, ein and en, as, for instance, in the Lorraine Psalter, and Apfelstedt, the editor 4 of the text attributes the value of e to these various spellings. Occasionally the more Southern oin has crept in, but the former value is confirmed by the modern forms of the words in question. In the patois spoken between Metz and Belfort a -f n has become f, while e -f- n is i, except after a labial, in which case it has become o or on; cp. Horning, Die Ostfranzosischen Grenzdialekte zwischen Metz und Belfort, Franz. Stud., y (1887), pp. 9 and 29-30. Outside of the Burgundian-Lorraine region and yet closely related to it we may cite the Roman de Fortune et Felicite of the 14th century by Renaut de Louhans studied by Nagel 1 Cliges von Christian von Troyes, p. lxii. 2 Cp. Forster’s edition of this text, Altfrz. Bibl., v, pp. xxvii and xxxi. 3 Franzdsische Studien, xii (1889). * Lothringischer Psalter, Altfrz. Bibl., iv, pp. xii-xiii and xxi. 666 JOHN E. MATZKE. in Zs. f. rom. Phil., xv, pp. 1-24. We there find such rimes as plainne (plena) : encienne :fontainne : certainne, terriennes : vainnes, sains ifisiciem, chastellains : gar diem, biens : maim which, though not entirely conclusive, yet seem to point rather toward the pronunciation |n(e). In the direction of the Wallonian we have the Pohne Moral and the Chronique of Philippe Mousket. Neither text allows satisfactory con¬ clusions. The former seems to demand din for both a and e -f- n, but the argument is based entirely on the orthog¬ raphy, 1 and is unsatisfactory. Philippe Mousket rimes ain and ein, and often ain is written for both. 2 The present dialect of Namur, East of Tournai the probable home of Philippe Mousket, has g for both syllables; cp. Nieder- lander, Zs. f. rom. Phil., xxiv, pp. 7 and 22. Toward the West and Southwest the din region also had its definite limits. We may note the Roman de Carite et Miserere written probably at Molliens-Vidame near Amiens ; cp. the edition of this text by van Hamel, Paris, 1885, p. cxcv. Here we find in strophe 41 of the Miserere the rime pena : peine a, which gives definite evidence for the pronunciation peine at least. The modern dialects around Amiens vary between | and a; cp. Siitterlin, Heutige PiJcard- isch-Franzosisohe Mundarten, Zs. f. rom. Phil., xxvi, pp. 289 and 297. Before leaving this portion of our study it is necessary to discuss briefly the Estoire de la Guerre Sainte written soon after the year 1191 by Ambroise, an ordinary knight of the third crusade in the army of Richard. Gaston Paris localized this poem at Evreux. The text shows the usual confusion in the rimes of ain(e) and ein(e), aindre and eindre, aint(e) and eint(e) and no proof for the pronunciation could be 1 Cloetta, Poeme Moral , Rom. Forsch., hi, pp. 49, 58 and 61. 2 Link, Die Sprache der Chronique Rimee, Erlangen, 1882. AI AND El IN FRENCH BEFORE NASALS. 667 drawn from these, though Gaston Paris, p. xxvii of the edi¬ tion, accepts din as the common value. The difficulty arises through the rimes of aine and erne with iene. We have here such pairs as Estienes ipaienes 10488, cristiaine :paiene 6353, teriane : cristiane 41, 3711, *3975, cristiane : paaine 2323, 3929, chaane (catena) : cristiane 3388, chaane ipaiane 3935. On the basis of the last two pairs Gaston Paris believed (p. xxvii) that a double pronunciation of iene must be accepted, viz., on the one hand the normal one in paiene crestiene, and on the other paiaine crestiaine . It is certain that the local¬ ization of this text at Evreux does not actually conflict with the argument which may be based upon the modern patois of this region; cp. above p. 664. But on the other hand the evidence for din is absent, except such as may be drawn from the orthography ; and it will be seen that this may be ex¬ plained otherwise. It should be noted in the first place that with the exception of chaane the ending ane is restricted to teriane , cristiane , and paiane , three common mediaeval words, whose Latin forms were familiar to every copyist and might readily affect the French form without influencing the pronunciation. If this point of view be valid, chaane would then be merely an adaptation of the word to its mate in rime. If the scribe said cristiene and wrote cristiane , he could also write chaane and mean chaaine, which he couples with demaine, 1. 9006. A double pronunciation is imaginable for cristiene , but scarcely for chaeine. In the next place it should be observed that aine meant %ne for the scribe, which is evident from the following series of rimes : Avesne : regne 6177, 6637, Charlemaines : regnes 8479, regne : cheveitaigne 8607, cheventaine : lointaine 7087, cristiaine : paiene 6353. Unless we are much mistaken, therefore, the matter is much less complicated than Gaston Paris imagined, and we have in the language of Ambroise in this particular another example of the type of R. Troie, B. Chron., and other texts examined above, p. 651. 668 JOHN E. MATZKE. The present investigation has shown, I think, two things. On the one hand a continuous line of assonances and rimes has been cited proving the development of ain in the direc¬ tion of ein > en , belonging to the Norman and subsequent Francian dialect, and on the other evidence has been brought forward showing that ein had become ain and an in a por¬ tion of the Picard and neighboring speech forms. IV. AIGNE— EIGNE. While the mingling of aine and eine in rime is frequent in French texts after the Brandan , the similar fusion of aigne and eigne became customary much more slowly. Of the Anglo-Norman texts in our list the first to show it is the Mist. Adam. Here we have, 11. 618-621, the follow¬ ing sequence of rimes with partial assonance: enseigne: feigne : guerre : pleigne. The inference that both aigne and eigne were pronounced ene it is of course impossible to avoid. Yet in spite of this long silence of the rimes on this point, scrutiny of the earlier texts shows that this pronunciation must have been common since the beginning of the 12th century. In the Brandan aigne rimes with aine (semaine : eumpaine 592) and eigne with eine (meinet : enseignet 714, 1114; cp. also 11. 215, 1110, 1252). Since aine and eine are identical in sound, it follows that the absence of the rime pair aigne : eigne cannot be due to a difference in pronuncia¬ tion, and the same inference is legitimate also for the other texts of this period; cp. St. Laurent , ovraigne : peine 67, Adgar , overaigne : peine 2—99, 23—237, desdeign : vilain 30- 151. In the 13th century the fusion of the two syllables in rime is more frequent; cp. Angler y Gregoire , gaaingne : enseigne 147, enfreingne : destreingne 911, enseingne : remeingne AI AND El IN FRENCH BEFORE NASALS. 669 983, Char dry, Set Dorm., muntainne : enseinne 911, Guil¬ laume le Marechal, enseigne : remaigne 2947, enseigne : acom- paigne 3395, pleingne : enseigne 3823, plaingne : ensenne 4805, Bretaingne : ensaingne 6505. No deductions claiming a conscious separation of the sylla¬ bles in rime or inferences as to a difference of pronunciation should, however, be drawn from the fact that such rimes even at this time are not more frequent. This proportion is determined entirely by the nature of the words in point. Those with aigne are on the whole more frequently employed than those with eigne, and this comparative relation is quite well illustrated by the Ipomedon of Hue de Rotelande. In the 10578 lines of this poem I have noted only one example of fusion of the two syllables, desdegne \gregne 2389. At the same time eigne occurs only once in rime by itself, 1. 5161, and pure aigne rimes are frequent. On the continent the general picture is closely similar. The Eneas keeps aigne and eigne separate. There are 12 rimes in aigne (11. 365, 3147, 4109, 4535, 5003, 5563, 6983, 7099, 7781, 8033, 9509, 10007), 4 in eigne (11. 4523, 5569, 9485, 9897) and one of aigne with regne, regne : plaigne 1427. That eigne also might have rimed with regne is proved by enseigne : demeine 4523 and regne ipeine 2523. The B. j Rou contains 23 pure aigne rimes (11. 25, 427, 515, 661, 671, 1513, 1831, 2597, 2629, 2697, 3931, 3937, 4115, 4481, 5095, 6099, 7593, 7951, 8685, 8715, 8717, 8719, 9143), 2 pure eigne rimes (11. 1629, 3941) and only one instance of fusion, deigne : greigne 2607. We have seen above, p. 643, that Wace follows the same habit for ain and ein, but that it would be dangerous to base conclusions upon this fact, because fusion of these two syllables in rime is much more constant in the same author’s Brut, and we may add here that this same text contains a rime showing mingling of aigne with iegne, (vaigne : Bretaigne 6072), a type which we shall presently meet in other texts. JOHN E. MATZKE. 670 The majority of the Norman texts in our list presents the same appearance as those just considered. The Chron. Mt. 8. Mich, has 8 rimes in point, all in aigne (11. 49, 501, 565, 783, 1131, 1491, 1649, 2277). The last in this list is enfregne : maigne, but here the source of the first word is infrangere in the place of infringere. The Liv. d. Man. has only a single strophe (cciv) in aigne ( Espaigne: bargaigne : gaine : ateigne ), Marie de France does not mix these syllables in rime, either in the Fables or in the Lais with the exception of a single example to be cited presently. Beroul has 3 rimes in aigne (731, 2247, 4029), 2 showing fusion, 1 saine (*sanginat) : enseigne 777, enseigne : Montaigne 4017 and none in eigne. In Guillaume le Clerc’s Besant de Dieu and Bestiaire there are 14 rimes divided as follows : 5 in aigne, Bes. Lieu 2355, 3213, Best 149, 365, 1849 ; 4 in eigne, Best 1247, 2941, 3075, 3601; and 5 showing fusion: Bes. Dieu desdeign: meheign 1767, enpaigne (im- pingat) : plaingne 1903, ovraignes : enseignes 2095, Best, enseigne: remaigne 1571, montaigne : enseigne 2817. In the Clef d’Amours finally we note aigne pure 1437, eigne pure 2847 and ouvrengne : ensengne 2061, compaignes : enseignes 3133. That is to say, we have in these texts, taken as a whole, a majority of aigne rimes, due to conditions which throw no light on the pronunciation, a very much smaller number of eigne rimes, and constant examples of fusion between the two, showing that their pronunciation was identical. We may now cite the rimes in these texts which aid in determining the pronunciation. These are in the first place regne iplaigne, Eneas 1427, regne : peine, ibid. 2523, preigne (*prendeam) :feigne, Marie de France, Lanval 131, pleig- 1 In view of this fact fange : enseigne 3801 and enseigne : bamage 4109 must be looked upon as doubtful readings ; cp. also Muret, ed., p. xliv. AI AND El IN FRENCH BEFORE NASALS. 671 nent : blosteignent, ibid. Fables 23—25, entracompaignent : espraignent, Best. 2903. These must be considered in con¬ nection with certain rime pairs in Benoit de Sainte-More. In both poems commonly attributed to this author the commonest rimes are aigne : aigne, eigne : eigne and aigne : eigne, but in addition there are found a few in aigne or eigne with iegne as follows: teigne : remaigne, R. Troie 12985, teigne : chastaigne, ibid. 16851, teigne : enseigne, ibid. 18407, teigne imaigne, B. Chron. 17557, and also aigne : *prendeam as ovraigne ipreigne , B. Chron. 19475, compaigne : empreigne, ibid. 22601. That is to say we have here more extended evidence of a rime tendency just noted for Wace’s Brut and Marie de France. The Clef d’Amours points in the same direction with plengne : tiengne 613, ensengne : eontiengne 2737, prengnes : restraingnes 401, and prenge : avienge 1143, contingent: mesprengent 2067, prenge : retienge 2083, aprenges: contienges 2899, which are probably only variant spellings for the same phonetic value. Presumably this value was ene, but of course the rimes themselves contain no evidence and it is quite conceivable that teneam and *prendeam were pronounced tane and prdhe. There is, however, another set of rimes in Benoit showing mixture of aigne and eigne, with aine and cine, similar to those already noted for the earlier Anglo-Norman texts, and these throw very definite light on our problem; cp. Seigne (Sequana) : Bretaigne, B. Chron. 15044, Seigne : compaigne, ibid. 39751, chevetaingne : conpainane, R. Troie 20419, Heleine : chevetaingne, ibid. 25863, estreine : Heleine, ibid. 5069. Whatever vowel existed in aine or eine must have been heard also in aigne and eigne, and since this tended in the direction of ene, cp. above, p. 651, it follows that aigne and eigne were sounded ehe, and the rime regne : plaigne, Eneas 1427 agrees with this inference. We may conclude, therefore, that in the Anglo-Norman 672 JOHN E. MATZKE. and Norman dialects both aigne and eigne were pronounced ene, *prendeam was prene and teneam, veniam had become tiene, viene with further development into tene and vene. In our study of the history of ain and ein, after the determination of the value of these syllables in Anglo- Norman and Norman, we were led through the Escoufle and Guillaume de Dole into a series of texts where these syllables rime with oin. Only Rutebuef of our list shows the similar mixture of aigne and eigne with oigne. Since the latter was pronounced oene, there can be no question as to the value of the former. Note the following rimes : remeigne : enseigne, I— 71, remeigne :preigne, i— 71, preigne : deigne, 1-150, main- teingne : veingne : compeingne : esloigne , I—22, poingne : so- vrainne : moinne : essoinne , 1-163, avaloingnes : lontaingnes : essoingnes , 1—241, enseigne : besoingne , II-85, ouvraingne : vergoingne, n-176, coviegne : besoingne , n-184, souviegne : besoingne, ii— 294, doingne : viengne, II-376, praingne : be¬ soingne, II-311. Since regnum, written raine, raigne, or rlgne, rimes with this same class of words, cp. 1-101, 127, II— 90, 206, 254, 283, 365, it follows that all were pro¬ nounced alike. 1 The same fusion of the syllables in question with oigne is to be found in the list of assonances from Gaufrey, p. 316, already cited in part above, p. 655. To the words mentioned 1 Metzke, Herrig ’ s Archiv, lxv, p. 60, calls attention to the fact that proper names like Champaingne Alemaingne in the Francian documents studied by him are never written with e or ei, and that forms like Cham¬ pagne, montagne are not infrequent. This leads him to accept the modern pronunciation for the words in question. But the rimes which he cites from Geffroi de Paris and Gautier de Coincy, in addition to those from Rutebuef and other authors included in the present study, show such absolute equivalence of aigne and eigne, that it follows that their value must have been identical, and in a note on p. 62 he hesitatingly withdraws his previous conclusion. AI AND El IN FRENCH BEFORE NASALS. 673 there, add demaine : compegne : essoigne : ensengne : castengne : Couloigne : Sessoigne : grifaingne : Espengne : vergoigne : com- pengne : gaaigne : entente : jenne ( =jeune ) : essoigne : quaran¬ tine : besongne : demouraine. The earliest texts ascribed by Suchier, Altfrz. Gram., pp. 2—3, to this region cannot enter into the argument. The Vie de Saint Thomas has no rimes in point, and Marie de France, as we have seen, scarcely has either syllable at the end of the line. The still earlier texts, as the Voyage Chari, and the Couron. Louis, show only aigne in assonance with a ; cp. Charlemaignes : compaignes : France : remaignet : plaigne, Voy. Chari. 783, Charlemaigne :Alemaigne : Bretaigne : reiames : France, Couron. Louis 10, a fact to be expected from the age of these poems. That neither should contain an assonance showing eigne : e is probably to be explained in a similar way as the relative proportion of aigne and eigne rimes in the later texts. If we now look at the texts cited above, pp. 656 ff., in which evidence for the value of en = ain and ein can be found, we shall be able to observe some rather definite criteria for our problem in the case of some of them, while we shall find a great deal of obscurity in the case of others. Maistre Elie in his Art d’Amour rimes aigne:eigne and *prendeam ; cp. remaigne : ensaigne 856, compaigne : preigne 370, gre- vaigne : peine 737, ensaing : desdaing 785. The author of the Escoufle joins saigne (signat) : montaigne 5089, plaigne (planea) : prengne 1125, daigne iprengne 7839, compaigne : aviegne 5277, remaingne : aviengne 1675, remaigne : raigne (regnum) 2221 , Loheraigne : raigne 5477, i. e., aigne, eigne, cgne, iegne and regnum rime in such a way that the conclu¬ sion is obligatory that all were pronounced alike either ahe or ehe. Now in this same text ain and ein are probably en, cp. above, p. 655, and since losenge, which rimes with vos ain ge 2877, is coupled with aveigne 7459, we are justified 674 JOHN E. MATZKE. in accepting ene as the pronunciation of the author. Guil¬ laume de Dole , which furnishes the earliest example of the fusion of ain with oin rimes plaigne : Alemaigne 3771, 5562, enmainent : remaignent 4085, and Champaigne : apregne 5, regne : graine 7, regne : deerraine 4134. The remaining rimes in point have a as tonic vowel, as 11. 972, 1084, 1654, 2560, 3068, 3542, or e as 3582. If aine is correctly de¬ termined as eine, then regne must be rene or rene and aigne was pronounced ene as is proved by the filiation of regne : aine : eine : aigne. The R. Rose also joined ain and oin. Here we find aigne : eigne (Bretaigne : enseigne , 1-39), aigne : iegne (com- paigne : tiengne , i-9), eigne : iegne (enseigne : tiengne , I— 68), aigne : *prendeam (refr aigne : sorpreigne , i—101, prengne : chastengne , H—215), eigne : *prendeam (preigne : feingne, I- 321, prengne ifaingne , n-129). Since saine rimes with essoine , i—73, and poine with moine (monachus), i— 100, we may accept vaine : raine (regnum), i-l 5 as meaning vene : rene , and we may infer that aigne and eigne were pro¬ nounced ene. In support of this conclusion we may cite also lointaignes : taignes (teneas), i-59, lointaingne : tiengne , II- 216, and Loheregne : regne , I—25. The Miracles de Nostre Dame show maines : enseignes , xii— 580, montaigne : ensaingne, xx-449, Bretaingne : re- teingne, xvn-883, deigne : veigne , ix-20, deigne : esconveigne, xiii- 408, viengne : enseigne , xxi—535, xxii-1397, aprengne : tiengne , yi-392, appartiengne : mesprengne, viii- 1138. One rime, compaigne : espargne, m-1079, seems indeed to speak for the pronunciation ane, but Charles d’ Orleans has espergne : preigne : enseigne : apreigne, II-85 showing the correct form. For similar evidence cp. espergne : Auverne , B. Chron. 5039, espergne : cerne , ibid. 16258, taverne : espergne , R. Rose, 1-168, esparne : superne, Besant Dieu 3167, and even as late as Marot we can find espergne : Auvergne , n-14. AI AND El IN FRENCH BEFORE NASALS. 675 The remaining texts 1 of this list present the following examples: Eustache Deschamps — paine : aviengne : empraigne : aviengne, I-XVI, montaigne: souspraingne, I—lxxxii, aviengne : plaingne : demaine : maine and empraingne : souvieingne : praingne : incertaine, I—cn, aviengne : Bretaingne : souviengne : reprangne, I—CLVii, foraine : Bretaigne, n— cxcii, reviengne : repraingne : maintiengne : enseigne, ii-ccxlviii, layne : apprengne : ensaigne : craingne : aviengne, II—CCCY. Guillaume Alexis—Espaigne : peigne, Le a b c des Doubles 485, attaine (==atteigne) : hayne , Basse Temps de tout Homme et de toute Femme 2897, montaigne : enseigne , ibid. 3535, preigne : appartiengne , ibid. 3703, daigne : baigne, ibid. 4323. Villon—Auvergne : Charlemaigne, Grant Test. 382, Bre¬ taigne : enseigne : tiengne , ibid. 1629, paine: attaine (= atteigne ), Poes. Div. 190, douzaine : Estienne : paine : sepmaine , Grant Test. 1913, Boyne (= rogne) : paine, ibid. 1151, Roynes : regnes : Renes : estrenes, ibid. 414. Marot —Only contraignent : preignent, 1-146, and daigne : enseigne\preigne : appreigne, n-47, Epigram cxxn, Espaigne : baigne, 1-259, 345, montaigne : baigne, 1-470, 485, n-165, baigne : gaigne, II-152 show the older tradition. Most of the rimes in point agree with the modern pronunciation as 1 Christine de Pisan’s rimes are very unsatisfactory in this connection. The following list includes all those of interest, and no conclusions can be based upon them. The volumes examined contain no eigne rimes what¬ ever, and no fusion of aigne : eigne. All rimes in point are exceedingly rare ; cp. Alemagne : remagne , Livre du Due des Vrais Amants 717, remaigne : Alemaigne , ibid. 1709, mahagnent : empregnent , Epistre au Dieu d } Amours 645, prengne : Bretaigne : compaigne : Alemaigne, Debat de Deux Amans 1552. Nyrop, Gram. Hist., i, p. 196, cites Bretaigne: empreigne, Cheminde l. Estude 3695. I may add here, because of similar interest and limited scope, enseing : repreing, Guiot de Provins, Bible, 1440, praingne : Champaigne, Crieries de Paris, 39, souvienne : preingne, Mist, du Viel Testament 1168, advienne : preingne, ibid. 3338. 676 JOHN E. MATZKE. campaigne : Espaigne , 1-63, and 1-117, 153, 195, 197, 232, 434 and enseigne : enseigne : contraigne, 1-82 and 125, 178. The material presented in the preceding pages lacks without question here and there the precision that one would like to see in an argument of this kind. However, no other criteria are available. The answer to this vexing question must be sought in the rimes, and from the nature of things their meaning must be unraveled with care, and certainly without prejudice. This we have endeavored to do. We have shown that in the Norman dialect aigne and eigne must have had the value of ene. Passing into the Francian, we have found evidence of the same pronunciation through the mingling of these syllables with oigne , while at the same time they rimed with iegne and *prendeam. Then we have seen oigne disappearing from this group, but we have noticed the others holding together until Marot’s time. The rimes cited from Eustache Deschamps, Guillaume Alexis, and Villon show aine and eine mixed with them besides, and we have seen these latter in a previous chapter definitely used with the value of ene. The conclusion which is forced upon us is inevitable. From the end of the 12th century until the time of Marot aigne and eigne in the Francian dialect both had the identical value of ene. There now remains the problem how the modern readjustment of these syllables into ane and gne is to be explained. V. Before taking up the consideration of this question we may study the history of aigne and eigne in the region where ain and ein had the value of an. Engelmann, l. c., pp. 22— 23, cites a few assonances showing the value of ane for both syllables. These are lances : ensaignes : grifaigne : France , AI AND El IN FRENCH BEFORE NASALS. 677 Charrois de Nimes 975 ; estrange : demande : regne : France : Alemaigne, Prise d’Orenge 179. We may add porpensent: Losane : Bretaigne : prendre : lance , Aiol, ccxxii; descendre : ensengne : Franche : Losane : Charlemaine ipendre :feme, ibid . ccxxv; chatainne : demande : mainnes : France , Amt's et Amiles 517 ; chatainnes : Charlemainne : estraingnes : entrent y ibid. 2042 ; demorance : montaingne : lance y Jourdains de Blaivies 1646. The rimed texts present no similar evidence. The rime bangne : espargne , Jacques de Baisieux, Trois Chev. 247 is of interest mainly on account of the orthography of the first word, but the rime itself is not entirely above suspicion since espargne might be a graphic variant for espergne, though this is scarcely probable in this dialect. Demanois : espaignois , Chev. as deus espees 2767, is also striking for the same reason. The second word is espanois < hispanensis, and the sylla¬ ble an could be written aign only by a scribe for whom aign meant ah. Nor will the words with which aigne and eigne are coupled in rime serve to throw light on the problem. We find here, as before, iegne and *prendeam. Only regnum seems absent and limited to rimes with aine and eine. Not all the texts, however, join all these syllables. All four, aigne : eigne : iegne : *prendeam, are found in Ille et Galeron , Guillaume de Palerne , Richars li Biaus and Philippe de Beau- manoir; aigne : eigne : *prendeam rime in the Chevalier as deus Espees and the Roman de Galeran ; and aigne : eigne only are coupled in Durmart , Robert le Liable and Froissart's Meliador. I do not add examples here to illustrate the nature of these rimes. A few specimens of each type could give no idea of the actual exclusion of the others and their general appearance will become sufficiently clear from the citations given for another purpose below. It follows that we have the same groups of rime words to 678 JOHN E. MATZKE. deal with as before, and the question of their pronunciation cannot be solved with the evidence which they contain. The only remaining avenue of approach is the orthography. We are aware, of course, of the care that must be exercised in basing an argument on such data. Yet certain ortho¬ graphic habits can be observed in these texts which are incompatible with the pronunciation ene. 1. Eigne, iegne , and *prendeam are constantly written aigne. Cf. entresaigne : baigne, Ille et Galeron , 568, Bre- taigne: taigne (= tienne ), ibid. 305, ouvragne: vaigne (= vienne ), Baudoin de Conde, xxi—503, empraignent: espaignent, Guill. de Palerne 9199, convaigne: compaigne, ibid. 5201, con- paigne : ensaigne, Robert le Diable 3115, chaigne (=cingat) : deschaigne, Chev. as deus Esp. 781, Bretaigne : praigne, Gale- ran 1620, pr aigne : remaigne, Manehine 449, vaigne (— vienne ) : remaigne, ibid. 2069. A complete list of the available ex¬ amples would serve no purpose, for it would fail to give an idea of the proportion of aigne outside of the rime in comparison with the other ways of spelling the syllable, but the constant recurrence of the aigne form in these texts com¬ pared with its more restricted employment in the ene group, gives it nevertheless the force of a valuable argument at least for the speech of the scribes who copied the manuscripts. 2. The parasitic i is often and in some texts quite regu¬ larly omitted. Cf. Bretagne : adagne, Ille et Galeron 1683, remagne : Alemagne, ibid. 2663, dagne: ensagne, ibid. 382, desdang : faing, ibid. 5407, engagne : gaagne, Rich, li Biaus 4189, Espagne : compagne , ibid. 4899, Bretagne : estragne Meliador 2776. 3. Eigne and *prendeam are written egne and engne. Cf. ensengne : estraigne, Rich, li Biaus, 3887, pregne : com¬ paigne, Ille et Galeron 4803, adengne : daingne, Jean de Conde, xxxv-13, ensengne : mehangne, Jacques de Baisieux, V lettres de Marie 195, plaingne : pr engne, Guill. de. PaL AI AND El IN FRENCH BEFORE NASALS. 679 6697, ensenges: montaignes, Rob. le Diablc 637, 1809, en- senges : estranges, ibid. 899, ensegne : estraigne, ibid. 2853, ensengne : Bretagne, Meliador 6799, ensengne : acompagne, ibid. 7125. 4. These syllables rime with aine = ane. Cf. plaine: ensenge, Rob. le Diable 2347, fontaine : plaine (planea), ibid. 2351, Charlemainne : painne, Rich, li Biaus 15, vaingne: fontainne, ibid. 1253, plainnes (plenas) : compaingnes, ibid. 4529, amainne : Espaingne, ibid. 4769, maine : remaigne, Chev. as deus Esp. 2293, deschaine : remaigne, ibid. 1461. 5. Extraneus appears as estrange and estragne. Cf. estranges: blanges (verbal noun from blangier ), Rob. le Diable 4383, ensengnes : estranges, ibid. 899, 2123, ensegne: estraigne, ibid. 2853, ensengne: estraigne, Rich, li Biaus 3887, com- paigne : estrange, Durmart 3123, Montaigne : estraigne, ibid. 5353. Such rimes are incompatible with the pronunciation ene. While these points could not in themselves serve as final arguments for the pronunciation, yet taken together they contain a certain cumulative force, which points to the value ane for the texts in question, but the nature of the material on hand does not permit us to draw more definite conclusions. We refrain from undertaking to determine the geographi¬ cal limits of this phenomenon. The material for this investigation is here even less satisfactory than in the case of ain and ein. We may admit that it included Champagne. Chrestien de Troie rimes aigne : eigne : iegne : *prendeam. Cp. Bretaigne : ansaingne, Yvain 1, remaingne : praingne, Cliges 2553, ansaigne : apraingne, Yvain 4957, desdaing : praing, Erec 4025. For aigne : iegne there exists only the isolated rime ptaingne : vaingne, Cliges 3077. Usually iegne written aingne is kept distinct. Gace Brul6 rimes only aigne : eigne : *prendeam. Cp. plaigne : ensaigne, viii-18, apraigne : remaigne, viii-26, adaigne : apraigne, xvii-32. 11 680 JOHN E. MATZKE. However, Burgundy, as in the case of ain and ein , did not share in this development. The Lyon Yzopet has oigne for eigne , while aigne has remained; cp. compaigne: deplaigne 503, aplaigne : acompaigne 859, complaignent : raignent (regnant) 1121, and ensoigne : proigne (*prendeat) 319, 1417, besoigne: ensoigne 529, 3007, doigne (dignat) : ver- goigne 999, ensoigne :cyoigne 1165, enproigne : ensoigne 3387, besoigne : doigne 3505. There is no example of fusion of aigne : eigne > oigne. The material, collected by Gorlich, Burgund. Dial., pp. 34 and 63, shows that these examples represent the general habit of this region. Aigne is written aine, ayne, eigne, egne, enne, very rarely oigne, and eigne be¬ comes usually oigne, though there is occasional interchange with the former group in the orthography. The Estoire de La Guerre Sainte, which on the basis of its development of ain and ein we were inclined to place with R. Troie and B. Chron., agrees with this same group of texts for aigne and eigne. As in the R. Rou the majority of rimes in point show aigne; cp. 99, 347, 995, etc., in all 42 rimes. There is one pure eigne rime, 1. 6225, regne rimes with Charlemaine 8479, and with cheveitaigne 8607, and we must cite besides the isolated empraine (impregnat) : enpraine (*imprendeat) 1. 5. With this enumeration the evidence for the pronunciation of our syllables in this text is exhausted, and we believe it should be joined to the rimes of the Nor¬ man texts in general, and interpreted together with them. VI. We may now endeavor to explain the difference in the modern pronunciation of montagne, chdtaigne, enseigne and the like. From the preceding pages it has become evident that not one of the long list of texts, Marot included, makes use of the syllables aigne and eigne with their modern value. AI AND El IN FRENCH BEFORE NASALS. 681 Both rime without distinction during the whole of the Old French period and were pronounced either ene or ane, depend¬ ing upon the dialect to which the text belongs. Bearing this fact in mind, we may examine the history of a and e + n as outlined in the Grammars. According to Behrens 1 a parasitic i developed before n when it was final or followed by a consonant, but not when it stood in medial position, so that we shouldMiave compaing, bciing, enseint, but compagne, bagne, ensegne. The i, which so constantly appears also in the second group of words, is looked upon as merely graphic. It is evident that this explanation is intended to suit the modern form of the words in question ; for if it is correct, bain, gain, refrain are regular as well as gagner, Bretagne, Allemagne, montagne, while baigner, plaignons, craignons, and the like can find a ready explanation on the basis of analogy. However, it is overlooked that then araigne, musaraigne and chdtaigne are not provided for, and that the rule does not explain the constant union of a -j- n and e -j- n in rime in Old French. The further assumed difference between a or e -f- final n and a or e -f medial n it is difficult to test, since the i before n is usually written in either position. Where it is absent before the medial n, it may also disappear before the same sound in final position, as desdang : faing, Ille et Galeron 5407, song : beson, ibid. 5780, though it is true that such examples are rather rare, and ordinarily the i is written before final n, even where it is absent in the medial position. Suchier’s explanation 2 differs fundamentally from the preceding. He maintains that in the Francian dialect n developed a parasitic i before it when it preceded the accent as in plaignons, joignons, Bourguignon. The diphthong 1 Schwan-Behrens, Altfrz. Gram., § 203. 2 Altfrz. Gram., p. 72. 682 JOHN E. MATZKE. created in this way could enter the tonic syllable (baigne), and this happened particularly often in the case of ei. This new diphthong ei then shared in the common development of ei > di toward the middle of the 12th century. As earliest example of this development he cites ensaignet , Mont. Ps. 17-37. The evidence presented in this study precludes in my opinion the possibility of accepting this explanation of the problem. There is no proof that in the Norman and Anglo- Norman dialect ein and eign developed in the direction of din and din; on the contrary, all the evidence available points strongly to the conclusion that ain and aign became ein and ein. If this be so, ensaignet in the Mont. Ps. can only be the earliest evidence of the graphic confusion caused by the coincidence of the two sounds. Furthermore one is tempted to ask how Suchier would prove that accented a and e followed by n become an and en, while pretonic a and e -f- n developed into din and ein. Certainly the ortho¬ graphy of the Mont. Ps ., the Carnb. Ps., and the Q. L. D. P. permits of no such conclusion, for in these texts n appears as ign or gn in all positions, regardless of the accent, and the rimed texts give no evidence for the unaccented syllables. If eigne toward the middle of the 12th century became aigne, it would have to be shown in the next place when and where the modern readjustment of the pronunciation of these syllables was effected, and how eigne = afie changed back to ene again, while aigne = dne retained its Old French value. We have seen that no evidence of such a division is visible in the Old French texts. As far as the rimes are concerned, both syllables are identical in all words of this group from the end of the 12th century until Marot’s time. The pronuncia¬ tion of both developed in a uniform direction during the whole of this period, and the conclusion which we have reached is the only one justified by the evidence before us. AI AND El IN FRENCH BEFORE NASALS. 683 This value of ene for both aigne and eigne was slow to disappear. Malherbe still rimes compagne: dedaigne (Larmes de Saint-Pierre)) Tabouret 1 in 1584 states that of the words in eigne ( la pluspart peuvent rimer avec aigne’, and Lanoue 2 in 1595 reiterates ‘ ces deux terminaisons n’ont qu’une pronon- tiation’ Yet there is evidence that the modern pronunciation was becoming established during this same period. Meigret 3 writes accompany, anyao (agneau), montafies, Qharlemane, Champane, Hespanol, Montanart, Hespane, accompanant, only once accompaine and plenet {plaignent ) for words with Latin a, and dedenans, crenans, crenet (craignent), crenez (craignez), penons {peignons ), ensene {enseigne ), ensene for words with Latin e. Baif 4 has gai)era {gagnera ), montage, garner, honpai)’ {compagne), akonpai]eront but eqecos (agneaux), %i)el%s {agnelets) and beiqera {baignera), deir)a {daigna), anseii)ement {enseigne ment), answer {enseigner), anseii)e, deir^e {daigne), and many other examples equally regular from the modern point of view. Among the various influences which must have been potent in establishing the modern pronunciation we may mention in the first place that of the orthography. Though aigne was pronounced ene, yet the orthography to a very large extent had retained aigne, though it is true, as has been shown, that eigne is very frequently used in its place. The written form aigne might cause a certain hesitation as to the proper pronunciation of the syllable. The i might have been looked upon as belonging to gn, and just as aiile was pronounced al'e so it might have been felt that aigne should be pronounced ane. That this sort of reasoning actually did 1 Thurot, l. c., I, p. 330. 2 Thurot, ibid. 3 Meigret, Le Trette de la Grammere Frangoeze, published by Forster, Heilbronn, 1888. 4 Jean Antoine de Bail’s Psaultier, published by Groth, Heilbronn, 1888. 684 JOHN E. MATZKE. take place seems to follow from the statement of Palliot (1608) cited by Thurot, 1. c., I, p. 330. ‘ Je sgay bien qu’il y a des diphthongues quil vauldroit mieux laisser et n’en retenir que la premiere voyelle plnstost que de les 6crire ny proferer : tant il s ? y donne un mauuois air par des mal-embouchez et mauplaisants prononceurs. Comme celle d ’ai en Bretaigne , montaigne , Champaigne , aigneau: ou ils semblent avoir le mords trops serr6 et se gourmer par trop, a en faire la petite bouche, les pronongantz en ei, eigneau, Breteigne, monteigne , Champeigne.’ In the next place it is certain that during the period of reconstruction, when the Old French changed to the modern language, the larger portion of the words in aigne fell into disuse. Only a fraction of the Old French rime words in aigne : eigne has passed into the modern vocabulary. Of those which remained a certain number readily suggested the Latin word from which they derived, as Allemagne , Charle¬ magne , Romagne, Espagne , Bretagne . The Latin influence which pervaded the language at that time could without difficulty re-establish the original vowel in these words. Other words with similar ending were introduced at that time from the Italian or the Spanish, as campagne < Italian campagna , pagne < Spanish pano. Both words are inter¬ esting for their form. The former occurs first in Marot, cp. Diet. G&n. s. v., the first instance of the latter is found in the correspondence of the Pere Nacquard in 1650, cp. ibid. s. v. In both cases the influence of the traditional orthography is so strong that they are written campaigne and paigne. Still other words were influenced by their Italian, Spanish, or Provengal cognates. So quoquaigne 1 becomes cocagne 1 The word is quite rare in Old French literature. Godefroy cites it, ii, p. 164, from Aimery de Narbonne (: remaigne ) and the Enf. Ogier (: engaigne). Two other references can be found in vol. ix, Suppl. s. v., both in rime e ^ AI AND El IN FRENCH BEFORE NASALS. 685 under the influence of Italian cuccagna; champaigne changes to champagne under the influence of It. campagna, Sp. cam- pana, Prov. campanha; compaigne is affected by It. com- pagna, Sp. compaha, Prov. companha in spite of the different gender, and Old French gaigner loses its i through association with It. guadagnare, in spite of the noun gain ; and from these simple words the new pronunciation finds its way readily into the derivatives, so that we have compagnie, com- pagnon, compagnard and the like. Similarly montaigne becomes montagne under the influence of It. montagna, Sp. montana. The old orthography has here lived on in the proper name Montaigne , and thereby caused the continuance of the older pronunciation. Note also the proper name Cham¬ paigne in rime with peigne, Cyrano de Bergerac, I, scene 2, cited by Nyrop, Gram. Hist., I, p. 196. As far as I know, this list exhausts the modern words 1 in -agne, with the exception of the learned name Ascagne, (the Eneas, 1. 773, has Ascanius) and the geographical names Cerdagne and Mortagne. Other words withstood this influence and retained their original and regular form. These are in the first place the nouns araigne, musaraigne, and chdtaigne. The reasons for this isolation remain obscure. It. aragna and castagna, Sp. arana and castana might have exercised similar influence here as in the preceding list. That it was at work is shown by the rime compagne : aragne in La Fontaine, Fables, m, 8. The verbs finally have retained their original pronuncia¬ tion. Plaignons is determined by the forms of the paradigm without h, as plaindre; in others the orthography has been with aigne. The word occurs also in Joufroi, cp. Langlois, La Societe fran- gaise au XIII e si&cle, Paris, 1904, p. 42, but the text being beyond my reach I am not able to verify the reference. 1 Bagne is a comparatively recent importation from the Italian, cp. Did. Gen. s. v. 686 JOHN E. MATZKE. changed under the influence of the older confusion, as in atteindre, atteignons, or enfr eindre, O. Fr. enfraindre, and fraindre. For saigner and its derivatives saignee, saignant, saignement no apparent reason suggests itself, but baigner retains its old vowel because of bain. The words with e -f- n on the other hand have not varied in their pronunciation. The causes which influenced the change of aigne to ane would have served to strengthen the pronunciation of gne for eigne. Hence we have enseigne, teigne, peigne , and all the verbs in -eindre , as astreindre, Streindre, restreindre, teindre , feindre f peindre, and eteindre, which in certain forms of the paradigm contain the syllable gn. Some words in this group show am or aign through orthographic confusion, as Sardaigne (Sardinia) daigner (dignare), dedaigner, and contraindre-contraignons. The whole conjugation of craindre is due to analogy with the aindre verbs, and similar influence of the eindre class has determined the orthography and conjugation of geindre, empreindre, and epreindre. John E. Matzke.