Remarks on the ‘Anglo-Frisian’ Thesis (1995)

Remarks /1 August 2018/1 [Updated; revised and added sections in square brackets and blue] Remarks on the ‘Anglo-Frisian’ Thesis Patrick Stiles, University College London 1. The Close Relationship of English and Frisian 1.1. English and Frisian are generally regarded as being more closely related to each other than to any other Germanic languages.1 This view has been current from the beginning of modern scholarship, and, indeed, before (as documented, for example, by Bremmer 1982, which is a useful survey of opinions; Nielsen 1981a provides detailed research history[; note also Sjölin 1973]). In the early period of modern scholarship, Sweet (1876: 562–69) and, above all, Siebs (1889, 1901) can be mentioned. In more recent decades, the same message has been repeated from the various linguistic sub-systems. For phonology and morphology, there is Nielsen’s monograph of 1981a (second edition 1985; cf. also 1986). In the field of vocabulary, there are the extensive studies of Löfstedt (1963–1969) and Lerchner (1965; but cf. Århammar 1989: 104 Anm. 2), stray notes by Collinson (1922, 1950, 1960, 1969), and the more specialized discussion of legal terminology by Munske (1970, 1973). Similarities have also been found in alliterating paired formulas (Bremmer 1982: 83–85) and on the syntactic level (Hofmann 1982). 1 This article is based on a talk I gave at the first Föhrer Symposium zur Friesischen Philologie in October 1991. Nils Århammar, Rolf H. Bremmer Jr., Margaret Laing, D. R. McLintock and Keith Williamson have kindly discussed aspects of this article with me; naturally, they are not responsible for any (mis-)use to which I may have put their advice. I would also like to thank Volkert Faltings and Hans Frede Nielsen for their friendly forbearance and understanding while waiting so long for the manuscript. Remarks /1 August 2018/2 1.2. I list below some of the Old English–Old Frisian shared phonological and morphological features that are typically adduced. (1) Front reflex of PGmc. long e1: OE (WS) dǣd, (Angl.) dēd, OFris. dēd(e) ‘deed’; OHG tāt. (2) Fronting of WGmc. short a : OE dæġ OFris. dei ‘day’; OHG tag. (3) Rounding of PGmc. long e1 before a nasal: OE, OFris. mōna ‘moon’; OHG māno. (4) Rounding of WGmc. short a before a nasal: OE, OFris. lond ‘land’; OHG lant. (5) Loss of nasal with compensatory lengthening before a homorganic spirant: OE tōþ OFris. tōth ‘tooth’, OE, OFris. *gōs ‘goose’ (cf. MWFris. goes Wang. gôs): OHG zan(d), gans; OE, OFris. fīf ‘five’: OHG fimf; and OE, OFris. ūs ‘us’: OHG uns. (6) Rounded reflex of PGmc. long nasalized ã: OE, OFris. brohte ‘brought’; OHG brahta. (7) ‘Breaking’: OE reoht (rare2) OFris. riucht ‘right’; OHG reht. (8) Reflexes of vowels in unaccented syllables: OE, OFris. mōna ‘moon’ OHG māno; OE, Fris. sunne ‘sun’ OHG sunna. (9) Palatalization of velar consonants: E cheese, OFris. *tsēse (cf. MWFris. ts(j)iis Wang. sîz); G Käse. Also OE dæġ, OFris. dei; G Tag ‘day’. (10) Uniform plural ending in present present (and preterite) indicative verb paradigms: OE berað, OFris. berath; OHG berumes, birit, berant ‘we, you, they carry’. 2 OE reoht is a rare form because the diphthong was liable to monophthongization and raising (to reht and riht) by the changes of Anglian smoothing and palatal umlaut that were operative at the time of the early texts. Compare Campbell 1959: §§222, 227, 304–311. Remarks /1 August 2018/3 2. Three Models 2.0. However, I believe that although the DEGREE of relationship between English and Frisian has been extensively discussed, the NATURE of the relationship is in need of further clarification. That is to say: What model must be set up to account for the common features? I shall discuss three models, broadly corresponding to those that have been proposed in the literature, but shorn of some of their historical baggage and perhaps more sharply drawn. 2.1. According to the ‘convergence’ hypothesis, the shared features of English and Frisian are a relatively late secondary creation, the result of changes spreading back and forth across the North Sea after the Anglo-Saxon settlement in Britain, and affecting also (parts of) Old Saxon. This idea was first floated by Hans Kuhn (cf. 1955a and b, 1957). 2.2.0. The other two proposals conceive of the shared features as arising at an early stage, before English and Frisian had emerged as separate languages. These models thus postulate an intermediate proto-language in family-tree terms. 2.2.1. The ‘Anglo-Frisian’ thesis envisages a period of common development that is exclusive to English and Frisian—this view is most closely associated with the name of Theodor Siebs. 2.2.2. What I shall call the ‘Ingvæonic’ model holds that the common features of English and Frisian were not exclusive to them, but were also shared by (at least some strains of) Old Saxon. Consequently, no two of the three languages are more closely related to each other than they are to the third. Later developments may have obscured the participation of Old Saxon in the Ingvæonic changes. One of the longest-standing and most energetic proponents of this view has been Erik Rooth. More recently, it has been advocated by another Swede, Nils Århammar. 2.3. The ‘Ingvæonic’ position can be depicted as in the stemma below (the diagrams in this section only include those West Germanic languages that are extensively attested at the ‘Old’ period and thus exclude Dutch): Siebs apparently conceived of the ‘Anglo-Frisian’ sub-group as immediately subordinate to West Germanic and co-ordinate to Old Saxon and Old High German, but today the proposal is to be pictured rather as the result of an extended period of shared development continuing for some time after separation from Old Saxon brought the Ingvæonic stage to an end. This can be represented as a familytree as follows: Remarks /1 August 2018/4 The ‘convergence hypothesis’ is clearly incompatible with the ‘AngloFrisian’ position, but is not incompatible with the Ingvæonic model. Convergence can be depicted as a modified family-tree diagram, although the immediate antecedents of English and Frisian (and Old Saxon) are not an essential part of the theory. If one views the stemma as representing relationship in abstract space, the lines of descent can be made to converge for the period of common development. 3. Methodological Observations 3.0. Before evaluating these models, it as well to consider what is necessary to establish linguistic sub-grouping (cf. Hoenigswald 1966). 3.1. As regards ‘shared features’ (cf. §1.2), it is QUALITY that is important not just QUANTITY. Data become evidence in the light of a hypothesis. Large amounts of data have been assembled, but we must consider what precisely they might be evidence of. I do not think sub-groupings are to be established by the collections of assorted unanalyzed ‘shared features’ that have often been presented in the literature. Given general tendencies of linguistic evolution, similarities are to be Remarks /1 August 2018/5 expected from the further development of cognate linguistic systems (they share the same starting-point). Moreover, in the case of the Germanic languages and in particular of the West Germanic languages, there was a lot of scope for secondary contact. What matter are common innovations rather than mere similarities or resemblances (cf. Hoenigswald 1966: 6; Hock 1986: 442). The acid test of a period of common development, I would submit, is an ORDERED set of innovations shared by the languages in question (cf. Watkins 1966: 32: ‘The subgrouping question is partly one of relative chronology’; also Cowgill 1970: 114). Rigorous attention to relative chronology reveals that many of the similar changes in sister dialects are in fact separate (a convenient recent example is Kortlandt’s note on ‘Breaking’ in various Germanic dialects, 1994: 15). In order to establish genetic relationship in a family-tree sense, the common changes must be earlier in the relative chronology than any significant separate developments. If the common changes are later, then it is a case of convergence. Although common innovations are of prime importance, the common retention of an archaic feature CAN be a valid dialect criterion—unless restricted to non-contiguous relic areas, of course. Further, developments that create something new (e.g. a new class of phonemes produced by secondary split) are likely to be more significant than simplificatory changes (e.g. frequently observed conditioned sound-changes, loss of categories). 3.2. This leads to the question of the relative value of criteria from the different linguistic sub-systems. The two most important in my view are: • Unconditioned sound-changes affecting accented syllables; • Morphological innovations affecting inflectional endings or patterns of wordformation. The primacy of phonological criteria, at least to establish the family in the first place, results from ‘the fact that the classical “comparative” method… applies to sound-change only’ (Hoenigswald 1966: 6–7, cf. 1973: 26). Further, it is traditionally pointed out that phonology is the most closed and structured subsystem of language and least susceptible to accept borrowed material. This is less true of morphology, syntax and lexis, in ascending order. The importance given to morphological changes is perhaps somewhat unconventional (but cf. Leumann 1959: 390). However, they can be significant, provided of course that they are implemented with etymologically identical phonological material. Thus, the Germanic family of languages can be defined not only on the basis of the phonological criterion of the consonant shift (Grimm’s Law), but also on the basis of the morphological innovation of the dental preterite (interestingly, these features are picked out by Hock 1986: 450, also 578, as defining Germanic; similarly Anttila 1989: 303). Compare Cowgill’s discussion of Italic and Celtic, where three out of his five common Italo-Celtic innovations are morphological (1970: §83 and fn 30; a sixth feature that he added later is also morphological, 1973: §19). Remarks /1 August 2018/6 3.3. Finally, it must be remembered that our hypotheses are dependent on the quality and amount of data available. We have a duty to try to interpret the data that we have, even if they may be limited and appear to be contradictory. It is incumbent upon us to produce the simplest and most accurate interpretation of the facts as we know them (the principle of Economy). 4. The Convergence Hypothesis The ‘convergence hypothesis’ (cf. §2.1, above) is a valid theoretical position, but in my opinion, it does not fit the present case. It sets out to explain the common features of English and Frisian as the result of changes spreading across the North Sea after the Anglo-Saxon settlement in Britain. As I present it here, an essential element of the model is that the two languages had already diverged somewhat before they acquired the shared features. Kuhn himself actually keeps his options open, being prepared to place changes before, during and after the Anglo-Saxon invasion (cf. 1955a: e.g. 40–41, 44). However, in so far as any shared features may pre-date the settlement, they do not need the thesis of the linking role of the sea in order to explain them. Hence my particular framing of the model here. It intentionally leaves out of account such relatively late phenomena as the unrounding of y to e in the Kent–Essex–Suffolk region, which might best be seen in connection with similar developments across the channel. In fact, the driving force of Kuhn’s theory is his late absolute dating of the end of his North-West-Germanic speech community. This led him to propound the idea that English and Frisian could have undergone the special developments they have in common even as late as after the settlement of England. (For Kuhn’s view on the participation of Old Saxon in the changes, cf. 1955a: 24, 27, 36–43, 46.) The convergence hypothesis can be falsified by the fact that we are not confronted in Old English by a situation where the shared features listed in §1.2 fan out from the littoral facing Frisia and then dwindle as one gets further away from this region. Old Northumbrian ought, on this view, to be the Old English dialect that is least Ingvæonic in general and least Frisian-like in particular. This is not the case. Indeed, one can cite a feature in common, the loss of word-final -n, although I would not set much store by it. If the changes spread better by sea, then it is odd that they should have spread inland so efficiently. Are we to suppose that there were active and intensive links between the general populations of, say, Worcester and Merseburg? In order to rescue the thesis of the linking role of the sea, one has to abandon the idea that the two languages developed their shared features together, for one would have to assume the features had developed and spread throughout Anglo-Saxon England before they moved to the Continent. However, what matters is not where pre-English and pre-Frisian were when they developed their common features—although having them either side of the North Sea rather than side by side on the Continent does seem contrary to the Remarks /1 August 2018/7 principles of dialectology3. The crucial point against the ‘convergence hypothesis’ is that most of the shared features listed in §1.2 occur early in the relative chronology of the two languages’ sound-changes. Much of the evidence for this will emerge in the course of the argument. 5. Aim of the Current Article 5.1. If the convergence hypothesis is wrong, the problem is reduced to a question of whether there is evidence for English and Frisian undergoing an early period of exclusive common development. If so, the consequence is the ‘Anglo-Frisian’ position. If their early shared changes are not exclusive, the Ingvæonic model would appear to be appropriate. The bulk of this article will be occupied with an attempt to disprove the ‘Anglo-Frisian’ thesis. Obviously, I am not the first person to have argued directly against it. Yet, it continues to rear its head, none the less. However, I hope to arrive at the conclusion by a slightly unusual route. My arguments will be runological, as well as more straightforwardly linguistic. The latter will put a chronological slant on data that are usually considered principally from a geographical perspective. If one attempts to reconstruct the relative chronologies of the sound-changes of the two languages (cf. §3.1), a lot of the alleged common features are downgraded. I am not aware that anyone has tried to do this before. 5.2. If one seeks a metaphor for what I seek to do, I would somewhat irreverently adopt the one used by Cowgill when taking up the problem of the relationship between two branches of Indo-European, Italic and Celtic (1970: 114): ‘It might be said that in reopening the grave… I have found neither an empty coffin nor a living adult prematurely buried, but a stillborn infant.’ In the case of ‘Anglo-Frisian’, it is more a case of trying to nail down the coffin lid on Count Dracula, one of the undead, a vampire that won’t lie down and die. 3 The case of Ozeannordisch that Kuhn cites as a supposed parallel is a different kettle of fish. The settlements in question were in general smaller, much more on the coast and oriented towards the sea and ‘home’. Compare Lerchner’s trenchant remarks (1965: 301): ‘die…These von der Sprachraummitte des Meeres…mag für das Ozeannordische zutreffen, aber was in dem einen Fall stimmt, muß…für den anderen, hier das Nordseegermanische, noch lange nicht richtig sein. Die Bedingungen sind hier wohl auch tatsächlich voneinander unterschieden… Die ozeannordischen Übereinstimmungen umschreiben Übereinstimmungen zwischen Mutterland und Kolonie. Die nordseegermanischen Merkmale dagegen betreffen Gemeinsamkeiten politisch, ökonomisch und kulturell durchaus eigenständiger Gebiete. Das Meer trennt oder verbindet nicht abstrakt als physikalisch-geographischer, sondern konkret als soziologisch-historischer Faktor.’ Remarks /1 August 2018/8 6. Runological Arguments 6.0. Before coming to the linguistic evidence, I propose to stray into the field of runology. I will discuss two topics: the so-called ‘Anglo-Frisian’ runes4, and the final -u of Frisian inscriptions. 6.1. ‘Anglo-Frisian’ runes 6.1.0. Anglo-Saxon and Frisian inscriptions are characterized by the use of a runerow which differs from the 24-rune Older Futhark in that: it has more runes; some of the runes have divergent shapes; and certain runes have different values. Common to the Frisian and English traditions is a rune-row consisting of 26 symbols, two more than in the Older Futhark. Only the English tradition provides complete futhorcs and these essentially preserve the order of the core 24 runes (with slight variation from the 20th on). The 25th and 26th runes have the shapes A and a respectively. The shape a is the same as that of the fourth rune of the Older Futhark; the fourth rune of the English Futhorc has the shape O. The Old English manuscript values of these three runes, together with their use in extensive and securely interpreted inscriptions, reveal that they had the following values: a æ, A a, and O o (both long and short). We have information only from Old English sources as to what their names were, so in the following I shall use their English names: a æsc, A ac, and O os. It is a feature of the runic alphabet that the name of a rune gives its value (the acrophonic principle). The conventional doctrine is that these new values arose in direct response to certain ‘Anglo-Frisian’ sound-changes (compare items 1 to 6 in §1.2). If this view is correct, then we have a powerful argument for a period of Anglo-Frisian unity, during which these runes were invented. There must have been an ‘AngloFrisian’, otherwise how do we account for the ‘Anglo-Frisian’ runes? However, I believe there are grave difficulties with this view. 6.1.1. The āc-rune Strangely, it seems largely to have escaped notice [see footnote 6] that there is a problem in assigning ‘Anglo-Frisian’ antiquity to the NAME ‘oak’ of the Old English 25th rune. As the cognate Frisian word for ‘oak’ is ek, the only way it and English ac can be united is under the earlier form *aik- (cf. further §8.4, on the phonological development). This suggests the superficially attractive theory that ‘Anglo-Frisian’ developed special runes for the a-diphthongs, A for *ai and 6 for *au. I think such an assumption is highly dubious; the Old English ‘ea’-rune 6 (number 28 in the extended English futhorc OE ea is the reflex of *au) which is the 4 The whole question of the ‘Anglo-Frisian’ runes will be considered in a joint study in preparation by myself and Christopher Ball, where fuller discussion will be found. forthcoming!] [still Remarks /1 August 2018/9 other main piece of evidence for such a theory does not occur in Frisian inscriptions. Yet the theory that ‘Anglo-Frisian’ developed runes for the a-diphthongs, anyway fails to rescue the situation. This is because the Frisian value of the rune A is a, just as in Old English. It is quite implausible to suppose that at the ‘Anglo-Frisian’ stage a rune was devised to express the value a, but given two different names, depending on which bit of ‘Anglo-Frisian’ was to become English (‘oak’) and which Frisian (some other name—which we do not know—which yielded a word in Frisian beginning with a-). It is only slightly less implausible to imagine that an originally ‘Anglo-Frisian’ name was later changed in one or both traditions. However, if the name is secondary, then the whole supposed rationale for the rune as an ‘Anglo-Frisian’ creation is undermined. It would be possible to salvage a name ‘oak’ of ‘Anglo-Frisian’ date, however, by supposing that the original Frisian form of ‘oak’ was †*āk, but that the vocalism was analogically replaced from the associated adjective of material PFris. *ēkin (OFris. ētsen), from *aikinaz ‘oaken’. (As the gloss shows, the analogy has gone the other way in Modern English). If this is what happened, it did so at the Proto-Frisian stage, as all varieties of Frisian know only ēk for ‘oak’5. One would then have to suppose that the Frisian rune-name was not affected by this development, but became fossilized in the form †āk, so that the value a was retained. All in all, a sequence of events which hardly makes for an unproblematic derivation. These considerations suggest that the A-rune was a specifically English creation subsequent to the sound-change of *ai > ā. The implication must be that at least the value was borrowed by the Frisians; the shape may have arisen as a variant at an earlier stage6. 6.1.2. The os-rune The NAME of the Older Futhark fourth rune, *ansuz, shows up attached to a shape O that—according to the standard view—was developed to express the sound-value o (compare, for example, Page 1973: 44–45). Yet the Older Futhark already had a rune which performed this function, number 24: o * ōþi/ala. There can have been no need for an additional rune to denote o until the completion of the later change 5 It should be noted that Dutch has an ‘Ingvaeonic’ dialectal form for ‘oak-tree’ aak, beside standard eik. It could reflect such an original Frisian form, but this can hardly be described as a certainty. It is better regarded as a non-Frisian Ingvæonic relict (Schönfeld–van Loey 1970: xxxvii and §65a). 6 For further details, see the forthcoming work by Ball and Stiles referred to in footnote 4. [Quak (1990: 358) and Seebold (1991: 508) have expressed the view that the āc-rune was an English invention and loaned into Frisian. Cf. also Hofmann 1995: 32–33.] Remarks /1 August 2018/10 of i-mutation, which gave this shape the value œ (in English at least). And imutation belongs to the individual history of each language, as will be seen below (§8). This objection that a second rune to denote o was unnecesary might be countered by appealing to the principles governing the transmission of runic lore. When a rune-name underwent a sound-change, its value was liable to change too (this is part of the acrophonic principle). In the case of the 24th rune just mentioned, ōþil became œþl by i-mutation. According to this view, it wasn’t a question of the rune-row needing an extra o; one arose willy-nilly as a result of sound-change. When short a before nasal was nasalized and rounded and then lengthened on the loss of the nasal before a homorganic spirant, cf. §1.2 (4 and 5), the outcome was a long ō, as in the Old English rune-name itself: ōs. The trouble with this argument is that the linguistic evidence fails to deliver the goods. It cannot motivate a value o of ‘Anglo-Frisian’ antiquity for the shape O, because, in Old English at least, the merger of *ã from PGmc. short a plus nasal plus spirant with PGmc. long *ō in fact takes place AFTER the relatively late change of i-mutation (compare §8.5, below). [Granted that *ã from PGmc. short a plus nasal plus spirant behaves the same as long *ã from PGmc. long ē before a nasal— and there is no evidence from Old English to suggest they do not—the post-imutation date of the merger of PGmc. long ē1 before a nasal with PGmc. long *ō ] is shown by the fact that their umlaut products have different outcomes when shortened: compare ModE bramble, beside unmutated broom the shrub ‘Sarothamnus scoparius’ (on their relationship, compare NED s.v. bramble) and OE blœtsian ~ bletsian ‘to bless’, where the stem-vowel can only be a mutation product (the verb is usually considered to derive from either blōd ‘blood’ or blōt ‘sacrifice’ by means of a suffix *-isōjan, compare Hallander 1966: 108–133). On the chronology of the merger of *ã of Ingvaeonic origin with PGmc. long *ō in English, cf. Luick 1914–40: §§111 A.1, 185 A.1, 186, 204, 207; Campbell 1959: §§127 fn1,193 (d), 285. Accordingly, there was no motivation at an ‘Anglo-Frisian’ stage for creating a rune to express the sound value o. (The particular shape may have arisen on the continent, but this datum cannot be used as an argument about a common linguistic origin.) Again, these difficulties cast serious doubt on whether this rune was indeed invented at a supposed period of ‘Anglo-Frisian unity’. 6.1.3. A further discrepancy may be mentioned, this time affecting rune-shapes. For the rune for h, the earliest English inscriptions use a single-barred form h, whereas Frisian only knows a double-barred form, as also found in later English inscriptions. This early divergence does not point toward a shared inheritance[, but suggests borrowing]. 1 Remarks /1 August 2018/11 6.2. Final -u in Frisian runic inscriptions 6.2.0. The other runic topic I would like to broach is the problematic word-final -u in a-stem nouns in Frisian runic inscriptions. I list the clear examples here, in order of increasing likelihood that they are a-stem nouns. Westeremden A adugislu a personal name Schweindorf weladu (Welandu) ‘Wayland’ Folkestone-Glasgow æniwulufu a personal name Oostum habuku ‘hawk’, probably a personal name Oostum kabu (kambu) ‘comb’ Toornwerd kobu (kombu) ‘comb’ I leave out skanomodu on a solidus in the British Museum, London, because Hans Frede Nielsen has recently pointed out that it is most likely to be a feminine name and thus an ō-stem nm. sg. in -u (1993: 83–84; see §6.2.3, below). 6.2.1. How one analyses the final -u in the examples cited above depends in part on what case-forms one thinks they represent. The most consistent interpretation, and—in terms of the Frisian runic corpus—the most plausible, is that the final -u in these forms is the phonological reflex of PGmc. *-a(-) in word-final position. The -u in all these forms could thus continue nominative singulars in PGmc. *-az and accusative singulars in PGmc. *-an. This has been suggested by Wolfgang Krause (1968: §34 A), Klaus Düwel (in Düwel–Tempel 1970) and Hans Frede Nielsen (1984). A nominative analysis seems inescapable for Toornwerd, at the very least, where ‘comb’ is the only word on an object that happens to be a comb. I said just now that it was the interpretation that was most plausible in terms of the Frisian runic corpus. But if one looks beyond this narrow corpus, difficulties arise. It seems to me surprising that PGmc. *-a should be preserved as a phonological entity in these inscriptions, even as a ‘Murmelvokal’ (as Krause and Nielsen propose). Surprising in several respects, I would suggest. (1) If it were not for this interpretation of these few Frisian runic forms, the loss of PGmc. final short *-a would surely be placed as an early ‘West Germanic’ change. (2) Further, the loss of PGmc. *-a(-) in West Germanic languages is a change that occurs regardless of the length of the preceding syllable, yet Nielsen considers that position after a heavy syllable (or its metrical equivalent, two light syllables) is a conditioning factor for the u-reflex, 1991a: 301. (I do not propose to consider here the interpretation of medial o and u as composition vowels.) (3) Yet this apparent extraordinary archaism appears together with the characteristically—though not exclusively—Frisian sound-change of the monophthongization of PGmc. *au to ā, as in adugislu < *auda- on Westeremden A. (4) In addition, this seeming reflex of PGmc. short -a occurs, within the corpus as a whole, ‘alongside’ apparently shortened PGmc. long vowels which, logically, one would expect have shortened only after the loss of PGmc. short *-a. An example is provided by the masc. -an- stem nom. sg. in -a < PGmc. (trimoric) long -ô, as in the personal name hada on the Harlingen solidus. This is not an Remarks /1 August 2018/12 insurmountable difficulty, in so far as long vowels could shorten to -a while original short *-a is retained as a Murmelvokal, though, being more complex, it is hardly an ideal solution. (5) What is more, the spellings of the Frisian runic corpus suggest PGmc. *-a had merged with ‘WGmc.’ *-u, witness -h[i]ldu (on which, see §6.2.3, below) and habuk-, which is the interpretation Nielsen proposes 1984: 17–18). But this is counter to what we think we know about the development of the ‘Ingvaeonic’ unaccented vowel system, where -a and -u are kept apart (Nielsen reports my view 1991a: 302 fn3). Thus, ‘literary Old Frisian’ as recorded, for example, in Riostring manuscripts does not seem to reflect a form of language in which this merger took place. 6.2.2. On an alternative view, some of the final -u’s in supposed a-stem nouns are etymological instrumental endings (cf. e.g. Beck 1981: 74–75; Nielsen 1984: 13; Insley 1991: 174). Thus, for example, Welandu would mean ‘(made by) Wayland’. A possible candidate is op hÆmu ‘auf der Heimstätte’ on Seebold’s reading of Westeremden B (Seebold 1990: 421); although he regards it as feminine dative singular, the word seems to be masculine/neuter in Frisian, as in the other Ingvæonic languages and in North Germanic. An ‘instrumental’ singular in -u would constitute a morphological feature that points away from Old English and towards Old Saxon and Old High German. Both these languages have ‘instrumentals’ in final -u in a-stem-nouns, but there is no sign of this ending in Old English, where the fifth case ends in -i. 6.2.3. Other instances of word-final -u are found in feminine (j)ō-stem nouns. In the nominative singular of this class, it is the regular reflex of PGmc. -ō, which yielded *-u in the North and West Germanic languages, compare OE giefu. A possible example in the Frisian runic corpus is the skanomodu of the London solidus (cf. above, §6.2.0). In another instance in the corpus, final -u most probably denotes the dative singular of a jō-stem noun. mþ Gisuhldu (read as miþ Gisuhildu) Westeremden A The ending is usually derived from the Proto-Germanic instrumental singular in long -ō. Again, this is the form of the dative found in Old Saxon and Old High German (and Old Norse), whereas in Old English and Old Frisian the dative singular of o-stems ends in -e, occurring alongside -a in Old Frisian (cf. Nielsen 1984: 13–14). [On this topic, note Nedoma’s and Schuhmann’s articles of 2014, with references.] 6.2.4. It may be pointed out that, in so far as the corpus of Frisian runic inscriptions shows the distinctively non-English sound-change of *au to ā, it cannot readily be argued that the masculine and feminine datives in -u are representative of a form of language ancestral to both English and Frisian. 6.3. I turn next to linguistic arguments. These will be from the realms of morphology and phonology. Remarks /1 August 2018/13 7. Morphology 7.1.0. As I stated above (§3.2), I think shared morphological features can be strong evidence for establishing linguistic sub-groups. However, this criterion turns out not to yield very much in the context of English and Frisian, as there appear to be only three uniquely shared morphological features. Three statistically a lot better than two. 7.1.1. One is the innovation of the -a ending of the nom.-acc. pl. of u-stem nouns, thus OE, OFris. suna ‘sons’, replacing the Proto-Germanic ending *-iwiz, reflected in Go. sunjus, ON synir, OS suni. Striking though this morphological innovation is, it is difficult to fit into a chronology, so one cannot rule out the possibility that the ending was borrowed from one language into the other. 7.1.2. Another uniquely shared morphological feature is what is presumably the identical selection of the ending -e for adjective-adverbs: thus OE swiþe, OFris. suithe, beside OLFrk suitho, OS swido ‘very’, to the adjective seen in Go. swinþs* ‘strong’; also ON giarna, OHG gerno ‘eagerly, willingly’, to the adjective seen in OHG gern ‘eager, willing’. Still, it is possible, although unlikely, that the identical selection was made independently. [7.1.3. The third is the form of the conjuction AND: OE and and OFris. and, with *a-stem-vowel and no ending-vowel (cf. Klein 1996: 399). ] 7.2.0. On the other hand, some disagreements can be cited. 7.2.1. The neuter singular of the demonstrative pronoun ‘this’ is þis in Old English, but Old Frisian and the rest of continental West Germanic show a form *þit. I am more concerned here to note the distinction, than to discuss its origin. 7.2.2. The dative singular forms in final -u to o-stem nouns in Frisian runic inscriptions (§6.2.3, above) coincide with these case-forms in Old Saxon and Old High German. Likewise, if at least some of the examples of the mysterious final -u in a-stem nouns (§6.2.2, above) are to be interpreted as instrumental singular endings, this also points to Old Saxon and Old High German, which both have instrumentals in final -u in a-stem-nouns. In Old English, the fifth case ends in -i. 7.2.3. For ‘am’, Frisian has bim (~ bem, as in Old High German and Old Dutch), whereas all Old English dialects show — alongside other formations — reflexes of *bium, a form English exclusively shares with Old Saxon. Both West Germanic types are innovations, resulting from the contamination of two verb-roots. On the one hand, bim is PGmc. *im (as reflected in Gothic and Old Norse, from the athematic verb IE *ésmi) with pre-fixed b- introduced from the thematic verb *biu (from the IE root *bhuX-; X denotes an unspecified laryngeal). On the other hand, *bium is *biu with -m adopted from *im. (West Saxon Old English eom in turn is inherited *im remodelled on *bium; non-West Saxon (e)am is free of contamination with *bhuX-, having instead been reshaped on 2. sg. (e)arð, respectively eart in the case of Kentish.) Remarks /1 August 2018/14 7.2.4. Old English and Old Frisian display a morphophonological difference in the way they have reacted to the effects of fronting of short *a in noun paradigms. In the a-stem nouns, the Old English reflexes are phonologically regular, whereas Old Frisian has levelled the fronted vowel of the singular cases into the plural. m. OE OEFris. dæg dagas dei degar ‘day’ nt. fæt fatu fet *fetu ‘vat’ (On evidence for the OFris. pl. *fetu, cf. Löfstedt 1932: 11.) In the ō-stem nouns, by contrast, Old English largely eliminates the fronted stemvowel (cf. Campbell 1959: §589.1), while Old Frisian favours it: OE sacu, OFris. seke, by-form sake (cf. van Haeringen 1920: 44). The levellings are clearly distinct. However, the disagreement can be explained on the assumption that the phonetically conditioned outcomes of short afronting remained undisturbed until the end of a putative ‘Anglo-Frisian’ unity, with any levellings only taking place after the split into English and Frisian. That this may be the case is suggested by reconstructed OIsNFris. pl. *dagar, which underlines the forms of the modern dialects, unless its stem vocalism is the result of interference from Old Danish. At any rate, the fact that the two languages have levelled differently does not lend support to Kuhn’s ‘convergence’ hypothesis. 7.2.5. The Old English nom.-acc. pl ending of masc. a-stems is -as (from the etymological nominative), but there is no trace of this in Old Frisian. There, the most archaic ending is -a (from the etymological accusative, also found in Old High German—and in Old Saxon, in addition to the -os/-as ending corresponding to that of Old English). Other Frisian endings I regard as later replacements. But as this distribution is the result of the different selection of post-West Germanic variants, it must be relatively late and could easily be ‘post-Anglo-Frisian’. Again, the disagreement does not help Kuhn’s position, for if he were right, it is precisely in this kind of thing that similarities would be expected. [7.2.6. Old English has *hē2r for “here”, whereas Frisian appears to reflect only *hīr.] [7.2.7. The suffix vowel of the past participal passive of strong verbs differs, with Old Frisian having generalized -i- and Old English -a- (cf. Seebold 1967). It is true that this difference is the result of independent generalizations, but they were largely complete by the time of palatalization and i-mutation in Frisian at least. Thus OFris. bretzen “broken”; fenszen “gefangen”, genszen “gegangen” beside OE brocen, fangen, gangen. There is no hint of palatalization in Old English. Old Saxon shows both types of suffix vowel, cf. Gallée 1897.] Remarks /1 August 2018/15 7.3. These morphological features do not lead us very far. In fact, they pull in opposite directions. Matters would be clearer if it could be established that the innovation of the u-stem nom.-acc. pl. in -a was early. Anyway, I do not think we could establish a sub-group on the strength of this one feature alone. 8. Phonology 8.1. Preliminary Relative Chronology An attempt to reconstruct a common relative chronology (cf. §3.1) of some of the shared phonological features given in §1.2, produces the preliminary conclusions depicted below. I add i-umlaut. Preliminary relative chronology of features given in §1.2 (plus i-umlaut) (1) Long and short a develop back timbre and nasalization before a following nasal consonant. Before change (2), otherwise we would expect them to be fronted by it: *māno > *mãno; *land > *lãnd; *gans > *gãns (a) Some time thereafter, short vowel plus nasal plus homorganic spirant yields long nasalized vowel plus homorganic spirant: *gãns > *gãs; 3pl. *-ãnþ > *ãþ (cf. §1.2 [10]) (b) The development of the uniform plural probably follows (a), and is to be put early for reasons of linguistic geography (cf. Århammar 1990: 11). (2) Fronting of long and short a. Before change (3), because *k and *ʒ are palatalized before the reflexes: *dād > dǣd/dēd; *dag > *dæg (a) Short a (only) was also fronted in unaccented syllables. This left the way clear for unaccented o to unround to a (cf. §1.2 [8]) (3) Palatalization. Before i-umlaut (or at least the unrounding of i-umlaut products): *kǣsī > *cǣsi; *gæf > *ġæf; *dæg > *dæġ (4) i-umlaut: *kuning > *kyning (> OE cyning, OFris. kening) 8.2. Breaking However, the situation alters when we bring in the English and Frisian changes known as breaking (§1.2 [7]). In the relative chronology of Old English sound-changes, breaking is demonstrably earlier than palatalization as a phonemic change (Luick 1914–40: §637 A8). Consider WGmc. *kerl (cf. G Kerl), which yields OE čeorl ‘churl’ by breaking, NOT †čierl by palatal diphthongization, a change clearly dependent on the prior existence of palatalization. (Old English palatal diphthongization, for its part, is a change without a counterpart in Old Frisian.) The Old Frisian reflex is tserl. In Frisian, by contrast, breaking is to be dated later than i-mutation (cf. Campbell 1939: 105). Frisian breaking affects the short vowels e and i and only Remarks /1 August 2018/16 takes place before tautosyllabic _h (cf. Århammar 1960: 285, Hofmann 1972–73: 61)7. So, the form siucht ‘sees’ (from earlier *sihith) only has conditions for breaking as a result of apocope of i. It should be pointed out that the third singular verb ending -t, from and beside -th, is a relatively late phonetic development. Still, since _hs is a breaking environment (cf. OFris. *miuchs ‘dung’ Wang. miux; *thiuchsel ‘adze’ Wang. thiuksel, ModWFris. tjoksel ‘adze’), there is no problem in _hth being one. There are significant differences in the operation of ‘breaking’ in the two languages as well. Frisian breaking only affects the short vowels e and i. In Old English, the change affects all front vowels, both long and short, and takes place in many more environments; further, in the Anglian dialects, retraction rather than breaking takes place in some of the environments. Thus, in Old English, the stemvowel of WGmc. *arm was first fronted to yield *ærm which was then broken to 7 OFris. tsiurke ‘church’ is usually regarded as showing Proto-Frisian breaking. I doubt this, however. It would be the sole example of _rk causing the chang e (Siebs 1901: §28). In all other instances, tautosyllabic _h is the only environment (assuming, as the evidence suggests, that the metathesis in *threh < *therh ‘through’ is Proto-Frisian). If _rk were a Proto-Frisian breaking environment, one might expect such a word as *werk to be affected by the change. Further, Proto-Frisian breaking is otherwise represented consistently in the reflexes of lexemes within the Frisian daughter idioms, whereas the word for ‘church’ appears in the Frisian dialects in a variety of shapes. In terms of dialect geography, it is suggestive that the diphthongal form in question is found only in Old Ems Frisian (tsiur(i)ke, and the tsurke developed from it, so Löfstedt 1960b: 50) and in Mainland North Frisian (Löfstedt 1931: 140, 141 fn 3), which was taken there by colonists from the Ems Frisian area. In the most southerly Mainland North Frisian dialect, that of the Südergosharde, særk continues a form without diphthong; similarly, some Old Ems Frisian texts attest stray forms without diphthong (cf. Löfstedt 1960b: 50). The other branches have forms without diphthong (or loan-forms). In Island North Frisian (which probably also came from the Ems Frisian area), Sylt. säärk (C.P. Hansen) and Ferring-Amring sark point to earlier *tserk- (Helgolandic kârk is from Low German). In Weser Frisian, R1’s sthereke is reflected in Wursten Schiräck, Schreek, Wangeroogic sjirik). Modern West Frisian tsjerke reflects *tserk-. Indeed, Saterlandic, modern Ems Frisian in a peripheral and isolated position, disagrees with the rest of Ems Frisian, having sérke (< *tserk-). Cadovius-Müller’s Harlinger tzierck is unclear between earlier *i and *e, but it does not show ‘breaking’ at any rate (cf. riucht). Thus, it appears that the Frisian pre-form everywhere had the stem-vocalism -e-: *tser(i)ke. In the central Ems Frisian area, this developed post-Proto-Frisian to -iu-, later -u-. Marginal areas retained the stem-vowel -e-. The following proximate pre-forms can be set up: WFris. *tserk-, EmsFris. etc. *tser(i)k-, WeserFris. *tserek-, Saterl. *tserk-. (The forms with or without medial -i- may have varied within the paradigm.) It seems that, rather than showing Proto-Frisian breaking, Old Frisian tsiurke is the result of a secondary development in parts of Ems Frisian. It took place after the migration of the Island North Frisians (7th and 8th centuries) and the Saterlanders (probably in the 10th century, cf. Århammar 1969: 52, 59–60), but before that of the Mainland North Frisians (after about 1050). There seems to be no reason to suppose that the word was loaned from, or influenced by, OE cyr(i)ce as Löfstedt suggested (1960a and b) and Krogmann accepted (1960). Palatalization initially but not medially would be regular for indigenous Frisian *tser(i)ke, developed from ‘WGmc.’ *kirikon-. If the form were borrowed from Old English, then dissimilatory loss of the medial palatalization has to be invoked, cf. Århammar 1984: 149, note 14. Remarks /1 August 2018/17 earm in the West Saxon and Kentish dialects, but retracted to arm in the Anglian dialects (for fuller details, compare Luick 1914–40: §§133–159). In Frisian, however, it underwent no significant change after a-Fronting: OFris. erm. [The following difference can also be noted: WGmc. *sehan develops as follows in Old English: > *seohan (Bkg) > *sēon. In Old Frisian the development is this: *sea (loss of -h-) > sja. As English and Frisian breaking turn out to be quite independent changes, it would appear that a relative chronology of sound-changes common to the two languages can progress no further than point (2) in §8.1. 8.3. Palatalization 8.3.1. An Anglo-Frisian period of unity looks even shakier when palatalization is considered more closely. The change is clearly separate in Old English and Old Frisian (compare respectively Luick 1914–40: §637 esp. A8 and van Haeringen 1920: 37). The difference between OE ceapian ‘to trade, buy’ (with palatalization) and OFris. kapia ‘to buy’ (without) — both ultimately derived from Lat. caupo “tavern-keeper; hawker” [cf. Wissmann 1938: 20–25] — demonstrates that palatalization took place in each language after changes which are unique to it (cf. §8.4). Moreover, the conditioning factors for palatalization were not the same in both languages. For example, they differed medially in the case of PGmc. *k: OFris. tser(e)ke ‘church’ (etc., cf. footnote 7), dik m. ‘ditch’, rike ‘rich’ beside OE čiriče, dič, riče. Note, however, the remarks of Århammar 1984: 139–40. Again, if there was an ‘Anglo-Frisian’ intermediate proto-language, it may be asked why the two languages have different lexical distributions. For example, ‘lark’ has no palatalization in English, but does in Frisian. [Conversely, OE þæč “roof” shows palatalization, but Old Frisian thek does not.] 8.3.2. Further, palatalization as merely a phonetic tendency cannot be pushed back to an ‘Anglo-Frisian’ stage, as has been pointed out by Richard Hogg (1979: 90– 91). Phonetic palatalization of the velars in Old English was followed by phonemic split in the case of the stops k and g and the voiced fricative ʒ. Palatalization also affected the voiceless fricative h, although here it did not lead to phonemic split (cf. Hogg 1992: §2.60 and references). However, because breaking only took place before velar consonants, the voiceless fricative h must have been unaffected by any phonetic palatalization in English at the time of the earlier change of breaking (above §8.2). This is shown by the following examples, in which the conditions for palatalization existed before the operation of breaking (Bkg). pre-OE *wīh > Bkg *wīoh > OE wēoh ‘idol’ pre-OE *nǣh > Bkg *nǣah > OE nēah ‘near’ Remarks /1 August 2018/18 pre-OE *hlæχχijan > Bkg *hlæaccijan > OE hliehhan (with i-mutation) ‘to laugh’ In the verb ‘to laugh’, where the velar fricative is in a palatalization environment, there is even a following i-mutation factor.8 8.4. Frontings and Monophthongizations 8.4.0. The difficulties increase when we consider some sound-changes that were not given in §1.2, for the very good reason that they are different in English and Frisian. These are the development of the Germanic a-diphthongs *ai and *au. Typical Old English and Old Frisian reflexes are given below. WGmc. OE OFris. *braida- ‘broad’ brād brēd *brauda- ‘bread’ brēad brād The problem is that this divergence cannot be put at a later stage than the changes given in §8.1. The following argumentation is based on that of Campbell 1939: 90–91 (which is largely derived from Tolkien, and is repeated in Campbell 1959: §132 ). I shall present what I believe remains valid. 8.4.1. We can begin with the point that ‘Germ[anic] au was clearly not monophthongized in a period of Anglo-Frisian unity, for it remains a diphthong in O[ld] E[nglish]’ (Campbell 1939: 91). I should probably also mention here the runic inscription found at Caistor-by-Norwich, Norfolk, dated to the fifth-century and bearing the legend raïhan ‘roe(-deer)’, with what certainly seems to be a diphthongal spelling. The main thrust of the argument is as follows. The Old English reflex of Gmc. *au, which—despite the spelling ‹ea›—is to be analysed as /æa/, can only be plausibly explained on the basis of fronting of short a to æ. Analogously, the monophthongization of *ai to OE long ā is best understood if it precedes the fronting of short a. Yet, the long ā reflex presumably arose later than any fronting of WGmc. long *ā, otherwise the monophthongization product would be subject to the fronting. Welcome confirmation of the relative chronology 1. fronting of long ā, 2. monophthongization of *ai, 3. fronting of short *a is apparently offered by the stemvowel in the ‘z-cases’ of the Old English demonstrative pronoun: þær- (which has short stem-vowel in all varieties of Old English, not just in the ancestor of the 8 Hogg 1979: 91 uses as his example feoh ‘wealth; livestock; money’, ‘from earlier *feh’. This word is clearly an a-stem neuter in Old English (cf. Sievers–Brunner 1965: §75), but was a Germanic u-stem: *fehu (Go. faihu, OS fehu, OHG fihu; OFris. fia still bears traces of being a ustem, cf. Siebs 1901: §§133, 155. IV). It might be objected that, if the Old English word was still a u-stem at the time of breaking, the velar quality of the h was due to the following -u: *fehu > Bkg *feohu ® feoh (by change of stem-class). Remarks /1 August 2018/19 Vespasian Psalter language, where gen. pl. ðeara has back-mutation of a short vowel). Derivation of this form from PGmc. *þaiz- (cf. ON þeir-) would appear to require the ordering: *ai > long *ā > short * ă > æ by fronting. However, there is an alternative derivation, provided it is licit to posit proclitic use of the demonstrative (as a definite article) with concomitant loss of stress early enough: [but probabaly is not, as eOE still a demonstrative] *ai > long *ē in unstressed position > shortened to æ. When shortened, long *ē in unstressed position would have have been assigned to the æ-phoneme, as the Ingvæonic unstressed vowel system lacked a short ephoneme. 8.4.2. In Old Frisian, the two a-diphthongs develop in parallel, to the extent that both are monophthongized, so it is natural to place the monophthongizations together in a relative chronology. On the basis of the argument concerning *au and *ai in Old English, the Frisian long ā-reflex of *au might be best accounted for by putting the fronting of short a after the monophthongization in Frisian. It does not seem necessary to explain the reflex of *ai as long ē (or earlier long *ǣ) by placing the fronting of short a earlier, as such a monophthongization product of *ai is fairly common. I will just point out that in monophthongizing both a-diphthongs, Old Frisian shares a feature with Old Saxon against Old English [cf. Hofmann 1995: 28]. Indeed, some forms of Old Saxon even share the characteristic Frisian monophthongization of *au to long ā, beside more common OS long ō (on the extent to which the two monophthongization products may be regarded as ‘the same’, cf. Århammar 1990: 11 fn2 ). Accordingly, we cannot have a shared relative chronology beyond point (2) in §8.1. 8.5. Revised Relative Chronologies The foregoing considerations lead to revised relative chronologies for the two languages as follows (compare §8.1). 8.5.1. Revised relative chronology for English (1) Long and short a develop a back timbre and nasalization before a following nasal. Before change (2), otherwise we would expect them to be fronted by it. (a) Some time thereafter, the sequence short vowel plus nasal plus homorganic spirant yields long nasalized vowel plus homorganic spirant. (b) The development of the uniform plural probably follows (a), and is to be put early for reasons of linguistic geography Remarks /1 August 2018/20 (2) Fronting of long ā. (3) Monophthongization of *ai to long ā. (4) Fronting of short *a, including development of *au to æa. (a) Short a (only) was also fronted in unaccented syllables. This left the way clear for unaccented o to unround to a (cf. §1.2 [8]). (5) Breaking (6) Palatalization. After fronting of both long and short a to æ, because palatalization takes place before the reflexes. Before i-mutation (or at least the unrounding of i-mutation products). (7) Palatal Diphthongization. (Cf. §8.2, above.) (8) i-mutation 8.5.2. Revised relative chronology for Frisian (1) Long and short a develop a back timbre and nasalization before a following nasal. Before change (2), otherwise we would expect them to be fronted by it. (a) Some time thereafter, the sequence short vowel plus nasal plus homorganic spirant yields long nasalized vowel plus homorganic spirant. (b) The development of the uniform plural probably follows (a), and is to be put early for reasons of linguistic geography (2) Fronting of long ā. (3) Monophthongization of *ai and *au (4) Fronting of short *a (a) Short a (only) was also fronted in unaccented syllables. This left the way clear for unaccented o to unround to a (cf. §1.2 [8]). (5) Palatalization. After fronting of both long and short a to æ, because palatalization takes place before the reflexes. Before i-mutation (or at least the unrounding of i-mutation products). (6) i-mutation (7) Breaking I am aware that the chronology for Frisian is still somewhat provisional. It embodies the ‘standard’ view of the relative ordering of monophthongization before i-mutation (cf. Campbell 1939; Nielsen 1993). If Dietrich Hofmann is correct in reversing the order of these two changes (1964[; 1995: 26–27, 31]), then the relative chronologies diverge even more. Remarks /1 August 2018/21 8.6. Frisian Reflexes of the a-Diphthongs There is an additional complication in the treatment of the Germanic a-diphthongs. It is not directly related to questions of relative chronology, but it does cast further doubt on an ‘Anglo-Frisian unity’. Frisian actually shows a twofold outcome of the diphthong *ai. Alongside the reflex spelled ‹e› in Old Frisian manuscripts given in §8.4 above, there is a reflex spelled ‹a› in Old Frisian manuscripts, which merged with the reflex of the PGmc. *au diphthong. [W]Gmc. *raipa- ‘rope’ OE OFris. rāp rāp Århammar has recently drawn attention to the fact that the distribution of the dual outcome of *ai is Proto-Frisian (1990: 22). All varieties of Frisian have, for example, reflexes of PFris. *klāth ‘cloth’, *rāp ‘rope’, *hās ‘hoarse’; * ēk ‘oak’, *bēn ‘bone; leg’, *dēl ‘part’. This double reflex has no counterpart in English, but it does in Low German (cf. e.g. Wortmann 1960: 15–22). Thus the modern dialect of Gütersloh opposes raip ‘rope’ (developed from Low German long ē2a, the more open reflex) and bein ‘leg’ (developed from Low German long ē2b, the more close reflex). This phenomenon can hardly be explained as substratum influence from Frisian, because in many dialects the examples tail off from south to north (cf. Wortmann 1960: 19– 20). However, the distribution is not the same for all Low German dialects. Still, the fact remains that this feature of Frisian points toward Low German and not English. [Cf. further Hofmann 1995.] [8.7. Labial Breaking This Frisian change has no counterpart in Old English (although it has a close parallel in East Norse w-breaking), cf. OFris. siunga “to sing” < *singwan (OSw siunga). It was probably quite a late change in Frisian and may have been coterminous with (ordinary) breaking. ] Remarks /1 August 2018/22 9. The Ingvæonic Model and the Position of Old Saxon 9.1. Above, it has been established that an English and Frisian shared relative chronology of sound-changes cannot be constructed with certainty beyond change (2) in §8.5. Changes (1), (1a) and (2) would hardly suffice to establish an exclusive Anglo-Frisian sub-proto-language—even if these changes did not also form part of the relative chronology of Old Saxon. It would appear, then, that the linguistic evidence supports the conclusions drawn earlier on the basis of the runic material. Indeed, one could make out a case for a closer genetic relationship between Frisian and (some varieties of) Old Saxon than between Frisian and English, in so far as parts of Old Saxon can probably share the relative chronology in §8.5.2 up to at least change (4). The key point here is the treatment of a-diphthongs. Further, as both languages monophthongize both diphthongs, it is probably not necessary to decouple the fronting of short *a from the fronting of long *ā as in the English relative chronology, §8.5.1. This would enable changes (2) and (4) in the relative chronology in §8.5.2 to be collapsed. (Compare the discussion of Vleeskruyer 1948: 183, but, despite Vleeskruyer, it does not allow us to get Old English on board, because of the divergent treatment of PGmc. *au in Old English, cf. §8.4, above.) Whatever may be the case for Old Frisian and Old Saxon, an Ingvæonic common stage cannot be constructed with certainty beyond change (2) in the relative chronologies given in §8.5. In genetic terms, therefore, the Ingvæonic model would appear to be vindicated—so far as it goes. To revert to the question of degree and kind (§2), it remains to be seen whether the Ingvæonic model can account for the degree of the relationship. 9.2. Old Saxon is central to the whole question of Ingvæonic. However, a proper evaluation of the role of Old Saxon is hampered by our ignorance. Outside the biblical poems, the Heliand and fragments of Genesis, the attestation is sparse, so, for example, little is known of Old Saxon agricultural or legal vocabulary. Another difficulty is that in the sources we have, it is possible that the orthography masks certain genuine Old Saxon linguistic features (as argued with great energy and plausibility by Erik Rooth, cf. his most recent exposition 1981: 31–42). Remarks /1 August 2018/23 A further source of ignorance is the elimination of ‘genuine’ Old Saxon traits by progressive High Germanization of the language, a process beginning prehistorically and continuing to the present day, affecting phonology, morphology and lexis. Thus, to select a few examples: on the phonological level, uns replaces us, and ander replaces othar/athar already beginning in Old Saxon; on the morphological level, sik replaces the reflexive use of the personal pronouns in Middle Low German (cf. e.g. Foerste 1962: 17); on the lexical level, such Old Saxon items as kind and urkundeo betray themselves as imports from the south because they do not obey Old Saxon sound-laws (cf. W. Simon 1965: 25–46). As Jørgensen observes, 1957: 14: ‘Verfolgt man diesen Prozess weiter rückwärts, kommt man zwangsläufig zu einem Punkt, wo das betreffende Gebiet nicht nur sporadisch ingwäonisch, sondern rein ingwäonisch war’. However, in my view it may be misguided to take this to its logical conclusion. Old Saxon was never a monolith, rather, it was a dialect continuum. (I owe much of my view of Old Saxon to David McLintock, my doctoral supervisor.) 9.3. Rooth’s orthographic theory contains some truth, but is certainly exaggerated. It is true, for example, that fronted values æ or e of WGmc. long and short a could easily lurk behind the ‘Frankish’ ‹a›-spelling. But morphological and lexical variation can hardly be deemed ‘merely orthographic’. For example, in the a-stem nom.-acc. pl., the ending -os (alongside -as) is lacking in High German, whereas the Old Saxon ending -a, which corresponds to that of Old High German, is found only outside the Heliand tradition. However, the variant -o, which is found occasionally in the Munich (M) and Cotton (C) manuscripts of the Heliand (cf. Gallée 1910: §297 A6), could be a hypercorrection born of the attempt to make the language look ‘Frankish’—unless it is a scribal error for -os. Further, the replacement of the butan ‘except; unless’ found in M by neuan, nebon in C and the Straubing fragment (S) is clearly not a matter of orthography (cf. Sievers 1876: 72, Simon 1965: 54–55, Korhammer 1980: 91). Paradoxically, it is the Straubing fragment—of all Heliand texts the most free of the suspicion of having its sound-system masked by Frankish orthography—that most clearly reveals the shortcomings of Rooth’s hypothesis. When its phonology shows such little sign of slavishly imitating Frankish models, it seems excessive to suppose that its morphology has done so. Yet, even this most Ingvæonic of Heliand texts has at least one feature in common with Frankish (i.e. non-Ingvæonic). It has an Ingvæonic unaccented vowel system (cf. §1.2 point [8]), with spellings ‹-e› and ‹-a› corresponding to OHG -e; -a respectively -o, cf. Klein 1977: 479–498, 1990: 209. Yet, there is a morphological exception. In the fem. ō-stem nom.-acc. sg., instead of expected ‹-e›, we have ‹-a›, e.g. sorga ‘care’, just as in High German (which does NOT have -o in this category: sorga), cf. Taeger 1984: 367-68, Klein 1990: 209. It seems unlikely that the scribe of S should have faithfully retained the spelling of his exemplar in just this grammatical category. Again, S shows some preference for the non-Ingvæonic ‘long’ pronominal masculine and neuter dative Remarks /1 August 2018/24 singular forms (type themu, imu ‘to the/that’, ‘to him’), although the archetype seems to have lacked these (type them, im), cf. Klein 1990: 214–15. It is implausible to assume that the mixture of S does not reflect some linguistic reality, cf. Klein 1990. If the language of the Straubing fragment is ‘genuine’, then it seems unreasonable to deny this possibility to other varieties of Old Saxon, especially as, for instance, some of the same combinations of features displayed by S are found in manuscript M, cf. Klein 1990: passim. Some parts of the Old Saxon area were clearly more Ingvæonic than others.9 And the various Ingvæonic traits were not co-terminous in their distribution. 9.4. Old Saxon is something of a ragbag, sharing some features with neighbouring dialects (e.g. Heliand S has consistent ht < ft as in Dutch, cf. Taeger 1982: 12–13). It is the combination of features that constitutes Old Saxon, the language having made no exclusive innovations of its own. Indeed, Old Saxon is easiest defined negatively as that continental West Germanic that is not High German, nor Frisian, nor Dutch. In effect, this is demonstrated by Klein in his meticulous article of 1990. Old Saxon shows the classic properties of a transitional dialect. Thus, where Old English and Old Frisian have a form that differs from the one found in Old High German, Old Saxon often has BOTH the Ingvæonic and the non-Ingvæonic variants. I give a few characteristic examples. • Mention has already been made of the ‘long’ and ‘short’ dative singulars in the demonstrative pronoun and strong adjective (§9.3). Compare OE þām OFris. thām OS them; themu OHG demo. • In the inflection of class II weak verbs: 3 pls. OE laþiaþ OFris. lathiath OS Hel 2816 ladoiat*; *ladod OHG ladont ‘they invite’ (cf. Cowgill 1959: 1–3). • This is even true of the stem-vocalism of some lexemes: OE libban OFris. libba OS libbian; lebon* OHG leben ‘to live’, OE wicu OFris. wike OS wika; weke, OHG wehhe ‘week’, OE liccian OS likkon; leccon OHG lechhon ‘to lick’. Old Saxon largely agrees with Old High German in the implementation and chronology of i-mutation: only the ‘primary umlaut’ of *a is regularly represented and its operation is influenced by various umlaut-hindering consonants, and, more importantly, the umlaut-allophones were developed subsequent to the loss of *-i after heavy syllables, umlaut thus only being caused by ‘retained -i’ (cf. Simon 1965: 7–8). However, there are also forms in Old Saxon which reflect Ingvæonic imutation caused by ‘lost *-i’, for example, the plural form men(n) ‘men’ of Genesis III, the Lamspringe glosses and Heliand S, as well as the comparative adverb forms discussed by Simon (1965: 13–25; cf. also Klein 1990: 206–209). As mentioned Rooth’s orthographic camouflage theory is weakened by the fact that the Heliand archetype appears to stem from an area and a milieu that was genuinely not very Ingvæonic, cf. Klein 1977: 333–34; 1990: 201. 9 Remarks /1 August 2018/25 above (§8.4.2), parts of the Old Saxon area appear to share with Old Frisian the monophthongization of *au > ā alongside ‘usual’ ō (cf. Gallée 1910: §§95–96). Old Saxon shares with Old High German alone the sound-change of e to i before u, as in fihu ‘money’, filu ‘much’, but there are numerous exceptions, e.g. fehu, ehu- ‘horse’ (cf. Gallée 1910: §65 A.1). Yet all varieties of Old Saxon display the Ingvæonic features of the uniform present plural ending -aþ in verbs and a four-member unaccented vowel inventory, cf. Klein 1977: 529–537. On the other hand, as noted in passing above (§7.1.2), all varieties of Old Saxon agree with Old High German (and Old Low Frankish) against Old English and Old Frisian in having adjectival adverbs in -o ~ -a. To this can be added the dative singular feminine pronominal forms in final -u ~ -o (cf. Klein 1990: 217). 10. English and Frisian in the West Germanic Continuum 10.1. Not only is Old Saxon a dialect continuum, the same is true of continental West Germanic (after the departure of English), and was true of Ingvæonic and, indeed, Old West Germanic. As Jørgensen remarked (1957: 13): ‘Das ingwäonische Problem ist somit ein Problem der altwestgermanischen Dialektologie und Dialektgeographie.’ I give below a crude representation of the West Germanic languages as a dialect continuum, attempting to depict the Ingvæonic and non-Ingvæonic components of Old Saxon and Old Dutch: However, a stemma is not ideally suited to representing a situation in which several closely related languages show a series of identical or similar changes, but do not share identical chronologies for them (cf. Anttila 1989: 304; Hock 1986: §15.3, esp 450; also Hoenigswald 1966: 10–12). As has been seen (cf. §1.2, §8, especially §8.5, and §9.1), the Ingvæonic dialects have features in common that Remarks /1 August 2018/26 cannot be put into a single relative chronology. Where there has been diffusion of linguistic features between dialects (cf. also §3.1), the wave model, with its crisscrossing isoglosses is a more appropriate way of looking at the data. 10.2. Recognition of a dialect continuum puts the problem in a different light. If the ‘Anglo-Frisian’ thesis is wrong, does that necessarily mean the Ingvæonic model is right? For two members of a dialect continuum to enjoy an exclusive relationship (such as ‘Anglo-Frisian’) is a contradiction in terms.10 On the other hand, it will be remembered that one of the tenets of the Ingvæonic model as presented in §2.2.2 is that no two of the three languages are more closely related to each other than they are to the third. This may be true from the viewpoint of a stemma or family tree, but from a dialect-geographical perspective, it would be odd to say the least to find that three adjacent members of a continuum were respectively united and divided by exactly equal numbers of isoglosses. Of the (attested) West Germanic dialects, English—and within English, Anglian—was the most northerly, as is indicated by the features it shares with North Germanic (cf. Samuels 1971: 5–6, Nielsen 1981a: 33–37, and the references they cite). Old Saxon was adjacent to and—in historical times at least—located north of Old High German. This leaves Frisian somewhere between English and Saxon and probably westerly. The approximate relative positions of the ‘pre-languages’ can be represented as follows (cf. Samuels’s map 1971: 6). I furnish it with a selection of isoglosses [There are additional features and I am unable to reproduce the isoglosses here]. 10 It is true, one could conceive of English and Frisian being adjacent, with Frisian in a relatively isolated coastal position to the west, and contact having been lost between West and North Germanic to the north of English (see the isogloss map that follows in the main text). But in these circumstances, the dialect continuum would by definition be broken. Compare Morsbach’s pithy formulation (1897: 323), when criticizing Siebs’ ‘hypothese von einer e n g l i s c h - f r i e s i c h e n g e m e i n s p r a c h e oder m u t t e r s p r a c h e ’: Eine solche gemeinsprache aber hat es aber nie gegeben. Sie würde zur voraussetzung haben, dass die Angelsachsen und Friesen auf dem Kontinent (etwa im 3. und 4. jarh.) eine in sich abgeschlossene und sprachlich in allen einzelheiten übereinstimmende völkergruppe gebildet haben. … Geographisch scharf abgegrenzte dialektgruppen gab es damals so wenig wie heutzutage; sie sind nur bei völliger geographischer isolierung möglich. Allerdings steht das Friesische dem Englischen ohne zweifel sprachlich näher als das Sächsische… Remarks /1 August 2018/27 North Germanic gen sg. *fađurz *suna pl. *swiđæ *ånd *ai-reflex opener than *ē1? Anglian English ‘Saxon English’ 3. pl. *-aþ Frisian Saxon *bium Dutch *bim High German A closer relationship between English and Frisian would reside in their sharing more isoglosses with each other than with other dialects and/or having isoglosses unique to them. Some exclusive morphological agreements were cited in §7.111. The syntactic agreement noted by Dietrich Hofmann (1982) can be added. This material can be enhanced by sub-dialectal morphological agreements that could not properly be handled earlier, when the daughter languages in their entirety were being considered, compare Nielsen 1981b (also foornote 11 here). However, there 11 In the detailed context of a dialect continuum, the distinction between genesis and borrowing (invoked regarding the shared morphological innovation suna in §7.1.1) loses importance. When any linguistic change is viewed in detail, it is found to involve both the origin and propagation of innovation. • The extent of negative verb contraction in both languages is relevant as regards shared isoglosses. However, I am uncertain as to the chronology of this phenomenon, so judge it best to leave it out of account here, cf. Nielsen 1981a: 146–47. • Also worth mentioning is the sub-dialectal Anglian Old English and Frisian agreement of a class 6 type preterite paradigm for the class 4 verb niman: thus nom nomon (Riostring Old Frisian shows sg. nam, which may be comparable to the West Saxon Old English form, or could be the result of Low German influence). The class 4 verb cuman has a general OE and OFris. c(w)ōm c(w)ōmon paradigm. The Old Saxon forms are nam nāmun; quam quāmun, Hel S also cāmun. Nielsen is right to point out (1986: 72) that, since the singular forms with long -ō- arose because the rounding of WGmc. long *ā before a nasal (§1.2 [3]) in English and Frisian affected the plural, the analogy on class 6 (type fōr fōron) could have taken place independently in the two languages. However, because of the dialectal oddity of the partial participation of niman in the change and also on methodological grounds, I incline to the view that this as an analogy that took place once. • In this context, we can mention such agreements between sub-dialects as OE (Kt.) lærest : OFris. lerest; OE (WS) and OS pret. sg. funde. Remarks /1 August 2018/28 are also other relationships, as one would expect within a dialect continuum, compare the instances of Old Frisian–Old Saxon and Old English–Old Saxon agreement cited in §§7.2.2 and 7.2.3. It must be borne in mind that the entities we are dealing with in the continuum do not equate straightforwardly with the later languages, or even their dialects. The isoglosses have primacy. As one pursues the entities backwards in time, they dissolve into a haze of indeterminacy. The later West Germanic daughter languages have emerged from this steaming broth—Dutch, like Low German, is an amalgam of Ingvæonic and non-Ingvæonic features (it is a mistake to equate absence of the second consonant shift with Ingvæonic, cf. Campbell’s simplistic statement ‘West Germanic without Old High German is often called ‘Ingvaeonic’, 1959: §4). As Samuels remarks (1971: 6–7), pointing out that one should not reckon with ‘the transplanting of tribal units intact’: amidst all the mixtures suggested by the archaeological evidence, there must usually have been preponderances that determined the preference for one linguistic variant to another. Some anomalies within the later languages can be interpreted as the result of such twists of the kaleidoscope. For example, Old English shows traces of the Old High German and Old Saxon raising of e to i before u in two words, cwidu ‘cud’, and the fiður- combining-form of ‘4’, by-form of feoður- and variants (cf. Ross 1974: 123). These forms cannot readily be accounted for merely in terms of the spread of soundchanges leading to ‘regular’ isophones, but presuppose the borrowing of ready-made forms (cf. Simon 1965: 24, 26). A cline of shared features is in principle amenable to quantitative evaluation. This means that instead of the bald plus or minus required when identifying innovations, concepts such as more like and less like can be employed, and the features examined do not have to be fittable into a relative chronology (cf. footnote 11). However, the pre-requisite is for the continuum to be observable. The analysis is effectively a synchronic exercise; one has to be there to observe and identify. Matters are not so simple when we are dealing with reconstructed stages and the dialects in question are no longer all in situ. Something else to bear in mind is that the ‘shared features’ we can observe between the later languages in all probability date from different times. 10.3. These issues are of particular relevance to word geography, which has been adduced in determining the relationship of English and Frisian (compare the references cited in §1.1). In addition to the general provisos about the use of lexical material in establishing linguistic relationship mentioned in §3.2, the peculiar circumstances of Continental West Germanic must be borne in mind. Once again, our relative ignorance of early Old Saxon is a stumbling-block. Whereas later Frisian attestation can supplement the limitations of Old Frisian (cf. Århammar 1989), this does not hold to the same extent for Low German, because of the retreat Remarks /1 August 2018/29 in the face of expansion of High German features (cf. §9.2)12. This bedevils any inquiry. Ideally, one needs copious material to be able to argue from vocabulary. This would be of especial value in the present context, given the general pattern of the isoglosses, which often include part of the Old Saxon area (cf. §9.4, and given the fact that there are often differing distributions for each word. It is true that a large number of vocabulary agreements between English and Frisian can be observed. With Bremmer (1982: 83), one can reasonably doubt that the English and Frisians should have borrowed from each other everyday words denoting body-parts, basic agricultural terms, or the name of as common a bird as the lapwing, and, as he remarks, this is a powerful argument against the ‘convergence hypothesis’. However, one cannot be sure that all the words were exclusive to English and Frisian. In most cases, Old Saxon is silent. We have to ask ourselves to what extent we are saying something about a special relationship between English and Frisian, and to what extent we are saying something about the attestation of Old Saxon (cf. §9.2 above, and the remarks of Århammar 1986: 179). Moreover, as a general point, it is a mistake merely to seek data that appear to confirm the hypothesis of an especially close relationship between English and Frisian. One needs to be on the look-out for counter-evidence as well. However, it must be recognized that in the field of lexis, this is harder to find. English has lost common Germanic vocabulary items in its island home (e.g. *haberan- ‘oats’, *hrain- ‘clean’, *mēhan- ‘poppy’, *speht- ‘woodpecker’), so this will of itself produce agreements between Frisian and Saxon. But, as both the latter have been subject to High German expansion (and if borrowed items are loan-translations utilizing etymologically identical material or are accompanied by soundsubstitution, they may be difficult to identify as such), this factor will also lead to agreements between Frisian and Saxon. As Frisian is generally less affected by this process (and the position of Island North Frisian is important here), English and Frisian share relic-words, which may have been lost in (the parts of) Old Saxon that once had them. On the other hand, as English and Saxon are attested much earlier than Frisian, they may share archaic words that are absent from Frisian for no other 12 In so labile a field as vocabulary, the retreat before High German has been particularly intense, and is in evidence from the time of the earliest continental West Germanic texts. Indeed, High German itself experiences a northward movement of Upper German vocabulary, cf. Braune 1918: 364–67. The process in Old Saxon has been alluded to above (§9.2). Even Frisian, which resisted southern penetration on the phonological level with a high degree of success, has been subject to lexical penetration, which must dim the prospects for reconstructing West Germanic word geography. Thus, the otherwise archaic Riostring dialect of Old East Frisian has admitted kind and, with sound-substitution, reth ‘wheel’ and sletel ‘key’ (cf. Århammar 1969: 66–67). Lexical penetration is not confined to the Old period: the Ingvæonic type ‘Wednesday’ has been extensively replaced by the High German type ‘Mittwoch’ in the modern Frisian dialects, cf. Löfstedt 1966: 62-63, Århammar 1986: 103 note 35. Incidentally, the fact that Old English attests only an umlautless form of Wednesday, OE wodnesdæg, whereas Old Frisian attests forms both with and without umlaut, OFris. wernesdei and wonesdei, but both languages arrived at forms with umlaut, is a potential example of convergence (cf. §§2.1, 4). Remarks /1 August 2018/30 reason than that its attestation is too late to record them (a probable example is the Old English Mercian archaism tulge and OS tulgo ‘very’). It is instructive to consider these points with reference to a few English and Frisian relic-words (cf. the list in Århammar 1968: 73 fn 58, cf. 1990: 24–25; it should be pointed out that Århammar uses the words there for other purposes than the one I am using them for). I quote the items in (Old) English form. • key: This is rightly considered a classic Frisian–English agreement. However, Simon provides strong arguments for thinking this word once belonged at least to Old Saxon as well (1965: 39–43). It may be noted that it has been replaced by the loan sletel in the Old Frisian Riostring manuscripts (although the peripherally located modern Weser Frisian dialects of Wangeroog and Wursten record it). • OE ofergietan ‘to forget’: This is the rarer and more archaic of the two verbs for ‘to forget’ in Old English, the other being forgietan, the later general West Germanic word. The only attested sure cognate of ofergietan is Syltring auriit ‘to forget’ (on the prefix, cf. Siebs 1901: 1268). It is not just a theoretical possibility that this word might have been lost by Old Saxon. Since McLintock provides good reasons for regarding it as the old Ingvæonic term (1972: 79–86), we may conclude Old Saxon did lose it. • grind: This verb—recognized as an English–Frisian agreement already by Kosegarten (1846: 106)—does appear to be lacking from Old Saxon. However, derivatives in other Germanic languages indicate it was Proto-Germanic (cf. Seebold 1970: 240). • sleeve (OE sliefe, OFris. *sleve): As far as we can tell, this word is at the very least semantically unique to English and Frisian as a specialization of usage (cf. Onions et al. 1966: s. v.). • til ‘to’ preposition with the dative: This word is not mentioned by Århammar in the places cited (but is duly listed by Löfstedt 1967: 58; cf. also Miedema 1972–73). It is attested in Northumbrian Old English (already in the Moore MS of Cædmon’s hymn, c. 737, and thus earlier than any Viking influence) and Old East Frisian. As far as we can tell, it is lacking in Old Saxon. The two languages share this isogloss with Norse (where, however, til takes the genitive). But one has to ask: In such cases as *oƀar-getan and *grindan, does it make sense to describe as ‘Old Saxon’ or even ‘pre-Old Saxon’ the language-stage that still retained the word in question? There is the risk that allowing for the existence of given words in Old Saxon can be taken ad absurdum. One has to interpret the data that exist. Besides, it is probable that some of the agreements that have been cited in the literature WERE exclusive to English and Frisian, despite the silence of Low German. Such a situation is to be expected in the context of a dialect continuum. Remarks /1 August 2018/31 10.4. To draw up the balance sheet at the end of this investigation. There would appear to be no difference in kind in the mutual relationships of English, Frisian and Saxon (and Dutch); they were all members of the Ingvæonic division-cum-dialect continuum of West Germanic.[And this also holds true for the Old English dialects Anglian and West Saxon, which at an early stage were as distinct as the dialects that became the later languages.] Concerning degree of relationship, Frisian seems poised between English and Low German, which is what one would expect in terms of the Ingvæonic dialect continuum (cf. the figure in §10.2). As regards exclusive English–Frisian isoglosses: in the realm of phonology, no early exclusive innovations were found; in the field of morphology there are a handful; on the syntactic plane, one; in lexis ‘probably’ several. Århammar is correct to observe: ‘Wir können also wohl resultativ von einem Anglofriesischen sprechen und zwar vonab der Zeit der Entingwäonisierung des Festlandssächsischen und des Festland-niederfränkischen’ (1982: 90, cf. 1990: 10). And for this reason it will remain the case that the most convenient place to find a parallel for Old Frisian will be in Old English, and vice versa. 11. Summary of Results In this article I have attempted to show that the evidence does not support the notion of an ‘original Anglo-Frisian unity’ or sub-proto-language. This is because it is not possible to construct the exclusive common relative chronology that is necessary in order to be able to establish a node on a family tree. The term and concept of ‘Anglo-Frisian’ should be banished to the historiography of the subject. Rather, English and Frisian descend largely from adjacent dialect groupings in the Ingvæonic continuum and thus share a number of exclusive isoglosses (as do in their turn English and Old Saxon, and Frisian and Old Saxon). It does appear to be the case that the isoglosses that are recoverable from that period justify our regarding English and Frisian as ‘more closely’ related to each other than to Low German (in terms of sharing more isoglosses). Because of the circumstances attending the attestation of Old Saxon and Old Dutch—the relatively sparse documentation, possible orthographic complications, and the progressive retreat, already from prehistoric times, of ‘Ingvæonic’ features in the favour of ‘inland’ ones—the state of affairs from the Old periods of the three languages onward is that English and Frisian show a high degree of resemblance to each other because of their status as Ingvæonic relict areas. Department of Scandinavian Studies University College London GB-London WC1E 6BT Remarks /1 August 2018/32 References Anttila, Raimo (1989) Historical and Comparative Linguistics. Second revised edition. Amsterdam–Philadelphia. Århammar, Nils (1960) ‘Zur inselnordfriesischen Wortkunde’, in Fryske Stúdzjes oanbean oan Prof. Dr. J.H. Brouwer op syn sechstichste jierdei 23 augustus 1960. Eds. K. Dykstra et al. Assen. 279–286. Århammar, Nils (1968) ‘Die Herkunft des Inselnordfriesischen im Lichte der Wortgeographie’. Philologia Frisica anno 1966. Ljouwert/Leeuwarden. 49– 76. Århammar, Nils (1969) ‘Die friesischen Wörter für “Rad” (“Wheel”)’. Kopenhagener germanistische Studien 1: 35–84. Århammar, Nils (1981) ‘Ernst Löfstedt, Leben und Werk, mit einer Bibliographie’. Us Wurk 30: 79–100. Århammar, Nils (1982) ‘Neipetear’ ad Bremmer 1982. Philologia Frisica anno 1981. Ljouwert/Leeuwarden. 89–90. Århammar, Nils (1984) ‘Die Lerche (Alauda) im Friesischen: ein Beitrag zum Atlas Linguarum Europae. Über Inlautassibilierung, unregelmäßige Lautentwicklung und “Lehnlautungen”’, in: Miscellanea Frisica, edd. N.R. Århammar et al. Assen. 137–152. Århammar, Nils (1986a) ‘Neipetear’ ad Nielsen 1986. Philologia Frisica anno 1984. Ljouwert/Leeuwarden. 179–80. Århammar, Nils (1986b) ‘Der “Umlaut” der wfries. Dialektform mendei “Montag”. Us Wurk 35: 85–103. Århammar, Nils (1989) ‘Zur Rekonstruktion des altfriesischen Lexikons mit Hilfe der neufriesischen Dialekte’. Philologia Frisica anno 1988. Ljouwert/ Leeuwarden. 94–128. Århammar, Nils (1990) ‘Friesisch und Sächsisch. Zur Problematik ihrer gegenseitige Abgrenzung im Früh- und Hochmittelalter’, in: Aspects of Old Frisian Philology, edd. Rolf H. Bremmer et al. Amsterdam–Atlanta. 1–25. Bammesberger, Alfred (1991) ed. Old English Runes and their Continental Background. (Anglistische Forschungen 217). Heidelberg: Carl Winter. Beck, Heinrich (1981) ‘A Runological and Iconographical Interpretation of NorthSea-Germanic Rune-Solidi’, in: Proceedings of the First International Symposium on Runes and Runic Inscriptions. Ed. Claiborne W. Thompson. Michigan Germanic Studies 7: 69–88. Remarks /1 August 2018/33 Braune, Wilhelm (1918) ‘Althochdeutsch und Angelsächsisch’. Beiträge zur Geschichte der deutschen Sprache und Literatur 43: 361–445. Bremmer, Rolf H (1982) ‘Old English – Old Frisian: The Relationship Reviewed’. Philologia Frisica anno 1981. Ljouwert/Leeuwarden. 79–90. Campbell, Alastair (1939) ‘Some Old Frisian Sound-Changes’. Transactions of the Philological Society. 78–107. Campbell, Alastair (1959) Old English Grammar. Oxford. Campbell, Alastair (1960) ‘Once again Anglo-Frisian’, in: Fryske Stúdzjes oanbean oan Prof. Dr. J.H. Brouwer op syn sechstichste jierdei 23 augustus 1960. Eds. K. Dykstra et al. Assen. 3–4. Collinson, W.E (1922) ‘Notes on North Frisian (Sylt) Etymology’. Modern Language Review 17: 263–271, 392–402. Collinson, W.E (1950) ‘Some etymologies’. Modern Language Review 45: 355– 57. Collinson, W.E (1960) ‘Some English and Frisian Equations’, in: Fryske Stúdzjes oabean oan Prof. Dr. J.H. Brouwer op syn sechstich jierdei 23 augustus 1960, edd. K. Dykstra et al. Assen. 5–6. Collinson, W.E (1969) ‘Some Further English–Frisian Parallels’, in: Studia Frisica in memoriam Prof. Dr K. Fokkema 1898–1967 scripta, edd. H.D. Meijering et al. Groningen. 35–40. Cowgill, Warren (1959) ‘The inflection of the Germanic ō-presents’. Language 35 (1–15. Cowgill, Warren 1970) ‘Italic and Celtic Superlatives and the Dialects of IndoEuropean’, in: Indo-European and Indo-Europeans, edd. George Cardona et al. Philadelphia (113–153. Cowgill, Warren (1973) ‘The source of latin stāre, with notes on comparable forms elsewhere in Indo-European. The Journal of Indo-European Studies 1: 271– 303. Düwel, Klaus and W.-D. Tempel (1970) ‘Knochenkämme mit Runeninschriften aus Friesland’. Paleohistoria 14. 353–391. Foerste, William (1962) ‘Die Herausbilding des Niederdeutschen’, in: Festschrift für Ludwig Wolff zum 70. Geburtstag, ed. Werner Schröder. Neumünster. 9– 27. Gallée, Johan Hendrik (1897) ‘Zur altsächsichen Grammatik (1. As. participia auf -in. 2. hwô—hû ’. ZfdPh 29: 145–49. Remarks /1 August 2018/34 Gallée, Johan Hendrik (1910. Altsächsische Grammatik. Eingeleitet und mit Registern versehen von Johannes Lochner. Zweite völlig umgearbeitete Auflage. Halle–Leiden. Haeringen, C.B. van (1920) ‘Zur friesischen Lautgeschichte’. Beiträge zur Geschichte der deutschen Sprache und Literatur 44: 27–48. Hallander, Lars-G (1966) Old English Verbs in -sian. A Semantic and Derivational Study. (Stockholm Studies in English 15). Stockholm. Hock, Hans Henrich (1986. Principles of Historical Linguistics. Berlin–New York– Amsterdam. Hoenigswald, Henry M (1966) ‘Criteria for the subgrouping of languages’, in: Ancient Indo-European Dialects, edd. H. Birnbaum and J. Puhvel, Berkeley– Los Angeles. 1–12. Hoenigswald, Henry M (1973) Studies in Formal Historical Linguistics. Dordrecht–Boston. Hofmann, Dietrich (1964) ‘“Germanisch” ē2 im Frieisichen’, in: Festschrift für Jost Trier zum 70. Geburtstag, edd. William Foerste and Karl Heinz Bork. Köln– Graz (160–185. (Reprinted in Hofmann 1989.) Hofmann, Dietrich (1972–73) ‘Fries. tiuche, deutsch zeche, griech. dikh und Verwandte’, in Bydragen wijd oan de neitins fan Mr. M.G. Oosterhout (1920–1970). Us Wurk 21–22: 55–80. (Reprinted in Hofmann 1989.) Hofmann, Dietrich (1982) ‘Zur Syntax der Zehnerzahlen mit Substantiv in den altgermanischen Sprachen, insbesondere im Altfriesischen', in Earefrissel foar Prof. F.G.A. Galama ta syn santichste jierdei. Us Wurk 31: 85–106. (Reprinted in Hofmann 1989.) Hofmann, Dietrich (1989) Gesammelte Schriften. Band II. Studien zur Friesischen und Niederdeutschen Philologie. Eds. Gert Kreutzer et al. Hamburg. [Hofmann, Dietrich (1995) ‘Zur Monophthongierung von germanisch ai und au im Altfriesischen und in seinen Nachbarsprachen’, in: Lingua Theodisca: Beiträge zur Sprach- und Literaturwissenschaft; Jan Goosens zum 65. Geburtstag, edd. José Cajo, Ludger Kremer and Hermann Niebaum. (Niederlande-Studien 16/1,2). Münster–Hamburg: Lit. 23–35. ] Hogg, R.M (1979) ‘Old English Palatalization’. Transactions of the Philological Society. 89–113. Hogg, R.M (1992) A Grammar of Old English. Volume I Phonology. Oxford. Insley, John (1991) ‘Appendix 2: The personal name *Aniwulf/ *Æniwulf’, of Mark Blackburn ‘A Survey of Anglo-Saxon and Frisian Coins with Runic Inscriptions’, in Bammesberger 1991. 137–189. Remarks /1 August 2018/35 Jørgensen, Peter (1957) ‘Das Problem der Ingwäonen’. Philologia Frisica anno 1956. Ljouwert/Leeuwarden. 7–14. Klein, Thomas (1977) Studien zur Wechselbeziehung zwischen altsächischem und althochdeutschem Schreibwesen und ihrer sprach- und kulturgeschichtlichen Bedeutung. Göppingen. Klein, Thomas (1990) ‘Die Straubinger Heliand-Fragmente: Altfriesisch oder Altsächsisch?’, in: Aspects of Old Frisian Philology, edd. Rolf H. Bremmer Jr. et al. Amsterdam–Atlanta. 197–225. [Klein, Thomas (1996) ‘Zu altwestfälisch ande “und” ’, in: A Frisian and Germanic Miscellany, published in honour of Nils Århammar on his sixty-fifth birthday, 7 August 1996, edd. Adeline Petersen and Hans F. Nielsen. Nowele vols. 28/29, in association with the Nordfriisk Instituut. Odense– Bredstedt. 399–411.] Korhammer, Michael (1980) ‘Altenglische Dialekte und der Heliand’. Anglia 98: 85–94. Kosegarten, J.G.L (1846) ‘Die Wangeroger Sprache’. [Techmers] Zeitschrift für die Wissenschaft der Sprache 1: 93–111. Kortlandt, Frederik (1994) ‘On Breaking’. Nowele 24: 15–19. Krause, Wolfgang (1968) Handbuch des Gotischen. 3. Auflage. München. Krogmann, Willy (1960) ‘Afries. tisur(i)ke und tser(e)ke “Kirche”’. Us Wurk 9: 29–32. Kuhn, Hans (1955a) ‘Zur Gliederung der germanischen Sprachen’. Zeitschrift für deutsches Altertum 86: 1–47. Kuhn, Hans (1955b) ‘Friesisch und Nordseegermanisch’. Us Wurk 4: 37–46. Kuhn, Hans (1957) ‘Das Problem der Ingwäonen’. Philologia Frisica anno 1956. Ljouwert/Leeuwarden (15–27. Lerchner, Gotthard (1965) Studien zum nordwestgermanischen Wortschatz. Halle/S. [Leumann, Manu (1959) ‘Baltisch und Slawisch’, in: Kleine Schriften. Zürich. 389–398. (Originally appeared in: Corolla Linguistica: Festschrift Ferdinand Sommer, ed. Hans Krahe. Wiesbaden: Otto Harrassowitz. 154–62.). ] Löfstedt, Ernst (1931) Nordfriesische Dialektstudien. (= Die nordfriesische Mundart des Dorfes Ockholm und der Halligen II). Lund. Löfstedt, Ernst (1932) Zwei Beiträge zur friesischen Sprachgeschichte. Lunds Universtitets Årsskrift n.f. avd. 1. Bd. 28/2. Lund. Löfstedt, Ernst (1960a) ‘Afries. sziurke, tzerke usw. “Kirche”’. Us Wurk 9: 1–3. Remarks /1 August 2018/36 Löfstedt, Ernst (1960b) ‘“Kirche” im Altostfriesischen’. Us Wurk 9: 49–51. Löfstedt, Ernst (1963–69) ‘Beiträge zur nordseegermanischen und nordseegermanisch-nordischen Lexicographie'. Niederdeutsche Mitteilungen 19/21 [1963/65] 281–345, 22 [1966] 39–65, 23 [1967] 11–61, 25 [1969] 25– 39. Luick, Karl (1914–40) Historische Grammatik der englischen Sprache. Leipzig. McLintock, D.R (1972) ‘“To forget” in Germanic’. Transactions of the Philological Society. 79–93. Morsbach, Lorenz (1896) Review of O.F. Emerson, The History of the English Language (New York 1894). Anglia Beiblatt 7: 321–338. Munske, Horst Haider (1970) ‘Angelsächsisch-altfriesische Beziehungen in der Rechtsterminologie für Missetaten’, in: Flecht op 'e koai, stúdzjes oanbean oan Prof. W. J. Buma ta syn sechstichste jierdei, edd. Teake Hoekema et al. Grins/Groningen. 40–52. Munske, Horst Haider (1973) Der germanischen Rechtswortschatz im Bereich der Missetaten. I. Die Terminologie der älteren westgermanischen Rechtsquellen. (Studia Linguistica Germanica.) Berlin. NED = New English Dictionary (1888–1928. Eds. J.A.H. Murray et al. Oxford. Reissued in 12 volumes 1933. [Nedoma, Robert (2014) ‘Voraltfriesisch -u im Nominativ und Akkusativ Singular der maskulinen a-Stämme’, in: Directions for Old Frisian Philology, edd. Rolf H. Bremmer, Stephen Laker, Oebele Vries. (= AmbäG 73/ Estrikken 96). Amsterdam–New York: Rodopi. 343–68. ] Nielsen, Hans F (1981a) Old English and the Continental Germanic Languages. A Survey of Morphological and Phonological Interrelations. (Innsbrucker Beiträge zur Sprachwissenschaft 33). Innsbruck. Second edition 1985. Nielsen, Hans F (1981b) ‘Old Frisian and the Old English dialects'. Us Wurk 30: 49–66. Nielsen, Hans F (1984) ‘Unaccented vowels in the Frisian runic inscriptions’, in: Miscellanea Frisica, edd. N.R. Århammar et al. Assen. 11–19. Nielsen, Hans F (1986) ‘Old English, Old Frisian and Germanic’. Philologia Frisica anno 1984. Ljouwert/Leeuwarden. 168–80. Nielsen, Hans F (1991a) ‘Unaccented Vowels in Runic Frisian and Ingvæonic’, in Bammesberger 1991. 299–303. Nielsen, Hans F (1991b) ‘The Straubing Heliand-Fragment and the Old English Dialects’, in Language Contact in the British Isles. Proceedings of the Eighth International Symposium on Language Contact in Europe, Douglas, Isle of Remarks /1 August 2018/37 Man, 1988. Eds. P. Sture Ureland and George Broderick. Tübingen. 243– 273. Nielsen, Hans F (1993) ‘Runic Frisian skanomodu and aniwulufu and the relative chronology of monophthongization and i-mutation’. Nowele 21/22: 81–88. Onions, C.T (1966) The Oxford Dictionary of English Etymology. Edited by C.T. Onions, with the assistance of G.W.S. Friedrichsen and R.W. Burchfield. Oxford. Page, R. I (1973) An Introduction to English Runes. London. [Quak, Arend (1990) ‘Runica Frisica’, in: Aspects of Old Frisian Philology, edd. Rolf H. Bremmer Jr., Geart van der Meer and Oebele Vries. (= AmBäG 31/32 and Estrikken 69). Amsterdam–Atlanta: Rodopi; Groningen: SSFYRUG. 357-370. ] Rooth, Erik (1981) ‘Das sog. westgermanische ā im Altsächsischen’. Part III of Nordseegermanische Studien. II. (Filologiskt Arkiv 24). Stockholm. 30–42. Ross, A.S.C (1974) ‘Anglo-Saxon “e” to “i” before “u”’. N&Q 219: 123. Samuels, M.L (1971) ‘Kent and the Low Countries: some linguistic evidence’, in: Edinburgh Studies in English and Scots, edd. A.J. Aitken et al. 3–19. Schönfeld, M. and A. van Loey (1970. Schönfeld’s historische Grammatica van het Nederlands. Achtste druk bewerkt door A. van Loey. Zutphen. [Schuhmann, Roland (2014) ‘Zur Endung runisch-altfriesisch -u und zur Entwicklung der Endung nom.sg.m. a-St. urgermanisches *az’, in: Directions for Old Frisian Philology, edd. Rolf H. Bremmer, Stephen Laker, Oebele Vries. (= AmbäG 73/ Estrikken 96). Amsterdam–New York: Rodopi. 397–419. ] [Seebold, Elmar 1967. Die ae. starken Partizipia Praeteriti mit Umlaut. Anglia 85: 251-269. ] Seebold, Elmar (1970) Vergleichendes und etymologisches Wörterbuch der germanischen starken Verben. The Hague–Paris. Seebold, Elmar (1990) ‘Die Inschrift B von Westeremden und die friesischen Runen’, in: Aspects of Old Frisian Philology, edd. Rolf H. Bremmer Jr. et al. Amsterdam–Atlanta. 408–427. Seebold, Elmar (1991) ‘Die Stellung der englischen Runen im Rahmen der Überlieferung des älteren Fuþark’, in: Bammesberger ed. 1991. 439–569. Siebs, Theodor (1889) Zur Geschichte der englisch-friesischen Sprache. Halle/Saale. Siebs, Theodor (1901) ‘Geschichte der friesischen Sprache’, in Pauls Grundriß der germanischen Philologie. Erster Band. 2. Auflage. 1152–1464. Strassburg. Remarks /1 August 2018/38 [Siebs, Theodor (1931) ‘Die Friesen und die nächstverwandten Stämmen’. Mitteilungen der schlesischen Gesellschaft für Volkskunde 31: 44–84. ] Sievers, Eduard (1876) ‘Zum Heliand’. Zeitschrift für deutsches Altertum 19: 1– 76. Sievers, Eduard and Karl Brunner (1965) Altenglische Grammatik nach der angelsächsichen Grammatik von Eduard Sievers. 3. neubearbeitete Auflage von Karl Brunner. Tübingen: Max Niemeyer. Simon, Werner (1965) Zur Sprachmischung im Heliand. (Philologische Studien und Quellen 27). Berlin. [Sjölin, Bo (1973) ‘Anglofriesich’, in: RGA 1: 329–31. ] Sweet, Henry (1876) ‘Dialects and Prehistoric forms of Old English’ Transactions of the Philological Society (1875–76. 543–69. Taeger, Burkhard (1979–84) ‘Das Straubinger Heliand-Fragment. Philologische Untersuchungen’. Beiträge zur Geschichte der deutschen Sprache und Literatur (Tübingen) 101 [1979] 181–228, 103 [1981] 402–424, 104 [1982] 10–43, 106 [1984] 364–389. Vleeskruyer, R. (1948) ‘A. Campbell’s views on Inguæonic’. Neophilologus 32: 173–183. Watkins, Calvert (1966) ‘Italo-Celtic Revisited’, in Ancient Indo-European Dialects. Eds. H. Birnbaum and J. Puhvel, Berkeley–Los Angeles. 29–50. [Wissmann, Wilhelm (1938) Die ältesten Postverbalia des Germanischen. Göttingen: Hubert. ] Wortmann, Felix (1960) ‘Zur Geschichte der langen ê- und ô-Laute in Niederdeutschland, besonders in Westfalen’. Niederdeutschen Studien 6: 1– 23.