² Old English sound changes reconsidered in relation to scribal tradition and practice, Transactions of the Philological Society 108-37 (1939).

³ Manuel de Vanglais du Moyen-Âge, I. Vieil-anglais (Bibliothèque de philologie germanique, Vol. 8; Paris, 1945).

⁴ Some Old English graphemic-phonemic correspondences—ae, ea, and a (Studies in linguistics: Occasional papers, No. 4; Washington, 1951).

⁵ These scholars are not, of course, the first to question the diphthongal interpretation of OE digraph spellings. Many grammarians assume that in words like geoc, the e is diacritical; and as early as 1868, W. Scherer suggested that ea and eo in breaking and velarumlaut positions indicated no more than that the following consonants were pronounced in a certain way. See Zur Geschichte der deutschen Sprache 141 (Berlin, 1868).

⁶ They do not attempt an explanation of nēah, in which we have an earlier ǣ in breaking position, or of gēar, in which ǣ apparently underwent palatal diphthongization.

⁷ This list is the irreducible minimum; several other symbols might be needed, depending upon the mode of analysis.

⁸ We approve their shifting the second fronting of WGmc. a (Luick's ‘zweite Aufhellung‘) to a period after i-umlaut. See further R. Girvan, Angelsaksisch handboek 79, 81, 83 (Haarlem, 1931); and S. Kuhn, The dialect of the Corpus Glossary, PMLA 54.13-6 (1939).

⁹ E.g. K. Luick, Historische Grammatik der englischen Sprache 210 (Leipzig, 1921); K. Brunner, Altenglische Grammatik nach der angelsächsischen Grammatik von Eduard Sievers 90 (Halle, 1951).

¹⁰ Yet there is no such universal effect of i/j upon preceding consonants, as we may see from modern English fielding beside field, in both of which many speakers pronounce a velar l with a velar or centering on-glide.

¹¹ One point they make is of such a nature that an attempt at direct refutation would lead to no very conclusive or worthwhile results. It assumes that linguists are in possession of universal truths regarding languages in general, and that it is possible to say what can and what cannot exist in language. According to Miss Daunt, ‘in existing languages there is, so far as my knowledge and information go, no example of “long” and “short” diphthongs, differentiated only by quantity, being phonemic, and it seems very unlikely that Old English had what is now non-existent’ (110). Stockwell and Barritt, after discussing the vocalic nuclei represented by nama, nām, nearu, and nēar, say: ‘We think it is an accurate statement that no languages have been found that utilize more than three of these four theoretically possible distinctions, or of any other four that might be set up; i.e., three types of syllabic nuclei seem to be a generalizable maximum. On the a priori grounds that OE was a real language and ought to act like one, the set of four distinctions is objectionable’ (5). We wish merely' to offer the following objections. (1) Present linguistic knowledge is probably insufficient for such generalizations. There are literally hundreds of living languages which have not yet been anatyzed. (2) This sort of statement seems to deny the possibility that a language may have a unique feature. (3) A-priori reasoning loses its validity when it comes into conflict with objective data. The statements of Miss Daunt and of Stockwell and Barritt regarding Old English must ultimately stand or fall upon the evidence of the MSS.

¹² On the dates and character of early English and Irish MSS, see E. A. Lowe, Codices latini antiquiores, Pt. 2 (Oxford, 1935); E. M. Thompson, An introduction to Greek and Latin palaeography 371-402 (Oxford, 1912) ; W. Stokes and J. Strachan, Thesaurus palaeohibernicus l.xiii-xxvi, 2.ix-xl (Cambridge, 1901-3).

¹³ R. Thurneysen, A grammar of Old Irish 56 (rev. ed., trans, by D. A. Binchy and O. Bergin; Dublin, 1946).

¹⁴ Ibid. 57.

¹⁵ Loc.cit.

¹⁶ J. E. B. Gover, A. Mawer, and F. M. Stenton, The place-names of Devon 1.xxxiii (London, 1931-2). Because of the scarcity of ME literacy texts from this area (especially texts which are linguistically reliable), it is necessary to depend largely upon place names for evidence of local pronunciation.

¹⁷ Studies in the ME dialects of Devon and London (Göteborg, 1944).

¹⁸ Studies in Old English fractured ea (Lund, 1948).

¹⁹ Frances M. Mack, Seinte Marharete (EETS, OS 193; London, 1934). Further evidence of the same sort will be found in E. Einenkel, The life of Saint Katherine (EETS, OS 80; London, 1884), and in S. T. R. O. d'Ardenne, Seinte Iuliene (Paris, 1936). Examples used here have been verified in Ragnar Furuskog, A collation of the Katherine Group (MS Bodley 34), Studia neophilologica 19.119-66 (1946-7).

²⁰ This discussion, limited to two OE dialects and their ME developments, makes no claim to completeness. An examination of other dialects would reveal further evidence; for example, breaking took place in Kentish as well as in West Saxon, and the Kentish diphthongs may have been preserved as late as the 14th century in such forms as zyalde ‘sold’, hyealde ‘hold’. See R. Morris, Ayenbite of Inwyt (EETS, OS 23; London, 1866).

²¹ See, for example, R. Jordan, Handbuch der mittelenglischen Grammatik 86-8 (Heidelberg, 1934); Luick, op.cit. 333.

²² Bohman, op.cit., and B. Sundby, The dialect and provenance of the Middle English poem The Owl and the Nightingale (Lund, 1950).

²³ British Museum MS Cotton Nero D.4. Incidentally, their suggestion that previous scholars were unaware of ea-spellings after c, g, sc in the Lindisfarne Gospels is unfair to the ‘handbooks’ which they appear to find deficient. The phenomena in question are dealt with in Brunner-Sievers 63-4; Luick 161; Girvan 68-9; K. Bülbring Altenglisches Elementarbuch 62-3 (Heidelberg, 1902).

²⁴ The three glossaries mentioned appear in Corpus Christi College, Cambridge, MS 144; MS Épinal 17; and MS 42 of the Amplonian Library at Erfurt.

²⁵ In words of the type represented by cæster, gæt, and sceaft, 8th-century Anglian texts regularly have æ (also spelled ae and ę) ore; 9th-century Mercian has regularly e. The word entered in dictionaries as cæfl is attested (in early Mercian and late WS glossaries) in an instrumental form aæfli or caefli, apparently from *cafuli and, if so, having secondary i-umlaut. An early WS nominative reconstructed from this might be *cœfl or *ceafl or *cafol.

²⁶ These statements apply to the WS and Mercian dialects; the distinction was retained longer in Northumbrian and Kentish. In Northumbrian, on the other hand, the scribes did not distinguish ea and eo, although they rather consistently preserved the distinction between æ and e.

²⁷ Daunt 132; Mossé 31-2, 41.

²⁸ The relevant portions of this MS are accessible in S. J. Crawford, The Old English version of the Heptateuch (EETS, OS 160; London, 1922). Crawford studies the significant forms in an appendix to this edition, and in Trans. Phil. Soc. 41-7 (1917-20) ; see also Randolph Quirk, On the problem of morphological suture in Old English, MLR 45.1-5 (1950).

²⁹ This can be observed sporadically in the codices of OE poetry; e.g. freom for from ‘valiant’ (Exodus 14), weorn for worn (Azarias 185), weordum for wordum (Phoenix 425).

³⁰ There is a parallel case in the i-umlaut of WGmc. u, which in a specific context produced a predictable variant of u, written y and phonetically distinct from the then existing sounds in the OE phonological system. Although minimal pairs developed (e.g. mund : mynd), they are rare, and it is easier to demonstrate the phonemic identity of y by the use of analogous pairs, such as tyrnan : murnan. On this method of analysis, see B. Bloch, A set of postulates for phonemic analysis, Lg. 24.30-2 (1948); id., Studies in colloquial Japanese, IV. Phonemics, Lg. 26.86-125 (1950); K. L. Pike, Phonemics (Ann Arbor, 1947); id., Tone languages 50-4 (Ann Arbor, 1948). As Bloch points out (Lg. 26.96), ‘Although minimally different pairs are not necessary to prove that a given phonemic difference is distinctive in a particular language, they illustrate such differences more strikingly than other examples.‘

³¹ Stockwell and Barritt attempt to explain away some of the pairs made possible by metathesis (31-2). Their first argument is a restatement of the contention that the ME development shows the sounds undifferentiated, a matter already dealt with in this article. Their other arguments are equally weak. They allow a single late spelling earn (for xrn) to carry as much weight as an abundance of early and irreproachable spellings with æ. They argue that cærse and gævrs were so spelled because the scribes wished to distinguish the unpalatalized c and g from the palatalized varieties; yet the scribes made no such effort in scores of similar cases, such as cēne ‘bold’ and genge ‘going’.

[A reply to this article by Stockwell and Barritt will appear in a later number of the journal.]