¹ See Wilhelm Streitberg, Die gotische Bibel ³ xix ff. (Heidelberg, 1950).

² Herbert Penzl, JEGP 49.217–30 (1950), shows the importance of emphasizing orthography and the relationship between orthography and phonemic pattern in considering Wulfila's dialect. For criticism of this approach see V. M. ZhirmunskiÎ, Voprosy jazykoznanija 8.477 (Moscow, 1959).

³ A number of scholars have felt that diphthongal values for the Gothic digraphs ai and au involve unlikely interpretations of Gothic orthography. For discussion see W. H. Bennett, Lg. 25.15–21 (1949). For additional bibliography see Fernand Mossé, Bibliographia Gotica, First Supplement, Mediaeval studies 15.176 (1953); Mossé and Marchand, Second Supplement, MS 19.185 (1957). See also Eric Hamp, Lg. 34.359 f. (1958), and ZhirmunskiÎ, op.cit.

⁴ J. F. Kräuter, Zur Lautverschiebung 120 f. (Strassburg, 1877), suggests that diphthongal glides were designated by w and not by u, but—as pointed out by Jellinek, ZfdA 36.272 (1892)—fails to account for the digraph iu.

⁵ Lg. 34.353–58 (1958). See also Eric Hamp's discussion of the problem (Lg. 34.359–63).

⁶ To make this work out, Gothic ƕ and q must be interpreted as unit phonemes /h^w/ and /k^w/, rather than as clusters /hw/ /kw/ (see Bennett, Lg. 35.429–32 [1959]); /i/ /j/ /u/ /w/ must be interpreted as four separate phonemes (see Penzl, JEGP 49.224 f.). Where the orthographic combinations au and iu designate phoneme sequences, it is assumed that each component of the sequence is a syllabic, e.g. ga-u-laubeis (John 9.35), ga-u-ƕa-seƕi (Mark 8.23), swa-u (John 18.22), bi-u-gitai (Luke 18.8), hailidedi-u (Mark 3.2), ni-u (Matt. 6.26); cf. the similar sequences aiu, aio, ou, e.g. habai-u (Luke 14.28), qimai-u (Mark 15.36), armai-o (Gal. 6.16), lailo-un (John 9.28), waiwo-un (Matt. 7.25), ƕo-uh (2 Cor. 6.15). I see no reason to believe that the iu occurring in certain verbal forms is an orthographic simplification of iju, viz. sium 1 pl. (2 Cor. 2.17A), siuþ 2 pl. (2 Cor. 3.2A), as implied by Streitberg's statement, Gotisches Elementarbuch ^3–4 52 (Heidelberg, 1910): ‘i für ij vor Vokal deutet auf schwache Artikulation des Übergangslautes.‘ It is just as reasonable to assume that in the scribal dialect these forms were pronounced with hiatus, si-um and si-uþ like ni-u and hailidedi-u. As has been pointed out by Jacobsohn, ZfvS 47.84 (1916), the variants without j occur only in the epistles and in a part of the gospels (Luke) where aberrant forms of various types are common. For discussion see Kluge, Elemente des Gotischen 13 (Strassburg, 1911); Wrede, Stamm-Heyne's Ulfilas ^13–14 303; Jellinek, Geschichte der gotischen Sprache 63 (Berlin and Leipzig, 1926); Mossé, Manuel de la langue gotique ² 143 (Paris, 1956); Schulze, ZfvS 41.175 (1907); Gaebeler, ZfdPh. 43.40 f. (1911).

⁷ Vocalic w does not occur in the accented syllables of native words. Nonsyllabic w does not combine with a preceding vowel to form an unaccented syllable in native words except perhaps in lasiws (see below). It is assumed that these restrictions do not apply to the loan words.

⁸ The values represented by u and oi seem to have been identical in Wulfila's Greek; see E. H. Sturtevant, The pronunciation of Greek and Latin ² 52 (Philadelphia, 1940). In my opinion, the fact that Gothic w rather than a digraph ∗oi is used in spelling forms like Lwstrws dat. pl. for Lustrois lends strong support to Bennett's sound substitution theory. If Wulfila had needed a neutral symbol to designate a syllabic which did not occur in Gothic, mechanical transliteration of the Greek digraph oi would have been the most logical solution.

⁹ In a few instances Gothic au is substituted for u; see the discussion in my paper Gothic au in Inflectional Syllables, Lg. 34.37 (1958).

¹⁰ It is assumed that ju for iu reflects a difference in pronunciation (see Penzl, JEGP 49.225; Jellinek, AfdA 28.23–4 [1902]), e.g. Judas (John 12.4), Iudas (Matt. 27.3) for Ioúdas, Judaius (1 Cor. 9.20) for Ioudaíous, Iudaius (John 18.35), judaiwiskaize adj. (Tit. 1.14) for ioudaïkoîs, iudaiwisko adv. (Gal. 2.14) for ioudaïkôs. The examples suggest that varying stress patterns (and uncertainty regarding the stress patterns) could have resulted in the loss of syllabicity. Very little is known about loanword accentuation and its effect on Gothic transcriptional practices. A few details are discussed by Kluge, El. d. Got. 8 f., 43.

¹¹ The case for the proposed solution becomes even stronger if it is assumed that eu and au were diphthongs in Wulfila's Greek (see §3 below).

¹² Perhaps lasiws should be included in this list.

¹³ The possibility of diphthongs is mentioned by Fernand Mossé, Manuel de la langue got.¹ 41 (Paris, 1942); Bennett, Lg. 25.19 f.; Hamp, MLN 71.266 (1956). Mossé includes eiw: ‘eiw existerait seulement dans les impératifs (non attestés) hneiw∗ “baisse-toi”, speiw∗ “crache”.‘ This would seem to indicate that Wulfila depended on position to signal the difference between tautosyllabic and heterosyllabic [], e.g. ∗hneiw versus hneiwan (Luke 9.12), lew (Rom. 7.8) versus lewa (Gal. 5.13), sunsaiw (Luke 4.39) versus aiwins (Matt. 6.13). If so, how were borrowed forms like aiwaggelida (Gal. 4.13) and Daweid (John 7.42) pronounced? Would Goths who had had little or no contact with the Greek language observe one rule of syllabic division when reading native words and another rule when reading Biblical loans?

¹⁴ Moulton writes (Lg. 24.84): ‘We cannot know whether there was any real phonetic difference between [] and [w].‘ Eduard Sievers, in PG ¹ 1.413 f., tries to distinguish between a [] which occurs in falling diphthongs, e.g. niu-ja, hau-ja and a [] (designated by the w symbol) which comes after the Silbengrenze (Druckgrenze), e.g. Pa-wlus, ai-wlau-gi-a, le-wjan, bal-wjan. Jellinek (ZfdA 36.270) shows that this theory is untenable by citing Esaw, lew, lasiws. Lwstrws, waurstw (John 6.29), and þiwadw (Gal. 4.24) might be added to this list.

¹⁵ For other proposals see A. G. van Hamel, Gotisch Handboek 30 f. (Haarlem, 1923); Joseph Wright, Grammar of the Gothic language ² 12 (Oxford, 1954); E. Prokosch, A comparative Germanic grammar 135 (Philadelphia, 1939); A. M. Sturtevant, GR 32.314–8 (1957).

¹⁶ Moritz Heyne, Friedrich Ludwig Stamm's Ulfilas ⁶ 250 (Paderborn, 1874); R. Bethge, in Dieter's Laut- und Formenlehre der altgermanischen Dialekte 197 (Leipzig, 1898): ‘stimmhafte bilabiale spirans’; Jellinek, ZfdA 36.276 f.: ‘ein … u-haltiger labialer spirant’, and Geschichte der gotischen Sprache 39 (Berlin and Leipzig, 1926); W. van Helten, ZfdA 37.121 f. (1893) and IF 14.69 ff. (1903); Ernst Kieckers, Handbuch der vergleichenden gotischen Grammatik 8 (Munich, 1928); Wilhelm Braune, Gotische Grammatik ¹⁵ ed. by Karl Helm 28 f. (Tübingen, 1956); Wolfgang Krause, Handbuch des Gotischen 63, 102 (Munich, 1953). As recently as 1958 (Lg. 34.354) I was inclined to admit the possibility of a spirantal w in Gothic.

¹⁷ Streitberg, GE ^5–6 60 (Heidelberg, 1920), F. Mossé, Manuel de la langue got.² 41 f.

¹⁸ Apparently this was suggested for the first time by Wilhelm Weingärtner, Die Aussprache des Gotischen zur Zeit des Ulfilas 50 (Leipzig, 1858). See also Braune, Got. Gr. ¹ 17 (Halle, 1880) and Jellinek, ZfdA 36.278 (1892).

¹⁹ This is suggested by the use of ou and b in names from dialects which could hardly be expected to have had the spirantal reflex of Germanic [] and by the occurrence of spellings like Alboúinos Batáouai, Batáboi Batáoi Batao.oi. See M. Schönfeld, Wörterbuch der altgermanischen Personen- und Völkernamen (Heidelberg, 1911).

²⁰ Braune, Got. Gr. ¹⁵ 29.

²¹ Pron. of Greek and Lat. 88. It is evident that Greek b and Gothic w did not designate sounds which were close approximations of one another. In Biblical loans, b transcribes Greek b; w is never used for this purpose.

²² Sturtevant, Pron. of Greek and Lat. 142–3; Max Niedermann, Phonétique historique du latin 158 (Paris, 1945). For the Latin spellings cited in this paragraph see Braune, Got. Gr. ¹⁵ 29, and Schönfeld, Wörterbuch 83, 253 f. For additional material see Franz Dietrich, Über die Aussprache des Gothischen während der Zeit seines Bestehens 79 (Marburg, 1862).

²³ Lg. 33.538–44 (1957). I am indebted to Bennett for calling my attention to Austin's article and for a number of other very helpful suggestions.

²⁴ Lg. 35.428–9.

²⁵ Although we cannot be sure how u was pronounced in Wulfila's Greek (see Lg. 35.428 fn. 8), failure to use Gothic ei, i, or e in transcriptions suggests that lip-rounding was a characteristic feature. The assumption that the Gothic digraph iu designates a central vowel (without lip-rounding) would explain why the spellings w and iu are never confused. This solution is a compromise between the two possibilities mentioned in my earlier paper (Lg. 34.356 ff.).

²⁶ This theory (combined with the opposition Westgoten–Ostgoten) was proposed by Ferdinand Wrede, Über die Sprache der Ostgoten in Italien 165 f. (Strassburg, 1891). For discussion see Bennett, Lg. 25.17 ff. Another possibility would be to check for scribal errata involving w and another consonant. I have found only one such spelling and evaluate it like H. C. v. d. Gabelentz and J. Loebe, Grammatik der gotischen Sprache 40 (Leipzig, 1846): ‘Aus dem Schreibfehler vidvor [i.e. widwor] statt fidvor Cor. 2, 12, 2 darf wol kaum auf eine Annäherung des v an das f geschlossen werden.‘

²⁷ See Wright, Gr. of the Gothic lang. ² 107, and, opposing this formulation, A. Walde, Beiblatt zur Anglia 22.261 (1911).

²⁸ C. C. Uhlenbeck, PBB 30.318 (1905), suggests that uhtiug could be a scribal error. Sigmund Feist, Vergleichendes Wörterbuch der gotischen Sprache ³ 515 (Leiden, 1939), remarks: ‘Bildung dunkel.‘ Theodor von Grienberger, Sitzungsb. der kaiserl. Akad. der Wiss., Phil.-hist. Cl. 142.145 (Vienna, 1900), explains lasiws by deriving Gothic -iws from IE -īwos (as in Latin captīvus). For other interpretations see Jellinek, ZfdA 36.277, Walde, Beiblatt zur Anglia 22.261 and A. H. Sturtevant, JEGP 24.195 f. (1925).

²⁹ Gotisch Handboek 69.

³⁰ Footnotes 6 and 10.

³¹ Griechische Lehnworte im Gotischen, Sitzungsberichte der königl. preussischen Akademie der Wissenschaften 36.746 (1905).

³² See Wilhelm Luft, ZfvS 35.309 (1899).

³³ Op. cit.

³⁵ It is generally agreed that the symbol is based on Greek upsilon. In his very interesting article, Das akrophonische Prinzip und Wulfilas Alphabet, ZfdA 86.265–75 (1956), James Marchand suggests a derivation of the Gothic symbol based on its nonsyllabic function.