and Themaleo with the somewhat doubtful two; Ekko and Erro (= Hero) with woo (= woe). Yet, strange to say, it is upon these foreign words, yielding such inconsistent evidence, that Mr. Ellis chiefly relies. His view is mainly, if not even exclusively, based on the single rhyme of schoo with principio (p. 266) !

56 Recapitulation

on 'O' words.

The facts then as to these 'o' words may be

briefly re-stated thus: There are words with similar written terminations which clearly pair off into two classes, which in Chaucer refuse to rhyme with each other; of these classes the vowels are fixed by universal English tradition as (0) or some modification of that sound for the first, and (u) for the second; this tradition being confirmed by French and Italian tradition for the first, and by German and Dutch tradition for the second. The hypothesis that they were sounded with (0) and (u) respectively satisfies all the conditions of the problem, save only the very few exceptions above noted. It has been suggested that those two os are simply the Italian ò and 6. But there is this grave objection, that those two can rhyme; in Dante they rhyme habitually; while, as we have seen, in Chaucer and other English poets the two classes are kept distinct. 57 Could, would,

should.

This seems to be the proper place to remark on the forms could, would, and should.

There

can be little doubt that the similarity of grammatical use of these three words has affected the spelling of all three, and exceptionally the pronunciation of one of them. The pronunciation of would and should, except that in quite modern times we drop the 7,* has been the same for at least three centuries, the vowel being (u) or (uu). Thus Gil, 1621, writes shüld, with i= (uu); and Hart, 1569, writes uld and shuld (with u Gil's i)—or (through carelessness or misprint?) uld and shuld; and Chaucer's forms and those of other E. E. writers are wolde, wold, woolde, sholde, scholde, schold, shulde, ssolde: only these, I believe; at any rate, none with ou. And the A.S. forms were wolde and

* Cooper, 1685, condemns woudst and widst, for wouldst, as belonging to the "barbara dialectus."

58

sceolde, never with ú. ancient orthography vowel was not (ǝu).

So far therefore as the testimony of and of these orthoepists goes, the Could on the other hand is in A. S. cúde, and in Chaucer and his contemporaries coude, coupe, cowde (or with k or th); and it has these forms only, the vowel being the same as in dún, down, hús, house, &c. Nowa-days we sound all three words alike. We may therefore not unreasonably infer that the o in the A.S. wólde and sceólde had the accent (though I do not find it so written in the dictionaries*), and that these words have always in the "Englisce spræc" had the sound of (uu) or later (u), and yet, as to their written form, they borrowed a u from coude, which nevertheless failed to affect their sound: that coude on the other hand, in sound, but not in writing, exchanged its diphthongal (ou) for the (uu) or (u) of its comrade auxiliaries; as it also, in too slavish imitation, assumed the /, which was radical to them, but to which it had no claim. It may be added that this / in could is sometimes sounded in the West of England; and, curiously enough, Hart also sounded it. At least we find the word, even in his phonetic writing, as kuld, or kuld, or once (by mistake, no doubt) kould.

Now, as I have above assigned at least plausible reasons for believing that the of A.S. and Chaucer's long i (I am speaking of the written symbols now) were sounded (əi), and not (ii), I shall not be expected to accept Mr. Ellis's view as to ai and ei, that these were both sounded (ai). For if so, we could not but have had i and ei or ai at least occasionally rhyming. There is not an instance of the kind in Chaucer, nor have I noticed one in any other poetry, always excepting the two words die and dry, which had also the other but rarer forms dey and dreye. The latter of these I cannot account for: the former is simply

If

* Grein is the best authority as to accents, yet not always correct. he is right in refusing the accent to wolde and secolde, as analogous forms to the Ger. wollte and sollte, then these are the only A.S. words I have met with, which have an unaccented o that becomes (uu) in the later stages of the language.

59

ded as in modern

French.

the O.N. deyja, ek dey. What sound then is represented The symbols or what sounds-by ai and ei (or ay and ey)? ai and ei soun- These symbols are at present pronounced alike in French; and that they were pronounced alike in early English (as Mr. Ellis admits) the rhymes of Chaucer and the frequent interchange of these digraphs in writing one and the same word, seem conclusively to prove. And if again we appeal to tradition, the traditional sound in both countries is, with certain exceptions, (ee) or (ee), as in vain, vein; faites, veine. But let us as before examine a few of the words themselves; for, as I have remarked of previous classes of words, when a few are fixed, the rhymes of Chaucer and other E. E. poets will show that these few draw a multitude of others with them.

words.

Now the verb dey (= die), as has just been pointed out, is the O.N. deyja (deei Ja). May (= maid) is the O.N. mey. Obey is from obedire through obéir. Journey, valley, chimney, Proof from the are the Fr. journée, vallée, cheminée, which have O.N. and O. Fr. had that termination (as written) unchanged for originals of certain of these at least the last six centuries. The words lay (= law) and fey (= faith) are in O.Fr. leis and fei (feis) or feid or fé, the former of which-to glance at the derivation, a point which Mr. Ellis far too commonly overlooks-is evidently the Latin legi-s with the guttural dropped, and the latter a syncopated form of fidei; and these in the Chanson de Roland are in assonance with reis (=king, from regi-s), fedeil (from fidelis), mei (Lat. mei), meis (Lat. mensis), creit (Lat. credit), &c. Array (vb.) was in O.Fr. aréer. Moneye was moneie (Burguy), or monnoie, and oi we know was pronounced (oee) or (wee), with no (a) in it, Nobleye is noblee in Kelham. From the noun preye is the verb preer in the Conquest of Ireland. And these twelve words draw with them way, away, alway, they, say, day, lay, (vb.), biwrey, may, May, jay, play, abbaye, tweye, and many besides; all of which indiscriminately and constantly rhyme with one another. In the Ormulum, which, according to Dr. Morris, exhibits the Lincolnshire dialect of the early part of the thirteenth century, we find a distinction be

tween da33 and ma33, and þe33 and we33e, of which there is no trace in Chaucer. (See § 65.)

60 The endings If then these -ay words are fixed, the pre(eed), (eeth), and terites of the verbs among them, deyed, pleide,

61

(een), thus ascer

evidence in many

tained; with fur- affrayed, preyde, &c., will fix other words rhymther independent ing with these, such as maid (O.Du. meeghd) and brayde. And in like manner we are taught how to sound fayth by deyth, seyth, layth.

cases.

Next, words in (een). The infinitive of sey (see) is of course seyn (seen). Sweyn, as I have above observed, is the O.N. sveinn, which is nearly (sweeidn). Tweyn and reyne (rain) are certainly contracted from twegen and regen, the twe33enn and re33n of the Orm., in neither of which is there any vestige of an (a) sound; nor is there any original (a) in atteyne, from attingo, distreyne, from distringo, desdeyne, from disdignor (= dedignor), peyne from pœna, Mawdeleyne, from Magdalene, feyne, from fingo, veyne, from vena. And these enable us to fix vein (adj.), certeyn, Spayne, soverayn, agayn, brayn, greyne, cheyne, compleyne, &c.; for all these indiscriminately and constantly rhyme with the above and with one another. And as to certaine and vilaine, these are found in the old French song by Le Vidame de Chartres rhyming with maine mène, from mener, Du. mennen, where there is clearly no (a).

Evidence

and the asso

poetry.

=

Now in Old French we find, as Mr. Payne from Meigret, has pointed out, words occasionally spelt with nances of O. Fr. e which more commonly have ai, as lesser, reson, treter, cler, set (= sait), &c.* And if we turn to Meigret we find that many of the words usually spelt with ai are by him phonetically spelt with e: grammére, jamés, més, &c., and in particular fés, fét, fére, fézons, &c. Moreover in the Chanson de Roland this verb faire and its compounds are commonly assonant with e words-perte, perdet, nuveles, apelet, tere, bele, &c. In like manner repaire and esclairet are assonant with deserte, herberges, and other e

* See also the numerous rhymes of ai with e words in French poetry, which Mr. Payne has collected: Transactions of the Philological Society for 1868-9, p. 387 sq.

words; and heir (= Eng. heir) with reis, mei, &c. And these words in Old French songs rhyme with de bon aire and paire (=pair). And with debonaire, paire, repeyre, heir, faire (s.) we have rhyming in Chaucer faire (adj.), eir, dispeir, &c.

But as Meigret fixes for his age the sound of jamais, and already in the Chanson de Roland jamais is in assonance with desert, Samuel, apres, and other e words; jamais, together with fais and fait (see Wright's Political Poems, i. 302) fixes Caleys, which in turn is found rhyming with paleys and deys, and these again with burgeys, harneys, &c. And in the Ch. de Rol. deiz and palefreiz are associated with soleilz, aveir, franceis (It. francese), dreit, &c., and never with any (a) words.

62 Three excep

tions in French :

=

Do I mean then to deny that the written Palsgrave on ai. ai was ever (ai)? Certainly not. There are for instance those three words mentioned by Meigret, which modern fashion, consistently with their derivation too, would or does write with puncta diæresis—aÿmant, adamant or loadstone (now aimant), ajdant* (now aidant), and haïr; to which païen, païs (now pays pai-is), traïr, traïson, traïtre (now traître) may be added, the sound of the last of which is clear when we find it in assonance with olive, ocire, &c.; and the forms païs, traïr, &c., are capable of like proof, which rhyme elsewhere supports. Then again faire itself, though no doubt commonly fere, occurs once in the Ch. de Rol. (in accordance with its etymology also-facere) in assonance with (a) words— Carles, marche, message; and so repaire once with visage, esguardent, &c.; though in the Conquest of Ireland this verb has always ei. And we know by instances just mentioned (aimant, aidant, traître) that the (ai) can pass into (ee); of which too the O.Fr. gueter (now guetter) and the phrase aux aguets from O.N. gæta (gJaaita), to watch, is a proof, and ay = ever, from O.N. a (aai). And when Palsgrave says that “Ai in the frenche tong is sounded . . . a distinctly and the i shortly and confusely," one can have no difficulty in seeing that

* 'A Paris dans le peuple on dit souvent aïde. Littré, s.v.