while there are many more exceptions than those he specifies,* his rule applies without doubt to all words in which the ai is followed by ll (as even in modern French, e.g., travailler), or ge (in which modern French keeps the simple a). The word sage or saige, for instance-and Palsgrave directs a faint i to be inserted after the a in words in -age, even if it is not written-occurs in the Ch. de Rol. in the form saive (i.e. sapiens) in assonance with (a) words, marche, Carles, &c.; as also bataille, vaille, asaillet, occur only in assonance with (a) words.

63 No exceptions

64

in

Chaucer,

But is this ever the case in Chaucer? For

where ai words some time I imagined it might be so, that Chaurhyme with one cer's travayle and batayle would be sounded

another without

distinction. like the French words, but veyl, sayle, ayle (vb.) as at present; but having run through the Cant. Ta. once more expressly to examine the rhymes with this termination, I am forced to a different conclusion; for I find the words with a radical (a) are twenty-four in number-vitaille (= victualia), hayl (= hagel), aveyle (from valere), &c.; those with a radical (e) or (i) are six-veyle (from velum), sayle (from segel), mervaile (from mirabile), and chamayle (Ch. de Rol., cameil, from camelus), &c. But these six words rhyme with one another even less frequently than with the others: six times with one another, ten times with the former class. It is therefore impossible that there can be a distinction in the pronunciation of these classes.

Was -ail then sounded with (ee) or with (ai)? With (ee), I reply. First, the analogy of Chaucer's spelling of the words in ai, aid, -aith, -air, -ais, points to (ee). Secondly, in six of these words the etymology will fully account for the sound of (e), not for that of (a). Thirdly, of two of the same words (cunseill and merveill) the as*He specifies futures in -ray, as sounded like -rey, but also in the Introduction, p. xviii., he says: "in stede of ai, they sounde most commenly ei." And so Meigret gives us ey, sey, eymer, eyt, vrey, &c. And we must not forget that a multitude of words now written with ai had oi (oee) in Palsgrave's time -anglois, francois, monnoie, poyement, &c., and especially all imperfect and conditional tenses.

show the sound to

have been (ee).

sonances of the Ch. de Rol. are solely with (e) words. Etymology, as- Fourthly, these same assonances show that one sonances, &c., all of the others (like faire mentioned just above) was wavering in its sound even in French: ventaille (Chaucer's adventayle) is associated with hastet, vasselage, &c., and elsewhere with sele, perdre, &c. Fifthly, I find apareillez, so spelt in the same poem. Lastly, if the Lancelot of the Laik may be quoted as an authority, I find there the forms batell and travell-clearly an (e) and not an (ai) sound—and bataill rhyming with the adjective haill, which is the O. N. heill (Heeidl).

65 Palsgrave again

on ai.

Time forbids me to examine the rest of the ai terminations, -aim, -eint, eise, ait, -eive, all of which, either from analogy alone, or for that and other reasons, I believe to have been pronounced with (ee); but one inquiry must not be omitted. What of Palsgrave's assertion that rayne, payne, fayne, disdayne, were pronounced like the French ai as opposed to ei, namely, the "a distinctly and the i shortly and confusely"? Why, I take his words to exhibit a simply local or temporary fashion, which did not take a firm hold even on himself; for he in his vocabulary writes peyne, as he also gives both cheyne and chayne. But from whatever cause, and to whatever extent Palsgrave distinguished ai from ei in English, such distinction was utterly unknown to Chaucer.

66 Many of these

ai words may

Obviously, that ai in Chaucer's time was the have had an older representative of (ai), and that at an earlier period it had been so, are widely different propositions. The former I deny; the latter, in many cases, I admit.

sound (ai).

Though the proofs are numberless that from the 14th century, or earlier to the present day, chain has been sounded like pain, and the latter, from pana, has no radical (a), nor is likely to have been ever sounded with (a); chain on the contrary, from catena, had a radical (a), and one cannot doubt that (kaena), (tshaena), (tshainǝ), were early stages through which the word passed. So facere, that is (fakere), passed through (faere), (fairǝ), to the modern faire (feer).

In like manner, though no (a) remains in the modern day, it exists in the Germ. Tag, Du. dag, O. N. dagr, M. G. dags, &c., as well as in the A. S. dag and dah, from which dæg and da33 (Orm.) lead on to day (dee). Many other words now sounded with (ee) or (ee) are shown by their etymology to have undergone like change of sound,--air, Lat. aer, chair, Lat. cathedra, Spain, Lat. Hispania, champaign from campanus, &c.

67 Summary

of

arguments on ai words.

In taking leave of these ai words it is important to observe that, varied as are the sources of information to which I appeal, there is little clashing as to the general results they yield, which lends to the several results most weighty confirmation, based as they are on entirely independent evidence. Rhymes in Early English, Early Scotch, Early French; orthography, especially of the Ormulum; distinct statements of old grammarians; assonances in Early French poetry; etymologies; modern pronunciation of German, Dutch, Icelandic, French; and above all, the pronunciation of most of the English dialects*-all these for the most part harmonize in the conclusions which they dictate. Early rhymes habitually associate these words-may, dey, lay, fay, obey (above § 29). Icelandic pronunciation fixes the first two; assonances and etymology fix the other three; modern French pronunciation also bears witness to the last; and these sources of information all give us the same sound, while modern English pronunciation fully accords both as to these and others that rhyme with them. We shall find entirely independent, though less various evidence as to the vowel sound in knee, see, he, me, &c.; and these words

* I must notice at least in a foot-note the objection that in Middlesex and some adjoining counties words written with ai are often sounded with (ai) or (xi) or (ææi). But in fact this sound being given to words with the simple a as well as to words with ai-to pane, lane, mane, as much as to pain, lain, main the argument proves too much, and therefore nothing. If sail, A.S. segel, Ger. Segel, O.N. segl, &c., with no radical (a), is now locally sounded with (ææi), the simplest solution is that this (regen) has become (reen), then (ren), then (reein), and that this the prevailing pronunciation has then been corrupted into (rææin).

though Mr. Ellis would sound them (kne), (se), (He), (me), &c.—never rhyme with the class we have just been discussing.

tainly not (ee),

sound.

some

Now it needs but a slight acquaintance with Chaucer to 68 'A' words: discover that many pairs of words which rhyme Chaucer's a cer- now-one word containing one of the last disbut (a) cussed diphthongs and the other the simple a-never rhyme in Chaucer. Thus travayl, aveille, apparaille, never rhyme with dale, vale, tale; nor eyr, despeir, faire (s. or adj.), debonaire, with fare, care, snare, tare; and so on. Moreover many of these words with the single vowel are of French derivation, and there is no reason to suspect that tradition has not preserved in them in French the true pronunciation of (a); and hence it is likely that such words, though now sounded with (ee) or (ee), yet, having certainly undergone some change, were sounded in the 14th century with some (a) sound; so that also the Dutch faam, naam, dal, taal, aap, staat, waar (s.), waken, maken, at least approximately represent the English pronunciation of these words for several centuries.

But a change having taken place in the sound of so large a class of words, is there any means of ascertaining when change that change took place?

The

69
from (a) to the

earlier in Scot

through O. N. in

fluence.

present (ee) It was certainly effected much sooner in land than in En- Scotland than in England, and mainly, I begland, probably lieve, arising from the fact (see above, § 48) that the North Angle dialect was so close akin to the Old Norse. In the Lancelot of the Laik, in Ratis Raving, and other early Scottish poems, we find words rhyming habitually which never rhymed in Chaucer, nor even in Ben Jonson, though some of them did frequently in Spenser. Thus grace, place, pace, or paiss, space, cass, face, all (a) words in Chaucer, rhyme with fadyrless, makless, perches (= purchase), wantonase, gudlynes, lawlynes, meknes, rychess, &c.* Maade (vb.), degrade, raide

* Mr. Murray suggests, with some plausibility, that the (a) and (e) classes of words met on the common ground of (x), the -ess of -less, -ness, &c., being sounded much like the English ass (æs), knowledge as (nolædzh), and so on. This, however, seems to apply only to the short vowels.

E

THE UNIVEKSHY OF MUHIGAN LISKAKIES

(vb.), rhyme with paid, affraid, saade (= said),* arayd, and manhed; visage and rage with knawlege; schame, name, blame, with thaim (O.N. þeim), and hame (O.N. heim); declare, spare, are (vb.), with mare or mair (adj.), debonaire, fare or fair (adj.), repar (vb.), aire or are or ere or eire (adv.), aire (s.), 3ere, frere, hair, &c.; estate, debait, &c., with blait, from O.N. bleyta, and hate (= hot, shortened in later Scottish into het) O.N. heitr, and have and craif (A.S. habban and crafan, but O.N. hefi and kref) with laif or lave or laiffe (O.N. leifar) and resaif. So in Barbour, who was contemporary with Chaucer, we find slain, which elsewhere and most frequently rhymes with again, as it might in Chaucer, rhyming repeatedly with ane = one (O.N. ein), gane gone (O.N. geingit), and tane = taken; none of which rhymes would be admissible in Chaucer. Can it be that toward the close of Elizabeth's reign the probability, and afterwards the fact, of a Scottish succession to the throne, aided and accelerated, if it did not even cause, the change of pronunciation in England ?

take place

=

It seems very unlikely that mere court influ70 The change in England did not ence could have thinned down a full bold (aa) through Stuart into (ee) or (ee) in the mouths of the sturdy. influence. Englishmen whom the early Stuarts ruled; and there are many indications of a rugged spirit of independence among the people that was quite prepared to resist court influence even in smaller matters than shipmoney and episcopacy. Yet in Milton and Dryden such rhymes as maid shade, fail ale pale, spare air bare, praise amaze, state wait, are sufficiently common to suggest a

No argument can be based on the mere spelling of the Scottish words, if Mr. Murray's view is correct that the i or y in these digraphs in Middle Scotch simply indicated the length of the vowel preceding. This view however still leaves it an open question what that preceding vowel itself was— whether (aa), (ee), or (ee)—in these words. But it will be observed that the argument in the text is based on the words themselves, irrespective of modes of writing. In Chaucer the past tense made, however spelt, never rhymed with saide, however spelt; and I should argue that the radical (a) in the former, and the radical (e) in the latter, sufficiently indicate an original distinction which in Middle Scotch has been blotted out.