¹ Lg. 35.575–97 (1959), with references to previous treatments.

² E. Haugen, ed. and transi., Language monograph 25.13–4 (1950).

³ Hockett's general presentation of the history of the term ‘short diphthong’ for ea, eo, io (577) is essentially correct. The first scholar to recognize the monophthongal nature of Old English æ and oe was Rasmus Rask in his Angelsaksisk Sproglære tilligemed en kort Læsebog (Stockholm, 1817). Rask, who had, for all practical purposes, only Hickes before him, was able to systematize the vowel changes in Old English rather well, primarily because of his familiarity with ancient and modern Scandinavian languages. Thus, he recognized the historical relationship between [u] and [y], [o] and [ø], [a] and [æ] [e], because there is a regular alternation between these sets of vowels in Old Icelandic and in his mother-tongue, Danish. Even the symbols for these sounds presented no problem for him, since Danish still uses æ and y to designate the i-mutation of a and u, and a relationship between the Danish symbol ϕ and the early Old English oe was clear to him through his studies of Old Norse paleography. Rask therefore assigned to the Old English symbols æ, oe, and y the phonetic values of Danish æ [æ], ϕ [ϕ], and y [y] (5–6). He also recognized the alternation between æ, e, i and ea, eo, io, ie in Old English (2 and 12–3), but the Scandinavian languages provided no basis for assigning a monophthongal pronunciation to the latter digraphs; indeed, such correspondences as OE eorð: OIcel. iorð, Dan. jord ‘earth’ dictated against such an assumption. As a result, Rask retained the traditional view (dating from Hickes) that ea, eo, io, and ie must represent diphthongs. Jacob Grimm leaned heavily on Rask in his treatment of Old English in the second edition of the Deutsche Grammatik (Göttingen, 1822) and thus passed on to succeeding generations the division of as, oe, y and ea, eo, io, ie into monophthongs and diphthongs respectively.

For a detailed discussion of the early scholarly treatment of the mutated vowels in the Germanic languages, see ch. 1 of my dissertation, The Investigation of I-Mutation in the Germanic Languages (University of Illinois, 1961).

⁴ Hockett makes the point that /α/ 'was not necessarily rounded in the physiological sense,' but 'occupied the structural position indicated on the chart' (576).

⁵ I find Hockett's presentation of the development of Proto-Germanic /au, eu, iu/ to Old English /a-, ə, (i·)/ (595–7) very convincing, and much superior to the traditional view that the Old English reflexes were still actual diphthongs; cf. e.g. A. Campbell, Old English grammar §37 (Oxford, 1959). A distinct tendency of these Proto-Germanic diphthongs toward assimilation of the two elements, often leading to monophthongization, is apparent in Old Saxon (see J. H. Gallée, Altsächsische Grammatik ² §§89–108 [Halle-Leiden, 1910]), Old High German (see Braune-Mitzka, Althochdeutsche Grammatik ⁸ §§43–9 [Tübingen, 1953]), and Old Norse (see A. Noreen, Geschichte der nordischen Sprachen ³ §§40–5, 106–8, 155–7, and 160 [Strassburg, 1913]).

⁶ O. Höfler provides convincing arguments against the wave-theory of the spread of mutation from one Germanic dialect to the others and assumes a phylogenetic development in the various dialects conditioned by the gradual increase in strength of the Germanic stress accent (Akzentballung): PBB 77.30–66, 424–76 (Tübingen, 1955) and PBB 78.1–44 (Tübingen, 1956), as well as in Mnēmēs khárin: Gedenkschrift Paul Kretschmer 1.158–74 (Vienna, 1956).

⁷ Det philologisk-historiske Samfunds Mindeskrift 179 (Copenhagen, 1879).

⁸ Arkiv för nordisk filologi 73.120 (1958).

⁹ See J. W. Marchand, Lg. 33.346–54 (1957).

¹⁰ See W. F. Twaddell, Monatshefte für deutschen Unterricht 30.177–81 (1938), Lg. 24.139–51 (1948); H. Penzl, Lg. 25.223–40 (1949), Arkiv 66.1–15 (1951); J. Fourquet, Word 8.122–35 (1952), Travaux de l'Institut de Linguistique 1.109–24 (1956); V. M. Žirmunskij, Doklady i soobščenija Instituta Jazykoznanija Akademii Nauk SSSR 5.79–106 (1953), [Schirmunski] PBB 79[suppl.].374–87 (Halle, 1957); M. Steblin-Kamenskij, Arkiv 74.105–11 (1959); K. M. Nielsen, Arkiv 75.1–78 (1960); H. Kratz, JEGPh 59.463–79 (1960).

¹¹ Fourquet, Travaux 1.112–9.

¹² See Braune-Mitzka §27, Anm. 2; G. Manganella, Annali: Sezione germanica 1.139–51 (Istituto universitario orientale; Naples, 1958).

¹³ In the following, /-i/, /-a/, and /-u/ will be used to indicate all vowel phonemes in following unstressed syllables and consonants following the root vowel which have a similar mutating effect.

¹⁴ It should be noted here that these are not the only allophones to have developed. There seem to have been innumerable variants, some of which have dropped entirely from view, while others have left traces in individual dialects; see §2.4, and cf. Fourquet, Travaux 1.118. Žirmunskij, PBB 79[suppl.].384, states on the basis of the evidence in modern German dialects: ‘alle kurzen und langen Vokale und alle Diphthonge der vorderen, wie der hinteren Reihe werden im Althochdeutschen durch ein folgendes -i affiziert und zeigen engere (geschlossene), d.h. palatalisierte Lautvarianten, die als positionsbedingte Lautschattierungen in der Schrift nicht besonders bezeichnet werden.‘ For the purpose of comparing the development in Old Icelandic and Anglian Old English, however, the allophonic system outlined above is adequate.

¹⁵ See Höfler, PBB 78.16–26.

^15a [The inverted v is used here, in the absence of the proper IPA letter, to denote an unrounded vowel with the tongue position of o.—Editor.]

¹⁶ Concerning the feasibility of such a system, cf. Hockett, Manual of phonology §§244–2443, and N. S. Trubetzkoy, Grundzüge der Phonologie 106 (Prague, 1939).

¹⁷ Cf. e.g. A. Heusler's ‘labial umlaut’ as an inclusive term for u- and w-umlaut: Altisländisches Elementarbuch ⁴ §§67–75 (Heidelberg, 1950).

¹⁸ See Heusler §59 and §68; the traditional view of the chronology of i- and w-umlaut is complicated by Axel Kock's theory of different periods within each, although Kock's ‘older’ i-umlaut is placed before his ‘older’ u-umlaut. The ‘older’ u-umlaut is supposed to have occurred before ‘younger’ i-umlaut, however, so that Kock can agree with Wimmer's derivation of OIcel. see A. Kock, Arkiv 4.157 (1888).

^18a Kock's three-period theory of mutation in Scandinavian was formulated in order to account for the discrepancy in development in the long and short stems, e.g. *dōmiðō > OIcel. dϕmda, but *taliðō > OIcel. talþa. This theory, presented in all its ramifications in Kock's Umlaut und Brechung im Altschwedischen (Leipzig and Lund, 1916), enjoyed wide acceptance for over half a century, but strong opposition to it was voiced from the very beginning. In the last fifteen years, this opposition has steadily increased. The weaknesses in Kock's hypothesis have been adequately demonstrated in innumerable articles and monographs, of which I wish to cite here only the following: D. Seip, Maal og Minne 1919. 85–90; A. Sommerfelt, Festskrift til Hjalmar Falk 42–9 (Oslo, 1927); E. Neuman, APhS 4.193–246 (1929–30); B. Hesselman, Omljud och brytning i de nordiska språken (Upsala, 1945); H. Andersen, Arkiv 61.157–70 (1946), and Festskrift til Peter Skautrup 9–15 (Århus, 1956); the comments by T. Johannisson, A. Janzén, and H. Andersen in “Nordistmødet i København,” APhS 19.1–61 (1947); H. Penzl, Arkiv 66.1–15 (1951); and M. I. Steblin-Kamenskij, Arkiv 74.105–11 (1959). My presentation of the simultaneous rise of the various umlaut allophones necessarily implies rejection of Kock's period theory. There thus remains the problem of accounting for the lack of umlaut in such forms as OIcel. talþa. To my mind, no entirely satisfactory explanation has yet been offered, although there seems to be some merit in various proposals in which the stress patterns and types of internal juncture in words with long and short roots are credited with permitting or preventing mutation. Alois Walde was apparently the first to propose that there was a different type of juncture in long and short stems and to deny (in contrast to Kock) that the presence or absence of mutation is dependent on the time of the syncope or apocope of /-i/ (Die germanischen Auslautsgesetze 189–90 and 198 [Halle, 1900]). Closely allied to Walde's theory are the rhythm and stress hypotheses of Sommerfelt and Neuman. The juncture and stress theories have the virtue that they permit an explanation for the failure of mutation allophones to develop in the short stems without positing an artificial division of mutation into two periods separated by an interval in which the process was dormant. If open juncture between the root and the following syllable was a hindrance to assimilation (and thus to the formation of umlaut allophones), then we cannot agree with Penzl's statement, ‘If a form without umlaut is not analogical, it can only go back to a form from which the i-sound was lost before the development of those allophonic variations that led to umlaut’ (Lg. 25.229 [1949]). The problem is, of course, too complex to be discussed here in detail, but it may be pointed out that the anticipatory assimilation which is mutation is limited by junctural phenomena to vowel phonemes within single words. When the open juncture (in this case external) is removed, mutation can occur, as seen from the classic examples taken from Otfrid's Old High German: meg ih ‘kann ich’, drenk ih ‘trank ich’, werf iz ‘warf es’, geh imo ‘gab ihm’ (Braune-Mitzka, Ahd. Grammatik ⁸ §26, Anm. 3 [Tübingen, 1953]).

¹⁹ Indicated in the orthography about 1550; see Noreen §103.

²⁰ Similar to the system proposed for Chaucerian Middle English by Hockett, Manual of phonology 86. For the Icelandic changes, see Noreen §§89, 91b, and 103.

²¹ Lg. 34.65 and 69–70 (1958).

²² Haugen, Language monograph 25.13–4.

²³ H. Spehr, Der ursprung der isländischen schrift und ihre weiterbildung bis zur mitte des 13. Jahrhunderts 108 (Halle, 1929).

²⁴ Spehr 108.

²⁵ In spite of the First Grammarian's testimony that the symbol ø represents e + o, Spehr believes that ‘In Wirklichkeit stimmt dieerklärung wohl nur für (108, fn. 6).

²⁶ oe has been inserted by the editor in place of the manuscript's sign for oc ‘and’, which must be a scribal error; B. M. Ólsen, ed., Den tredje og fjærde grammatiche af handling i Snorres Edda 47–8 (Copenhagen, 1884).

²⁷ Ólsen 48, note to line 35.

²⁸ See my dissertation §1.3.

²⁹ Note that Hockett's /ə/ = my /∧/, Hockett's /i/ = my /m/, Hockett's /a/ = my /ə/, and that I operate with four tongue heights in contrast to Hockett's three.

³⁰ See Sievers-Brunner, Altenglische Grammatik §10 and §14 (Halle, 1942).

³¹ The numbers indicate the origin of the phonemes in comparison with the Proto-Germanic diagram above. I have not included here the development of /a/ before nasals, which may well have resulted in a new back allophone; see Sievers-Brunner §79, and Campbell §40.

³² In this comparison of the system proposed by Hockett with the one I am suggesting, I shall use the Old English designations of the phonemes for the sake of clarity, since I have found it necessary to use some of Hockett's symbols with values different from those which he intended.

³³ An intermediate stage æ is usually assumed in the handbooks for the development from Proto-Gmc. a to OE ea; cf. Sievers-Brunner §84, Anm. 1, and Campbell §139, fn. 1.

³⁴ Cf. Sievers-Brunner §93 and §108.2, and Campbell §205 and fn. 2.

³⁵ See note 14.

³⁶ See Sievers-Brunner §105.

³⁷ Anglian e instead of æ before rc, rg, rh is a younger development; see Sievers-Brunner §120, Anm. 2.

³⁸ In the case of the Anglian forms cited, it is usually maintained that i-umlaut did not occur; see Sievers-Brunner §107, Campbell §201.

³⁹ Cf. K. Brunner, Die englische Sprache 1.212–3 (Halle, 1950).

⁴⁰ See Jordan-Matthes, Handbuch der mittelenglischen Grammatik ² §65 and §66, Anm. 3 (Heidelberg, 1934).

⁴¹ Brunner 1.212–3.

⁴² Jordan-Matthes §65.

⁴³ It is significant that the traditional view of Scandinavian breaking as a type of umlaut has been seriously questioned in recent times by John Svensson, Diftongering med palatalt förslag i de nordiska språken (Lund, 1944), and Valter Jansson, Nordiska vinnamn (Upsala, 1951), who believe the breaking to be essentially a spontaneous diphthongization. While the new interpretation has met much opposition (see ‘Nordistmødet i København’, APhS 19.3–61 [1947], I. Hoff, NTS 14.315–40 [1947], M. I. Steblin-Kamenskij, Studia linguistica 11.84–91 [1957], K. M. Nielsen, APhS 24.33–45 [1957]), the fact remains that no one has yet been able to propose a reasonable theory which would explain why u- (w-) umlaut in Old Icelandic changes a > o, æ > ϕ, e > ϕ, and i > y (all monophthongs), while breaking affects only e and results in a diphthong (cf. L. Posti, SSUF 1946–8, 39–59, and J. Fourquet, Travaux 1.111–3).

⁴⁴ Cf. Sievers-Brunner §95, Anm. 2; Campell §203; Marchand, Lg. 33.351–4; and H. Weyhe, PBB 31.43–78 (1905–6).

⁴⁵ Cf. e.g. C. B. van Haeringen's dilemma, De germaanse inflexieverschijnselen (‘umlaut’ en ‘breking‘) phoneties beschouwd 153 (Leiden, 1918).

⁴⁶ Cf. R. Kögel, Literaturblatt für germanische und romanische Philologie 8.108 (1887); M. Joesten, Untersuchungen zu ahd. (as.) ë, i vor u der Folgesilbe und zur 1. Pers. Sg. Präs. Ind. der starken e-Verben (Kl. III_b, IV, V) (Giessen, 1931) ; Marchand, Lg. 33.346–54 (1957).

⁴⁷ See Noreen §7.1, and Altschwedische Grammatik §§65–71 (Halle, 1904); P. Skautrup, Det danske sprogs historie 1.124–5 and 127–8 (Copenhagen, 1944).