the principles of alliteration are much less strict than in Old English. Alliteration is becoming less of a structural necessity and more of an ornament. This may be seen, in the first place, in the fact that unstressed words and prefixes may bear the alliteration, as in the following lines:

63 On hade bozt hym a borz, he sayde, by hys trawbe.

114 Ay þe best byfore and bryзtest atyred. Cf. bifore, 918, 978. 127 And rehayte rekenly þe riche and þe poveren. 197 Bot never get in no boke breved I herde.

Another indication that the alliteration is becoming mere ornament is the tendency to crowd as many alliterating words as possible in a single line. Lines with three alliterating words in the first half are very common, but many lines have even more. The author of Purity is especially fond of such superabundant alliteration, for example:

113 Wheþer þay wern worþy oper wers, wel wern þay stowed. 661 Penne sayde oure Syre þer he sete: 'Se, so Sare lazes.' 1681 His hert heldet unhole, he hoped non oper.

In Purity, as in most contemporary alliterative poems, double alliteration is not unusual. This is generally parallel-a a bb, as in

299 Sem soply þat on, þat oþer hy3t Cam.

Here, and in 345, 1304, 1573, 1622, it is essential in the structure of the verse, since these lines would otherwise have to be considered defective; but elsewhere it is added to the regular alliteration as additional ornamentation, as in

25 Me mynez on one amonge oþer, as Maþew recordez. 493 Myryly on a fayr morn, monyth be fyrst.

Transverse alliteration-a ba b-appears in

515 For I se wel þat hit is sothe þat alle mannez wyttez,

1571, 1727, 1807. This list does not include those lines where double alliteration is possible. No alliteration appears in the second half-line of 28, and in MS. 520, 745 (emended in text).

and possibly in 228, 327, 1618. Inclusive alliteration— a b b a―appears in

608 Hit is ebe to leve by be last ende,

and perhaps in 67, 735.

Most of the peculiarities of the poet in the sounds and combinations of sounds which are used for alliteration are shared by his contemporaries.1 Alliteration on the same vowel, which is avoided in Old English poetry, is frequently found in Purity, for example, 241, 277, 411, 713, 1006, 1035, 1301, 1470. Vowels alliterate freely with h before vowel, whether it be of Germanic or Romance origin. Likewise, no distinction is made between w and wh- whyte, for example, alliterating with wynnes and worschyp (1120). In general, sk, sp, st, as in Old English, do not alliterate with s; and this is also the case with sch and sm.2 Finally, the poet exhibits a particular fondness for alliterating certain groups, such as cl (634, 839, 858, 965, 1400) and str (307, 880, 1199, 1540). Sometimes two consecutive lines have the same alliteration, but the practice of the author of Morte Arthur of grouping from three to five lines together

1For the characteristics of the Gawain-poet in alliteration, see Trautmann, Über Verf., pp. 29-31; for further discussion of the alliteration, see Fischer, pp. 38-48, Thomas, pp. 59-64, and especially Schumacher's excellent dissertation on the alliteration of the whole ME. school.

2 Sp alliterates with itself without exception; sk 14 times with itself, but once irregularly with s (523); st 35 times with itself, and only once with s (999), since 995 may be considered double alliteration (Schumacher, p. 110). The curious fact may be noted here that in excused, 62, the sk alliterates, whereas in excuse, 70, the alliteration is with k; sp is always the alliterative sound in expounde (1058, 1492, 1565, 1606, 1729). Sch alliterates twice with s (58, 566) and once with sk (600); sm alliterates only with itself except at 566, and possibly 1019 (see note). Other peculiarities are the alliteration of z with s (1169), cf. Pat. 470, Gaw. 517; and ch with k? (464), but see Schumacher, p. 169.

by this means, is wholly unknown to the author of Purity and Patience.

VIII. DIALECT AND LANGUAGE1

It is difficult to determine precisely the dialect of this group of alliterative poems. The criterion of rhyme is not altogether satisfactory because of the paucity of rhymes in Gawain, and the license in the use of rhyme required by the metre of The Pearl.2 The peculiarities of vocabulary, the large number of Scandinavian loan-words, for example, which have led some editors to place the poems in Lancashire or Cheshire,3 are not sufficiently distinctive to fix the origin of the poem within such definite geographical limits. On the other hand, if these tests of the dialect of the original poem yield only vague results, it is certain that the dialectal traits of the manuscript as we now have it are overwhelmingly West Midland in character.

1

Among the characteristics that point to this dialect are

1A new investigation of the language of this group of poems is not attempted in this section. The studies of the language of these poems by Morris, Schwahn, Fick, and Knigge (see Bibliography), though old, are still valuable, and should be consulted for details. Only the chief dialectal characteristics and a few other peculiarities of the language, knowledge of which may save the reader trouble, are mentioned.

2 On the rhymes of The Pearl, see Fick, p. 8.

3 Bateson (p. xxxii), following Morris, attributes the poems to Lancashire, but the evidence for this on the basis of the vocabulary of modern dialects is entirely insufficient. Brunner (Archiv 132. 185) rightly criticizes this view, but he seems to me too skeptical in rejecting any conclusions about the original dialect of the poems. Wyld (Engl. Stud. 47. 47) objects to Lancashire on other grounds. These tests of dialect are based chiefly on the characteristics enumerated by Morsbach in his Mittelenglische Grammatik (Halle, 1896), pp. 15-7, esp. p. 15 n.; and Wyld in his Short History of English (London, 1914), pp. 122-3. The only characteristic generally

4

(1) the frequent representation of OE. eo by u, as in urþe, 150; brurdes, 1474; (2) the frequent use of o before nasals, as in honde, 174, 734 (as well as hande, 34, 155), and mon, 124, 183, etc. (as well as man, 51, 180, etc.); (3) the representation of OE. ỹ by u (ui, uy),1 as well as by y, as in fust, 1535, beside fyste, 1723; luper, 163, beside lyperly, 36; huyde, 915; kuy, 1259; (4) the participial ending -ande as in the North, in contrast to the -ende of East Midland; (5) the occurrence of the pronominal form pay (from Old Norse), together with the native forms her, hem, and the preservation of the feminine nom. ho; (6) the use of the curious form schyn, 1435, 1810, as pres. 3 pl. of schal.2 The evidence of the manuscript, to be sure, is not proof that the dialect of the original, as well as that of the scribe, was West Midland, but there is every likelihood that they were not very different. Morris declared3 that 'the uniformity and consistency of the grammatical forms is so entire, that there is no internal evidence of subsequent transcription into any other dialect than that in which they were originally written.' It would perhaps be safer to say that most of the dialectal characteristics of the manuscript are undoubtedly those of the poet himself. The evidence of the manuscript, together with that of the vocabulary and the rhymes of the other poems by the same author, make it almost certain that Purity was actually composed in the West Midland dialect.4

found in West Midland, and not usual in the poems of this group, is the occurrence of u for e in such endings as -us, -ud, for -es, -ed; but there are at least two examples of this-flemus (pres. 3 sg.), Pur. 31, and exorsismus (pl.), 1579.

1

1 That OE. y was often kept (written u, etc.) in West and Central Midland, has been shown by Wyld, Engl. Stud. 47. 1-58. On the basis of his investigation Wyld would assign the poems of this group to Derbyshire rather than Lancashire (p. 47).

2 Cf. Morris, Early Engl. Allit. Poems, p. xxvi.

Early Engl. Allit. Poems, p. viii.

It has been suggested by some students of the language, for

Certain Northern traits are generally assumed to be due to the fact that the poet was writing in the northern part of the West Midland area. Among Northern characteristics are (1) the ending of the pres. 2 and 3 sg. in -s, -es (-es); (2) the occasional ending -et in the preterite of weak verbs, as in bounet, 1398; tulket, 1414 (Schwahn, p. 18); (3) the appearance of contract forms like ma, 625; tatz, 735; bos, 687; (4) the representation of OE. hw by qu (as well as by Southern wh), as in quite, 1440; quo, 1650; and (5) the representation of OE. a by a (as well as by Southern o), as in halde, 652 (but holde once, 315).

Certain other peculiarities of the manuscript-phonological, orthographical, and morphological-may be noted here for the convenience of the reader.

I. Phonological and Orthographical.

(a) Vowels. ĕ frequently becomes i: kynned, 915, 1072; fyber, 530, 1026; rydelles, 969, but redles, 1197 (e here shortened from OE. a, ē). Still more frequently i appears as e: wekked, 855, beside wykked, 570; prevenest, 1571, beside pryvenest, 1639; þeder, 64, 461, beside þider, 45, 61. The spelling y (for OE., ON. ĭ, OF. i) is much commoner than i. OE. ūg, ōg, and āg (final) are all spelled either ow or o, the last two frequently ogh: bowe, 45, 67, but boz, 1551, 1750; innoghe, 669, but innoze, 808; wowes,

instance Fick (p. 9), that the present form of the poems shows traces of being a copy by a Southern scribe. But it is no longer necessary to assume this, since most of these characteristics are now known not to be peculiar to the Southern dialect. Thus the representation of OE. y by u is now known to be West and Central Midland as well as Southern (see above, p. 72); the retention of -iin the OE. weak verbs of the second class, as in lyvyes, wonyes, is not necessarily an indication of Southern dialect (see Schüddekopf's Sprache u. Dialekt William of Palerne, Erlangen, 1886, p. 104, and Boerner's Die Sprache Roberd Mannyngs of Brunne, Halle, 1904, pp. 218-9); and the participles in -ing which occur in The Pearl and Gawain are at this late period Midland as well as Southern (Kaluza, Engl. Gram. 2. 191).