PRONUNCIATION OF OLD NORSE.

Rask considers that the modern pronunciation is practically the same as the ancient, except in a few instances, hence in the following table the modern forms as already explained, are given in Rask's column, and his supposed ancient values are bracketed. Rapp gives an opinion upon nearly every letter in the alphabet, and although he did not consider that he had arrived at a result sufficiently definite to give an example, he has transcribed a large number of words into his alphabet, a selection of which is subjoined. Grimm's pronunciation is not easy to be determined, and the sounds which I have given must be therefore considered to be in great part conjectural. The vowels are taken from the third, and the consonants from the second edition of his Grammar.

On these conjectures generally I make no observation, except to remark that I feel doubtful as to the value which Rask meant to ascribe to the old u. He says: "u, without accent, may perhaps have had the sound of the short English u in nut, but, the Danish o in hos, the Swedish o in sporde, menniskor lärorik, etc." These sounds are certainly not identical, and I have been accustomed to consider them as (o, o, u) respectively. Grimm assumes the English u to be a sound between German o and ö, whatever that may mean. Neither he nor Rask, therefore, had mastered the English (e, a) sounds. I have represented Rask's ancient u by (o, u) doubtfully, but believe that the latter is more probable.

1

Letters.

Modern & Grimm. Rapp. Letters. Modern & Grimm. Rapp. [Rask].

[Rask.]

Old Norse words as pronounced by Rapp: á (AA) in, water, æ (ee) always, átta (AAtta) eight, auk (euk) also, auga (ouga) eye, bleikr (bleik r) pale, bleydi (bləydh'i) fear, blés (blees) blew, blód (blood) blood, bökr (book'r) books, bók (book) book, brúda (bruudh a) of brides, býd (byydh) invite, byggia (byg Ja) build, dagr degi dögum (dager, degi, dag um) day, to a day, to days, dóttir (doot tir) daughter, dýpi (dyypi) depth, ey (ǝy) island, eyk (eik) oak, fel (fuel) fell, fliúga (fliu ga) fly, fötr (foot r) feet, frí (frii) free, fullr (fol·r) full, fylli (fyli) fullness, gæs (gees) geese, gás (gAAs) goose, göra gjórt (gər a gJoort) to do, did, hálmr (khaalm r) halm, hlaup (khlaup) leap, hniósa (khnioo'sa) sneeze, hreinn (khroinn) pure, hvítr (khbhiitr) white, kaupa (kǝup a) to buy, kné (knee) knee, kránkr (kraaqk ́r) sick, liúga (liuga) to tell a falsehood, opt (opt) often, skapt (skapt) handle úngr (uuq g'r) youth, verd (bherd) price, vís (bhiis) wise.1

The following observations on the Old Norse pronunciation, based upon a phonetic examination of the structure of the language, its connection with the Teutonic branches and the usages of Old MSS., are drawn up from notes kindly furnished me by Mr. Henry Sweet, of the Philological Society (suprà p. 539, L. 9).

1 The following is a translation of Dr. Rapp's latest views on the subject (Vergl. Gramm. iii. 40). "Of the seven long vowels, the two strongest (ii) and (uu) have remained intact. The (aa) subsequently, as everywhere else, degenerated in the direction of (0). The mutates of (aa, uu) must here be (EE, yy). There must be an (ou) corresponding to the old German diphthong ei, but it is here written au, since the mutate, if written ey, could only mean (oy); the Norwegian dialects retain (ou). Long (00) afterwards became diphthongal, and its mutate coincides with a (EE). The third long vowel wavers between gothic (iu), becoming, when softened (geschwächt) (10), and confluent (yy). Isolated remains of (ee) subsequently passed into (Jee) as in Sclavonic; but the e which arises from reduplication need not necessarily be long. As regards the mutation of the short vowels, the change of (a) into (e), and of (o, u), into (y) is clear, but the mutation of (a) into through the action of a following or (o) is more obscure. We can theoretically assume an earliest period in which (a) remained pure, but it does not agree with the period of existing monuments. Hence we allow (a) to pass into (0) but entirely reject the usual assumption of the generation of of an-impossible-(c) from (o). The division (Brechung) of short (e) into (ía) and by mutation (ío), must also be

observed. As regards the consonants we assume h and s, here as elsewhere, to have been (kh, sj), though we write (s). The z was an abbreviation, generally for (ts), occasionally for (st), and by mistake for other combinations; the first alone must be retained. The pis initial as in Gothic, but medially and finally it is softened to ; as this also happens in most cases to the modern Danish d, both classes must be distinguished from out of the corruption of writing. This is the weakest point in northern philology. The old runic alphabet has only the aspirate p (th) and this is used medially even in the oldest manuscripts. The modern Icelandic and Danish (dh) is on the contrary not an aspirate but a spirant, which is more naturally developed from (d) than from (th). But since Scandinavian orthography is here irremediably confused, nothing remains but to restore the old essential organic pin all places where it is required by Gothic, Anglosaxon, and Friesic, and in other, partially doubtful cases, to leave d, so that the modern is altogether eliminated. The tt, which arose from an older (kht), must certainly be sharpened, [that is, make the preceding vowel short], since reduplication can mean nothing but confluence; the prolongation of the vowel in this case is a modern corruption, which even Grimm has overlooked, and similarly before ng, nk, and I followed by a consonant, etc."

When Icelanders first employed the Latin alphabet they had no written literature at all, and consequently no traditional orthography to transliterate, that is, no theoretical guide to mislead them. They had therefore, no means of writing except by ear, using the Latin letters in their accepted values, and modifying them for new sounds. Under such circumstances, it is scarcely possible that they should have

1) expressed one sound in two ways, as in the modern identities ý, iy, ei ey.

2) made á represent (au) to the exclusion of au,

3) have used au to express a sound (i) for which they had a form to hand, namely öi, unless indeed they had read in Grimm's Deutsche Grammatik (13, 474), that old Norse au corresponds to Gothic au, and had foreseen that the sound (au) would have been preserved in the German of the xix th century.

A comparison of the old cognates shews that the difference between a á, e é, etc., was originally purely quantitative. In modern Icelandic, as in Modern German, all short vowels before single consonants have become long, but in old German the length or shortness of a vowel was quite independent of the following consonant, as is proved by the metrical laws. In the same way the non-accentuation of faðir, father, in Icelandic originally meant that the vowel was short, and the accentuation of móðir, mother, that the vowel was long, as in Latin påter, māter.1 If this view be well founded, the vowels in each pair, as a, á, e, é, etc., must have had the same quality, but different quantities, a, e, etc., being always short, and á, é, etc., always long; and diphthongs must have had the sounds of their elements connected by the glide. The following sounds appear then to be the only possible.

(y, I, ǝ)

some modification of (0) The two principal criteria for 1) The palatisation of k, g, and or Umlaut, (um laut).

As Icelandic still possesses really doubled consonants, the device of doubling the consonant to indicate the brevity of a preceding vowel was not likely to occur to the writer. That the length of a vowel depends in any way upon the number of following consonants is a delusion, to be classed with the notion that all vowels under the stress must be long, and deducible probably from the false statement in Latin prosodies, that a short vowel might become

1) The palatisation of k, g, from (k, g) into (kj, gj) naturally takes place before front vowels (p. 13), while these consonants remain unchanged before back vowels (p. 13). Existing habits as to palatisation would hence determine

e, i í, y ý, æ*, ei, ey to be front vowels, and

a á, o ó, u* ú, ö* au* to be back vowels,

*

whereas those marked transgress this rule, a=(aai) commencing with a back vowel, and u, ö, au = (ə, œ, œœi) with a front vowel.1

2) Vowel mutation is the result of the partial assimilation of two vowels, not in juxtaposition, but in consecutive syllables, whereby the first or accented vowel becomes modified in the direction of the second. This may be expressed by such a formula as (a..i = e), meaning that (a) in the first syllable acted on by (i) in the second is converted into (e). The original sounds of these mutated vowels or mutates, have been so changed in Icelandic, that it is necessary to examine the other Teutonic languages where they are better preserved.

(a..i=E, e), giving (E); (i.. a=e, E), giving (e);

old Ger. hari (Hari), modern G. heer (HEEr) army. Gothic niman (nim‍an) modern G. nehmen (nee men) to take; the (1, e) forms are confused in modern German. (。 ..i=əh, ǝ, 1), giving (əh); old Ger. scôni (skooni), mod. G. schön (shœœen) beautiful. (u.. a=o, A), giving (0); Gothic stulan (stulan), mod. G. ge-stohlen (ge-shtoo

len), stolen.

(u ..i=1), giving (1)); old G. sundia (sund лa), mod. G. sünde (zynde) sin. In Icelandic we find, her, nema, stolinn, synd (Heer, nee'ma, stool in, sind) all with mutates. The equation of the last word with modern pronunciation is (u..ii) which is not a mutation at all. The old sound must have been (1) or (y), as these are the only possible intermediates. The vowel mutation also proves that the modern sound of a is inorganic.

(aa.. i=EE), old Ger. wûri (bhaari), Icel. væri.

(oo.. i=ah), Gothic ? fôrjan (foor Jan), Icel. færa, old fœra.

The genuineness of the sound (c) is made doubtful by the nonpalatalisation of k, and this doubt is confirmed by the equation (a..u=o), as in dögum for dagum. As both vowels are back, the result cannot be front. And the back sound of u is shewn 1) by the preservation of that sound in long ú, 2) the nonpalatisation of k before it, 3) the vowel mutation. The a.. u=ö, is merely a reversal of (u.. a=o) in stolinn, ge-stohlen, and both are quite parallel with (a..i=e, i..a-e).

The above conclusions result from the structure of the language,

1 The remarks on p. 206 shew that this criterion cannot be relied on so far as æ is concerned, and, indeed, the palatal action of a on k, g, while a, á, produced no such action, may have arisen from the anticipatory action of the second element (i). Nor is there any organic necessity for the palatalisation of k, g, before such obscure vowels, as

(a, c), as we see from the fact that although both sounds are used in different parts of Germany for ö, which is also frequently called (ee) or (ee), yet the k, g, of könig, Goethe, are never palatalised. This criterion can therefore only furnish an à priori probability.-A. J. E.

the following is almost positive evidence of the usages of the x th century. Þóroddr, the grammarian, circà 1160, remarks on the necessity of an A, B, C, and after stating that the English have made an alphabet for themselves by adopting or modifying the Latin letters, he proposes to perform the same service for his countrymen—oss Islendingum, saying:

"To the five original Latin vowels a, e, i, o, u, I have added four: [now o], [now e, a], [now ce, æ], y [now y]. Of these ? has the curve of a and the ring of o, because it is blended of their two sounds, being pronounced with a less open mouth than a, but a more open mouth than o; g has the curve of a and the whole figure of e, for it is composed of these two, being pronounced with a less open mouth than a, and with a more open mouth than e; & is composed of e and o, being pronounced with less open mouth than e, and with more open mouth than o; and y is composed of i and u, being pronounced with less open mouth than i and with more open mouth than u."

%

He proceeds to give examples, shewing that e and e short correspond to modern e, e long to modern é, e long to modern a, o to modern o, o to modern ö, and to a now a. And then he remarks that each of these vowels begets another by being sounded in the nose, which he marks by a point above the letter. This probably corresponds to the palaeotypic (.), not to (A). It is now quite lost. Hence Rask's imaginary nasality, suprà p. 550, 1. 3.

Þóroddr further states that each of these 18 vowels can be long or short, and proposes to mark the long vowels with an accent. His examples shew that he places this accent in those places where an accent (indicating a diphthong in the case of á, ó), now exists in Icelandic. Then he concludes by enumerating the diphthongs, describing accurately the nature of diphthongs in general. Among these diphthongs appear au, ei, ey, but not á, ó.

The older MSS. follow póroddr with some variations. Thus the diacritic is often written as a full letter, as ao for o, ae for e whence modern æ, and the diacritic is not unfrequently entirely omitted, so that e, o, are confounded with e, ọ.

The following examples shew Þóroddr's spelling compared with that now used, and the probable corresponding pronunciation. Abbreviations-p. Þóroddr's spelling, M. modern spelling, OP. old pronunciation, MP. modern pronunciation.

p. M. O.P. M.P. p. M. p. M. O.P.

M.P. p. M.