Sprachstrukturen
Herausgegeben von Herbert L. Kufner, Hugo Steger und Otmar Werner
Reihe A Historische Sprachstrukture...
83 downloads
1137 Views
25MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Sprachstrukturen
Herausgegeben von Herbert L. Kufner, Hugo Steger und Otmar Werner
Reihe A Historische Sprachstrukturen
3
Elmer H. Antonsen
A Concise Grammar of the Older Runic Inscriptions
Max Niemeyer Verlag Tübingen
ISBN 3-484 60052-7 © Max Niemeyer Verlag Tübingen 1975 Alle Rechte vorbehalten. Ohne ausdrückliche Genehmigung des Verlages ist es auch nicht gestattet, dieses Buch oder Teile daraus auf photomechanischem Wege (Photokopie, Mikrokopie) zu vervielfältigen. Printed in Germany. Satz: Rothfuchs Dettenhausen
Contents
Foreword
VII
Abbreviations
XI
1. 2. 3. 4. 5. 6. 7. 8.
The runes, their phonological values and transliteration . . . . Runic graphemes and their variations Dating and relative chronology Orthographic peculiarities Phonology Morphology Syntax Delimitation of languages/dialects represented in the Corpus.
Corpus of inscriptions 1-93 Northwest Germanic 94-98 East Germanic 99-106 West Germanic 107-121 North Germanic 118 West Nordic 116-117, 119-121 East Nordic Index of forms 1. Germanic 1.1 Runic 1.2 Gothic 1.3 Scandinavian 1.4 English 1.5 Low German 1.6 High German 2. Greek 3. Italic 4. Slavic .
1 6 10 12 14 19 24 26 29 29 73 75 78 85 83-85, 85-89 90 90 90 93 94 96 98 99 101 102 103
5. 6. 7. 8.
Baltic Celtic Indolranian Others
103 104 104 105
List of works cited
106
Index of inscriptions treated
110
VI
For inclusion in the present series, it has been necessary to reduce considerably the original scope of this grammar. As a result, much of the argumentation for the new readings and interpretations I propose in the Corpus of Inscriptions has had to be omitted. I hope to present that argumentation in detail elsewhere. It has also been necessary to reduce to a minimum the discussion of matters treated in the text and to restrict the bibliography to a listing of only works actually cited. Works of a general nature dealing with Germanic linguistics as well as etymological dictionaries have been excluded. In the revision, I have proceeded on the assumption that the reader will have had at least an elementary introduction to the study of the runes and will be familiar with Germanic linguistics. Under each inscription in the Corpus, I refer to the main handbooks in which it is treated and occasionally to individual treatments. An asterisk after the citation indicates that a reproduction of the inscription is to be found there. Further bibliographic references can be found in the works cited. From the very beginning, it has been my intention to focus on the linguistic significance of the inscriptions, which in my opinion has not come sufficiently to the fore in runic studies over the past decade, in spite of considerable interest in the inscriptions (cf. the corpus edition by Krause 1966, the introductory and general presentations by Elliott 1963, Müsset 1965, Düwel 1968, Krause 1970, and the discussions of the language of the inscriptions by Makaev 1965 and Krause 1971). Krause's corpus edition (1966) is indispensible, but despite this scholar's immense contributions to runic studies, it suffers from a lack of linguistic methodological rigor and from the author's convictions concerning the magical nature of the inscriptions, which often result in patently untenable interpretations and the assumption of linguistic forms and developments which are quite out of keeping with the evidence. Makaev (1965) has done yeoman service for the field by pointing out many of the shortcomings of previous work and issuing a clarion call for more rigorous methods, but he himself provides little in the way of reinterpretation and relies very heavily on the views of his VII
predecessors in his corpus. Krause's linguistic study (1971) is essentially a distillation and compilation of the views expressed in his corpus edition. There thus still remains the need for a work which looks to the inscriptions anew and attempts to interpret them from a strictly linguistic point of view. While other disciplines, such as archeology, mythology, prehistory, etc. are rightfully interested in the inscriptions, it seems to me to be beyond question that mythologists and prehistorians can make the best use of them only after a sober linguistic analysis has been presented. To tamper with the linguistic evidence by prematurely introducing unconfirmed assumptions concerning the social and religious milieu of the writers of the inscriptions is to condemn the entire field of runic studies to the status of conjecture. Unfortunately, this impression of runic studies, and particularly of the value of the linguistic evidence provided by the inscriptions, has become fairly well entrenched. I have tried to follow a rigorous linguistic approach. In the Corpus of Inscriptions are included only those which lend themselves to linguistic interpretation (i.e. inscriptions consisting only of uninterpretable sequences of runes or of single runes which may or may not have been used as ideographs are excluded). All of the so-called ,Proto-Nordic' and ,Gothic' inscriptions which can be interpreted are included, as are the earliest ,West Germanic' ones. From a comparison of these, I attempt to delimit and describe the languages and/or dialects as attested in the earliest native Germanic monuments and to point out the significance of the evidence for the study of historical Germanic linguistics. Also included are those inscriptions which present (in whole or in part) the 24-letter fubark itself. For each inscription treated, I give the name, the geographic location of the find, and the approximate archeological and/or linguistic dating, followed by the text in transliteration, a discussion of peculiar or difficult readings, the text in separate words, an etymological discussion of each form, a translation, and a concise list of references. Since proper names play an important role in deciphering and interpreting the inscriptions, each one is listed with references to the same or similar names in other dialects before the constituent elements are analyzed. It should be strongly emphasized that the PIE etyma reconstructed are intended only to suggest the PIE elements which enter into the formation of the word and should not be interpreted to mean that the forms themselves were actually present in PIE. Similarly, the translations of proper names indicate only the probable basic meaning of their elements, which may very well have been no longer apparent to those giving or bearing the names, since in many instances these nameelements are very archaic (e.g. biäawarifaz = ,defender of the covenant or oath'). VIII
The reader familiar with the inscriptions will find new readings, interpretations, and/or analyses of forms given in nos. 2, 3, 4, 6, 9, 10, 11, 12, 15, 17, 18, 21, 25, 27, 31, 34, 38, 39, 40, 41, 45, 46, 52, 53, 57, 62, 69, 72, 73, 76, 94, 99, 101, 105, 107, 108, 111, 114, 115, 117, 119, 120. In the morphology, I find it preferable to use the IE designations for stems (e.g. o- and a"-stems, instead of Gmc. a- and o-stems) and to speak of y- and wstems instead of the traditional i- and u-stems, since at least in early PIE, [i] and [u] were allophones of /y/ and /w/, respectively, and the new designations reflect more accurately the state of affairs in the PIE stem formants (i.e. full grades /-ey-, -oy-/, zero grade /-y-/, etc.). In dividing the inscriptions into the various linguistic groups, I have included under North Germanic also West Nordic (no. 118) and East Nordic (nos. 116-117 and 119—121), since all three groups are specifically Scandinavian. Within the groupings, the inscriptions are presented in roughly chronological order in so far as such an ordering is determinable. This book is dedicated to the memory of my grandfather, Anton Christian Antonsen, who aroused in me an abiding interest in things Scandinavian. To my wife, Hannelore, and to my daughter, Ingrid, I am indebted for help in the preparation of the Index of Forms and in proofreading. I also wish to express my appreciation to the publisher, Mr. R. Harsch-Niemeyer, and to the editors, Prof. Herbert L. Kufner, Prof. Hugo Steger, and Prof. Otmar Werner, for the kind understanding they have shown me.
University of Illinois at Urbana-Champaign October, 1974
Elmer H. Antonsen
IX
ace. adj. adv. Alb. Arm. attr. Av. Burg. Celt. cent. cf. cons. Da. dat. dem. dial. Dor. E EG EN encl. eNG ePG F fern. Finn. G Gall. Gaul. gen. Gk. Gmc. Go. Hitt. Horn. Ic. imp. inanim.
accusative adjective adverb Albanian Armenian attribute Avestan Burgundian Celtic century compare consonant Danish dative demonstrative dialect Doric English East Germanic East Nordic enclitic early North Germanic early Proto-Germanic French feminine Finnish German Gallic Gaulic genitive Greek Germanic Gothic Hittite Homeric Icelandic imperative inanimate
ind. Ingv. Ion. Ir. Jutl. Lat. Latv. lit. Lith. masc. med. MHG Mir. MLG Moselfrk. MW N neut. NG no(s). nom. Nw. NwG Ο OB ODa. ODu. OE OF OFris. OHG OIc. OInd. Olr. OLF ONw. OPers. OPr.
indicative Ingveonic Ionic Irish Jutlandic Latin Latvian literature Lithuanian masculine medial Middle High German Middle Irish Middle Low German Moselfrankish Middle Welsh noun neuter North Germanic number (s) nominative Norwegian Northwest Germanic object Old Bulgarian Old Danish Old Dutch Old English Old Prankish Old Frisian Old High German Old Icelandic Old Indie Old Irish Old Low Prankish Old Norwegian Old Persian Old Prussian XI
opt. OS Osc. OS1. OSw. OW OWF part. Pers. PG PIE pi. poss. pres. pret. pron. quant. R.
XII
optative Old Saxon Oscan Old Slavic Old Swedish
Old Welsh Old West Prankish participle Persian Proto-Germanic Proto-Indo-European plural possessive present preterite pronoun quantifier rune
Russ. S Scand. Serb. sgSG subst. vb. Sw. Thrac. Toch. transl. Umbr. V Ved. vs. W WG WN
Russian Subject Scandinavian Serbian singular South Germanic substantive verb Swedish Thracian Tocharian translation Umbrian verb Vedic versus Welsh West Germanic West Nordic
1.
The runes, their phonological values and transliteration
1.1 The earliest attestations of a Gmc. language are found in inscriptions in the so-called older or Gmc. fupark. This alphabet, undeniably related to the Mediterranean alphabets, is known in complete form from three inscriptions: 30 Kylver, 90 Vadstena and Motala (both from the same stamp), and 91 Grumpan, and in incomplete form from 89 Lindkaer and Overhornbaek (from the same Vorlage), 99 Aquincum, 104 Breza, 105 Charnay, and 106 Beuchte. These fubarks display a unique and uniform order, except that Kyler has ^ beforehand ^ before,^, whereas Vadstena-Motala and Grumpan have these runes in reverse order. For purposes of comparison, we can establish the following standardized fupark:
l
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
f u p a r k g w h n i j
a e p z s t b e m l
ngdo
1.2 The phonological values of most of the runes can be established with great certainty on the basis of their use in identifiable words, of comparative linguistic evidence, of their correspondence to letters in the Mediterranean alphabets (Marstrander 1928, Hammerstr m 1930, Arntz 1944, 30— 64, Krause 1970, § 21—9), and of the rune-names recorded in medieval manuscripts (Arntz 1944, 167-233, Diiwel 1968, 107-8, Krause 1970, § 15-20). The identification of R. 13 f and R. 15 Υ has caused the greatest difficulties. R. 15 Υ was long regarded as m (as in the younger fupark) and then later as an orthographic variant of R. 5 ft (Munch 1847, 333), since it corresponded to r in later Scand., e.g. ft f\^ ^ Υ = Ic. r nar ,runes'. Not until the latter half of the 19th cent, was it recognized as corresponding to ζ alternating with s in Gothic (Bugge 1865, Wimmer 1867, 32—4) and therefore the reflex of PIE */s/ through Verner's law. To distinguish Υ from ft, 19th cent, runologists devised the transliterations 1
= (original) r, Υ = R (derived from */z/). K.M. Nielsen (1970, 9) considers the deciphering of the fupark complete upon the recognition that Υ was not m, but rather an r-sound. Linguistically, | cannot be anything other than /z/ until it alternates with ft , attesting to the coalescence of /z/ with /r/. This does not occur until the NG period (see 5.6 and JacobsenMoltke 970, Steblin Kamenskij 1962). It is meaningless to speak of an .intermediate stage R' as an indication of the Scandinavian character of the early inscriptions (e.g. Diiwel 1968, 15). Andersen (1960, 393) transliterates Υ as R, but assigns it the value [z]. This meaningless practice should be eliminated, and I transliterate Υ consistently as z (in quoting other scholars, I retain their transliteration). 1.3 Still undetermined, however, was the value of R. 13 . Speculation has attributed to it the values eu (Wimmer 1874, 120), close e (Marstrander 1928), open e (v. Friesen 1933, 9), and ϊ (Krause 1966, 2) representing a sound ,between e and /', all four values derived from the OE name eoh ~ ih and the use of this rune in OE inscriptions, e.g. X ^ T N T f t H Gislheard (Dover stone). Elliott (1963, 83) states that l· in this inscription designates presumably the high front vowel between e and i generally denoted by this rune in common Germanic usage', but this ,usage' has been reconstructed on the basis of OE evidence! Steblin-Kamenskij (1959, 1962) refuted the notion of a vowel between e and / in Gmc. and concluded that J1 was superfluous from the very inception of the fupark, a point of view which is understandable if we compare the fit between the phonological system of the earliest inscriptions and the runes of the fupark: Vowels:
/i/ l
Consonants:
/ U /J\
/T/ J
/p/ £
/b/ g
/{/
N f /k/
/e~/ introduced still a third possible representation, ["] , on the basis of system balance. This stage has clearly been reached when we find e-spellings for older */x/ (whether from original */a?/ or */ai/), e.g. 24 tawide and 27 woduride. Simultaneous with the monophthongization of */ai/ > */u/ was that of */au/ > */5/, spelled ^ . Upon the raising of */o/ to */o~/, the % -rune became available as an alternate of f\ to represent [o] (ώ-umlaut of */u/) through system balance (see 4.8, and Antonsen 1970). The original function of J' was removed through reorganizations in the phonological system. Coupled with the reverse spelling -ai for original */«/, the presence of the 6th vowel graph is incontrovertible evidence that the fupark was devised to represent a language which had the PG vowel structure and had not yet undergone the monophthongization of vowel clusters in unstressed syllables. The history of writing in runes must predate our earliest inscriptions by a fairly long period, long enough to antedate the phonological changes which separate even Gothic from the rest of the Gmc.
speech area. This evidence is fatal for those theories which place the inception of the fupark as late as the 1st or 2d cent, after Christ or deny that the Goths could have been familiar with the runes. 1.6 Krause (1971, § 12, Anm.) notes that my assignation of the value */£E/ to J4 ,ist weder in der Namensform des Wortes für „Eibe" noch in der Verwendung dieser Rune in sinnvollen Wörtern begründet'. To be sure, my derivation is not founded on the OE name eoh ~ ih in medieval manuscripts. I have demonstrated how the rune became superfluous. It was retained in the fupark (and OE fuborc), just as superfluous c, w, , still occupy their places in the Scand. alphabets today. The f-rune also became superfluous in both the OE fuporc and the younger Scand. fuparks because of later phonological developments. Wimmer (1867, 32—4) saw the secondary nature of the OE name eohx for . The secondary nature of the Scand. name yr ,bow, yew' for this same rune is also clear. It could not have arisen until f had ceased to designate /z/ and been assigned the value /y/ in medieval inscriptions (Jacobsen-Moltke 980—1). There can also be no doubt that the OE name for J^ is secondary. As Krause himself points out, J1 is used in OE inscriptions with two values: i in Giselheard (Dover stone), and h in almehttig ,almighty' (Ruthwell cross). In neither case can the value have been original. In (pre-)OE, J" became superfluous after /a/ = ^ > /ae/ = ^ , which then became the normal designation of /ä;/ through system balance (see Antonsen 1967). Every instance of J'in our inscriptions is disputed or uncertain. Krause (1971, §12.2, following F. Ranke) assumes the contrast between PG */ei/ and */i/ was maintained after the cluster was monopthongized and that */ i / < */ei/ was represented by J". The only evidence for such a contrast is his own assumed value / for |! Beck (1972) has constructed an hypothesis concerning the age of the fupark based on Krause's interpretation of this one rune, but Beck's arguments must fall along with Krause's. On the basis of solid linguistic evidence, ,f can only have represented */»/, and the fupark can be transliterated as in 1.1.
2.
Runic graphemes and their variations
The runes of the 24-letter fupark occur in a number of varieties (see Arntz 1944, 65—72) determined by the chronological period, the geographic area, the substance in which they are carved, and perhaps also the ,school' to which the runemaster belonged, as well as by personal idiosyncrasies. In
many instances, we have only a vague notion of which shape was ,original' for any given rune. It will be necessary, therefore, to make a somewhat arbitrary choice of the basic shape in order to proceed with a feature analysis. I have chosen as a starting point the more common shapes found in inscriptions of Scandinavian provenience (see 1.1). 2.1 All runic shapes can be described in terms of staffs (vertical lines of full height), branches (horizontal or oblique lines), pockets (enclosed spaces), and crooks (bent vertical lines). Differences between curved and angular forms, e.g. P or f w, between straight or bent horizontal lines, e.g. [~| or [^ e, and between horizontal and oblique lines, e.g. f-| or [v] h, are not distinctive, since such variations are found in contemporaneous inscriptions and sometimes even in one and the same. A first glance at the runes reveals a basic distinction among those with two staffs, one staff, and no staff (see Table l).Of those with two, m and d display two branches proceeding in opposite directions. They therefore have the features [2 staffs, 2 branches]. The position of the branches on the staffs distinguishes m from d: in m they are at the top (i.e. above center), while those of d proceed from the top to the bottom. We may then define [top] as meaning ,proceeding from the top no farther than the center', [center] as ,at or proceeding from the center', while [bottom] will mean .proceeding from the top all the way to the bottom'. Thus, [bottom] includes a redundant [top] in all runes with this feature. Keeping in mind that e has the variants j~]and ["], while h can be |s|or |-|, these two runes differ only in the placement of the single branch at the top or center. The same features suffice to distinguish the runes with one staff and one or no branch: 1 is [1 staff, 1 branch, top], n is [1 staff, 1 branch, center], u is [1 staff, 1 branch, bottom], and i is [1 staff]. For runes with one staff and two branches, it is necessary to add another feature, [side], to differentiate those with both branches on one side (unilateral runes) from those with one on each side (bilateral runes): a, f, p are unilateral and distinguished by the features [top], [center], [bottom], respectively, while t, z, ä are bilateral with the same features. The runes with a staff and a pocket, w, J), b, differ in the position of the pocket, whereby the top one of the b-rune is redundant (cf. ^on 116 Gummarp). The r-rune is the only one with a staff and a zigzag (i.e. more than one crook). Among the staffless runes, g is the only one consisting of branches alone, ng the only one with a pocket alone, the only one with a pocket and branches. The remaining runes are those consisting merely of crooks: s, j, k. The k-rune has only one crook, while s and j have two (i.e. more than one, cf. the s-variants ^and £ ), these two
Μ
ο ο (Ν
CD
Ό S
Κ, l-i
«υ _g Μ
§
I c ο
ο
4^4
^
«4-1
Ul
ν^ν
which is an exact parallel of OpedaTs ^(, both displaying the features [pocket, 2 branches]. Another comparable variety is the ^ of 89 Lindka;r and Overhornbaek. The occurrence of reversed runes is certainly a consequence of the fact that runic inscriptions could be written from left-to-right or right-to-left, making the direction of individual runes variable. The use of inverted runes (contrary to Arntz 1944, 68) must be directly connected with the practice of combining runes to make use of a common staff, e.g. [^j^eker 75 Bratsberg, £ = az and ^ = ag 15 Kragehul. Such bind-runes are attested in the earliest inscriptions and must represent an ancient practice. To combine the a- and z-runes, it was necessary to invert one of them, i.e. ^ (attested in six inscriptions). Reanalysis of this bind-rune would result in the equations: ^ = a and j^ = z. Such reanalysis led to the inversion of other runes, even those which did not normally enter into combinations of this type, e.g. ^ , or to partial inversion as in ty(, 0(. Faulty analysis must also have been the reason for the appearance of staffs on originally staff-less runes, e.g. J^ k (cf. J^ ka). Considerable importance has been placed on such variations in determining the chronology of the inscriptions. It should be pointed out, however, that such variants also occur very early (e.g. 10 Vimose, 96 Vimose) and any dating on this basis is fraught with uncertainties.
3.
Dating and relative chronology
Since none of the inscriptions in the Gmc. fubark contain references to known historical persons or events, anything approaching an absolute dating can be done only with the aid of archeological findings. Archeologists themselves, however, often take recourse to a relative chronology based on changing styles in ornamentation and workmanship. The relative chronology of linguistic forms, where it is possible to establish earlier vs. later ones, is undoubtedly a better basis for dating. Ever since Engelhardt (1863) provided the first breakthrough toward an absolute chronology of the oldest inscriptions by proposing dates for the bog-find of Thorsberg, runologists have attempted to fit their conceptions of the timedepth covered by the inscriptions with those of the archeologists (see Wimmer 1887, 300-13; Jacobsen-Moltke 1013-20; Krause 1966, iii-iv; Makaev 12—5; K.M. Nielsen 1970). Most important are the archeological datings of 1 0vre Stabu (ca. 150 A.D.; Shetelig 1914, 5-14 and 253-4), 2 Thorsberg chape (before 200 A.D.; Norling-Christensen 1945), and the bracteates (400-550; Jankuhn in Krause 1966, 238). 10
I have no basis for arguing with the datings proposed on archeological grounds and therefore in general have retained the dates proposed by Krause and Jankuhn, although in a number of cases in which the dating is made on the basis of linguistic evidence, I have seen it necessary to diverge from Krause's relative chronology. Thus, the placing of 11 Möjbro in the middle of the 5th cent, cannot be defended on linguistic or runological criteria. The spelling -ai for the dat. sg. ending points to a date closer to that of 4 N^vling and 9 Vimose, while the forms of the runes do not contradict such an assumption. I place this inscription at about the beginning of the 4th cent. The clearly related inscriptions of 15 Kragehul, 16 Kragehul, and 17 Lindholm can be dated on archeological evidence as early as the 4th cent., and there is no linguistic or runological evidence to justify placing them in the 6th. This late date results from unfounded speculations concerning the interpretation of the inscriptions, particularly of 17 Lindholm. The much discussed inscription 21 Opedal is reinterpreted here and placed no later than the earliest Norwegian inscriptions on stone, i.e. the latter half of the 4th cent. The inscription on 45 Str^m, relegated to ca. 600 A.D. on supposed linguistic and runological grounds, should not be dated later than the bracteates and is accordingly assigned here to 450—550 A.D., as are also 12 Arstad, 46 Noleby, 51 F^rde, 52 Veblungsnes, 53 Eikeland, and the bracteates 89 Lindkaer and Overhornbaek 3, 90 Vadstena and Motala, and 91 Grumpan, all of which Krause dates between 550 and 600 A.D. Very important for the whole question of chronology is 114 Ellestad, assigned the date of 600— 800 A.D. by Krause because he assumes the retained stem-vowel before -z cannot be correct. This inscription is not discussed by K.M. Nielsen (1970), even though it presents evidence that the loss of initial /j/ is earlier than the syncope of /a/ before final /z/ (note that the epenthetic vowel before /z/ is written with e, not a or A, in 119 Stentoften and 120 Björketorp, where /a/ before /z/ has been lost; see 5.3). Ellestad must be older than the Blekinge inscriptions, i.e. somewhat before 600 A.D., and 115 Setre is approximately contemporaneous. It cannot be overemphasized that the dating of every inscription is, in the last analysis, hardly more than an educated guess. There is no absolute proof that early inscriptions on stone are younger than those on loose objects dated earlier. In most cases, archeologists are not equipped to date stone monuments, and even where graves have been found under or near one, there is no guarantee that grave and stone are contemporaneous. Concerning the inscriptions of the so-called transitional period (i.e. executed in the older fubark but with definite NG traits), there is good reason to posit the period of the bracteates (archeologically 400—550) as the approximate 11
time of the inception of the first phonological changes. Whether these changes occurred already around 400 or not until later in the 5th cent, cannot be determined. Since the bracteates show no trace of the loss of initial /j/ leading to the use of the j-rune for /a, a/ (see 4.7), inscriptions with such a use of this rune must be dated later than the bracteates, therefore 114 Ellestad and 115 Setre can be placed in the immediate postbracteate period, while inscriptions with syncopated /a/ must be somewhat later. (On so-called ,archaicizing' inscriptions, see 4.11)
4.
Orthographic peculiarities
4.1 In a number of inscriptions, NwG /if/ is spelled -ai: 4 talgidai, 9 maridai, 11 hahai, 12 winai, 60 aiwuidai, 105 fabai; see 1.5. 4.2 Nasals were normally not designated before homorganic obstruents (as also in Mediterranean inscriptions), e.g. 5 -hudaz = hundaz, but were written when open juncture intervened: 31 ungandiz, 41 unnamz. The only exceptions are 31 -gandiz, 44 brando, 58 tantulu and in the representation of the sequence /ng/ (see 4.3). In all the later dialects, PG */n/ before */h/ has been lost, in Ingveonic (North-Sea Gmc.) and Scand. also before */s/, and in Ingv. before all other fricatives. From the OE reflexes of PG */an/ Ο /δ"/) + fricative, it is clear that the preceding vowel was not only lengthened, but nasalized. Phonemically, then, the NwG reflex of PG vowel + */nh/ was still a vowel + /nh/, whether actually realized as a vowel followed by a nasal or a long nasalized vowel, cf. Burg. Hanhavaldi. In normalized spellings, I therefore write the nasal, i.e. hanhe, ansu-, and unp (see Antonsen 1972a, § 1.3.1). 4.3 A peculiarity of the fubark is the special rune Ο for the sequence /ng/ (see 1.3). Its status in intelligible inscriptions is clearly marginal, since in seven attestations of the sequence /ng/, only once does the rune occur in unadulterated form: 21 birgnggu = birgingu, and even here the preceding vowel is not expressed and the postnasal stop redundantly written g. In three cases, /ing/ is written with a bind-rune i + ng: 95 awings, 81 mairlingu, 98 marings, in two instances as i + g (with the obstruent expressed, the nasal unexpressed): 36 igaduz, 37 igijon, while once it is written i + n + g: 41 idringaz. 4.4 Another convention is the designation of long (double) consonants by a single graph, e.g. hallaz = 37 halaz. The practice has been extended to consonants separated by open juncture in 98 wihailag = wih hailag and 119 dudsA = dtfds sa, but apparently was not original in runic writting, cf. 18 minas staina, 41 unnamz, and 121 hAriwulfs stAinAz. Double designation of consonants is common in WG inscriptions, e.g. 105 iddan. Double vowel graphs do not occur, except for WG 99 aada- = *ada- = andaand laas- = las·. A single graph for two vowels over open juncture is common in the younger fubark, but does not seem to occur in the older, cf. 103 auja alawin (3 times). Another example of two graphs is probably 26 stain [a] ana (with the two words in
12
different lines). The oft-cited 27 arbijarjostez is a bahuvrihi-compound with the final vowel of the stem-formant regularly deleted before a following initial vowel, i.e. arbifarjostez (see stem-forms, 6.1.1), and not an example of one graph for two vowels. Only the enigmatic 56 niujil alu remains, where it is often asserted that the a of alu is also the final a of niujila. On bracteates, however, we have so many cases of faulty copying (Jacobsen-Moltke 792) that this one may represent a Vorlage *niujila alu with one a inadvertently omitted. 4.5 Copying and spelling errors are fairly common. Inversion of two runes: 2 owlpu= wolpu- (o as an ideograph for .inherited property' finds no support in the inscriptions and would require the name in the gen.), 81 mairlingu = märilingü, 94 -wnis = -wins, 106 buirso = buriso; the omission of runes: 26 fahido = faihido (cf. saita-), 38 -stadaz = -staldaz (cf. 29 -staldaz), 48 hite = halte (cf. hait . . .), 107 ehu = ehwü? (cf. ehwu), 112 eerilaz = 75 ek erilaz, 115 naA = nan A. An interesting error is found in 38 hlaaiwido, where the carver anticipated the following a while executing the 1, carved two branches instead of one, proceeded to carve the desired a correctly, noticed the mistake, and inserted the 1 by attaching its branch to the left-hand staff of the preceding h. In 60 frohila, the bracteate-maker overlooked one branch of d = W in his Vorlage and produced h. Crassest of all is 110 mk mrlaz for ek erilaz. There is no evidence that runecarvers ever consciously omitted vowels. 4.6 There are two instances of spellings reflecting the neutralization of the voice contrast after /s/: 3 aisgz = aiskz and 28 -gasdiz = -gastiz. 4.7 _ With the loss of PG initial */j/ in NG, the old jäeran-iune , > % became the ura-rune and, through the acrophonic principle, a new designation for /a, a/ in competition with the old ansuz-rune fc, e.g. 114 -mArAz = -märaz (with A as the transliteration of the former j-rune). A differentiation of the two a-runes is evident in certain inscriptions in NG and later, where fc was used for nasalized /a/ and the epenthetic vowel, e.g. 117 hAriwulafa, but the practice was not always observed, cf. 119 uhA and uha. 4.8 When PG */ae/ and */o"/ underwent raising in unstressed syllables to NwG /e~/ and /Ö7, the a-umlaut of PG /u/ = [o] ceased to be a conditioned variant of /u/ before these vowels. The simultaneous raising of stressed PG */57 >NwG /°/$ made possible the correlation of /o/ and /o~/ through system balance (see 1.5), so that /o/ could now be written : worahto = worhto, and also when leveled (= [ ]: dohtriz, holtijaz). In other environments, however, e.g. horna, [o] was still conditioned and therefore could be analyzed as /o/ (system balance) or as /u/ = [o] before a low vowel. In EN, WN, conditioning (nonnasal) /a/ was lost (see 5.4) and the spelling alternation u ~ o became arbitrary: /wulfz/ =117 -wulafz ~ 119 -wolAfz (OIc. ulfr), /worte"/ =111 worte ~ 109 wurte (OIc. orte), /körne"/ = 109 -kurne (OIc. korne), /runu/ =120 ronu, /ginnu-/ > 119 gin0-, and was even extended to /u"/ (and presumably /ö/): /rühäz/ = 119 -ronoz. This arbitrary interchange (nondistinction of the contrast: high ^mid) also spread to i and e and finally resulted in the elimination of M and from the younger fubark. 4.9 The /'-umlaut of nonfront vowels and the «-umlaut of NwG /a, a/ are phonemicized in EN, WN after the loss of conditioning */i/ and */u/ (see 5.4) and are designated
13
(through system balance) as the result of other changes in the EN vowel system (5.1): (examples from 119 Stentoften and 120 Björketorp) NwG /e/ > EN /ae/ e ~ i = [x] < NwG /a/ A, therefore -gestumz = [gaestumz] like felAh- ~ fAlAji- = /faelh-/ and -ekA ~ -Ak = /aek(a)/; NwG /ai/ > EN /5/ Ai (and e ~ i as long counterpart of /ae/) = [5] < NwG /ä I A: hAidz- ~ hidez- = /hiedr-/; NwG /iu/ > EN /y / iu = [y ] < NwG /Ü7 u: bAriutip ~ bAnitz = /bryt-/; NwG /au/ > EN / / Au = [9] < NwG /a~/ A: -lAusz ~ -lAs = l-lqs-l ^intermediate stage toward) /-l$s-/, with / / >_/tf/ as recorded in dAude ~ duds = /d$d-/ (system balance: /au/ > / / > / / Au = [ ] < / ( u; see 4.8 and cf. K.M. Nielsen 1960). 4.10 The breaking of NwG /e/ >EN, WN /jae/ and /J9/ is recorded in = /hjaerm-/ and 117 hAeru- = /hJ9ru-/, in which the are- rune (transliterated A) continues its original value /j/ in noninitial position (cf.^( = j 46 Noleby), while e = /ae/ (see 4.9) and /9/ (regarded by the runecarver as a positional variant of /ae/ before /u/); 120 herAmA- also = /hjaerm-/, but with the traditional, etymological spelling. 4.11 Spelling variants such as -runAz ~ -ronoz, -dAude ~ duds, hAidzfelAhekA ~ f AlAhAk, hAerAmA- ~ herAmA-, hAiderA ~ hederA have given rise to a theory of ,archaicizing' inscriptions, in which the runecarver consciously tried to reproduce outmoded spellings and forms to achieve a .chancellory style1, but erred in doing so (Jacobsen-Moltke 897;Makaev 16-19). There is no need, indeed no basis for such an hypothesis. Every spelling can be explained as well-founded synchronically in view of the phonological developments which have disrupted the phonologic-orthographic fit. Stentoften and Björketorp represent one and the same linguistic stage and both make use of .modern' and .traditional' spellings according to the carver's command of traditional spellings and his own analysis of the sounds he wishes to represent. (On d ~ b, f ~ b, -sz ~ s, see 5.6; on w ~ u, j ~ i, see 5.5)
5.
Phonology
5.1 Vowels in root syllables (see charts in §1.3, 1.4): PG */i/ (PIE */y/) >NwG /i/: ginnu, hinö, idringaz, it, skiba-, widu-, witanda-, writum; EG /i/: tila-; EN /i/: ginn-; NwG /e/ (a-umlaut) > EN /ae/: hasdra; EN /JE/: wäel« */wihla-/). P_G */e/ (PIE */e/) > NwG /e/: bera, ehwü, ek, -eka, em, erilaz, hlewa-, keban, lebro, swestar, bewaz; NwG /i/ (z-umlaut) birgingü, ist, ligi, birbijaz, brijöz, wiliz, (before nasal + cons.) finno, ingaduz, ingijön, (leveling) gibu (see 6.3), ik, -ika (see 6.2), wine « */wini-/); NwG /e"/ (lengthening) mez (see 6.2); EG /i/: -wins, WG /i/ (i-umlaut): -wid, -win; NG /e/: ehu, ek, eka, erilaz, heldaz; NG /i/ (/-umlaut): sigi·; EN /ae/: faElh-, aek, -xka, CEZ; EN /jas/: hjcerm-; EN /J9/: hjqru-. PG */a/ (PIE */a, o, a/) > NwG /a/: after, agila-, alja-, alu, ana, ansu-, arbij-, -arjöstez, brando, fara-, flagda-, frawa-, gakaz, -gandiz, -gastiz, hadu-, hagi-, hagu-, hagla, hagJu, haha, hakubo, hallaz, halli, hanhe, hart-, habu, hnabdas, hrabnaz, hrazaz, landa-, labodu, labu.jnagoz, magu, -markiz, ragina-, rannja, saba, sali-, salu, saralü, satido, sawilagaz, skabi, slaginaz, -staldaz, talgide, talijö, tantulü, tawide, tawidö, tawo, paliz, bar, brawijan, wagigaz, waja-, waje, wakraz, -warijaz, waruz; EG /a/: awings, marings, wara-; WG /a/: 14
da-, anda-, fabe, -gast; NG /a/: af-, akaz, alu, azina, fahi, halmaz, labo, , -s[ffya, pat; EN /a/: a/r/·, gaf, sä, satte, saz, staba, -barbä, bat; EN /ae/: (/-umlaut) tergal, gaestumz, hceri-, mtsgiu; / /: (w-umlaut) hqbu-. PG */u/ (PIE */w/ and syllabic resonants) >NwG /u/: gudija, -hundaz, -kundö, •mundön, un-; NwG /o/: (a-umlaut) borö, horna, worhto, (leveling) dohtriz, holtijaz, wolbu-; EG /u/: gutanio; WG /u/: buriso, unb; NG /u/: fcum-, -mttndiu; NG /o/: (aumlaut) körne, worte; EN /u/: n/ra7, -wulfa, -wulfs, -wulfz; EN /o/: (a-umlaut) borumz; EN /y/: (/-umlaut) -wylfiz. PG */f/ (PIE */T7) > NwG /!/: -widaz; see PG */ei/. PG */ / (PIE */e/) > NwG /ä~/: dälidun, fakaz, glä-, mäkifa, märide, märilingü, märiz, -nämz, -rädas, -radaz, swäba-, wage, wate; WG /ä"/: /äs·; NG /a/: -maraz. PG */57 (PIE */ö, ä~/) >NwG / "/: frödila, gödagas, höraz, töj-, wödu-, -wödz; WG /ö/: -göd; NG /ö/: hrözaz, hröze. PG */ü/ (PIE */ü7) >NwG /ü/: rü/iö, ränöz, üAa; /ü/: rrözäz, rühä" (gen. pl.), uha, Ute. _ _ _ _ _ PG */ei/ (PIE *_/ey/) > NwG_/ i /: bTda-, -gTslas, hiwigaz, Una, minas, minino, minu, ni, •ride', wigiz, -wisa, wiwaz, wiwilan, wiwio, wiz, writü; EG / i / : -rids, wih; EN / i / : n/7 cf. PG */!"/. PG */eu/ (PIE */ew/) > NwG /eu/: hleunö, leubaz, leubu, leugaz; NwG /iu/: (/-umlaut) niujila; WG /iuw/: (/'-umlaut, gemination) niuwila; EN /y/: (/-umlaut) brytib, brytr. PG */ai/ (PIE */ay, oy, ay/) > NwG /ai/: a/saz, a/sfa, aiwide, baijaz, faihidö, -faikinaz, halte, haitinaz, -hlaiban, hlaiwa, hlaiwidaz, hlaiwidö, -laikaz, laibigaz, saira-, staina, waiga, wraita; EG /ai ~ ie/: hailag ~ hätlag; NG /ä~/: (before /h/) fähide; NG /ai/: staina, taitr; EN /ä/: haedr-, stcenaz, bxiaz, wraet. PG */au/ (PIE */aw, ow/) >NwG /au/: au/a, hauköbuz, laukaz, rawn//az,_(PIE_*/okW-/) •augiz; WG /auw/: (gemination) -auwifa; NG /au/: mauna; EN /^/: -cf^rfe, -rf^cis, -
5.2
Vowels of medial syllables.
5.2.1 PIE and ePG medial syllables retained as medial syllables (for examples, see Index of Forms 1.1): PG */i/ (PIE */y/): NwG -id-, -ig-, -//- (see 5.5), -//-, -in-, -is-; WG -il-, -is-; NG -id-, -in-. PG */e/ (PIE */e/): NwG -w^-. PG */a/ (PIE */a, o, 3/)^NwG -a */-au/ > */-o7 >_NwG /-ö"/: hakubö, magöz; ePG */-an-/ >NwG^/-an/: Aaryan, keban, brawi/an, wiwilan; ePG */-oh-/ > NwG /-oh/: ingijön, -mundön; ePG */-5~-/ > NwG /-ü"/: birgingü; ePG */-um-/ >NwG /-urn/: writum; ePG */-eng-/ >EG /-ing/: awings, marings; ePG */-ag-/ >EG /-ag/: hailag ~ ?/ ^; ePG */-e-/ >EG /-i/: ftryh'i», lost in EN: brytr. (On the loss of final third syllables, see 5.4) 5.3 Epenthetic vowels (written a, A) occur within consonant clusters containing /h, r, 1, w/ (see Index of Forms 1.1), in EN also in Af*tz. EN ginAninAz has lost the
15
stem-formant /-u-/ and introduced an epenthetic vowel in the sequence /nr/. Epenthetic e occurs in the EN sequence dr ~ dz = /dr/: heedra, ha>dr-, a further attestation of the coalescence of PG */r/ and */z/ after apicals (see 5.6). 5.4
Vowels of final syllables. PG */i/ (PIE */y/, and ePG */e/ before final */z/, see below) is lost in third syllables: */-aj-i/ NwG wine; */-aw-i-/ NwG hakubö, magöz; */an-i-/ NwG -hlaiban, harijan, keban, brawijan, wiwilan; */-o"-i/ NwG birgingü; */-5h-i-/ NwG ingijön, -mundön; */-iw-i/ NG -mundiu, EN mteg[i]u; */-id-i/ EN brytib, */-iz-i/ EN brytr (cf. 5.2.2). After nonroot short vowels, it forms diphthongs (see PG */ai/). It is retained in NwG second syllables, but lost in EG, WG, and EN when nonnasalized (see ^-sterns 6.1.3; cf. EN brytr). PG */e/ (PIE */e/) > late PG */i/ before final */z/ (and then lost in thirdjyllables, see PG */i/): */-a-ez >-a-iz > - - >-ez/ NwG -arjöstez; */-ej-ez >-ij-i/ >-i-z/ NwG •gandiz; */-aw-ez > -aw-iz > -au-z > -ö"-z/ NwG magöz; */-an-ez > -an-iz > -an-z > -an-n/ NwG keban, brawijan, wiwilan; */-5n-ez > -5h-z > -on-n/ NwG ingijon, -mundon; */-ez >-iz/: NwG dohtriz, NG sigi-. It is lost finally in third syllables: */-ij-e, -ej-e/: NwG ligi, skabi; */-um-e/ NwG writum. It is retained in NwG after (on swestar, see 6.1.6) and lost in second syllables in EN gaf, wraet. PG */a/ (PIE */a, o/) is retained in NwG: ana, -eka, wraita and o-stems (6.1.1); in EG retained as connective, lost elsewhere (see 6.1.1); in WG retained as connective and when nasalized, lost elsewhere (6.1.1, 6.1.5); retained in eNG walha-, akaz, erilaz, halmaz, heldaz, hrözaz, -märaz, but lost in later NG except when nasalized: an, WN taitr, EN , hjcerm-, heedr-, heeri-, -tyss, -tysr, -wulfs, -wulfz, -wylfiz, wail-, (but NG azina, eka, mauna, staina; EN hasdra, üha, -wulfa). PG */u/ (PIE */w/ and syllabic resonants) is retained in NwG dalidun (*/-dund/) and w-stems (see 6.1.4); after short vowels, it forms diphthongs, see 5.2.2; retained in WG alu, alu-, in NG alu, -mundiu, and EN ginnu, mcEg[i]u, but lost in EN ginn-, sp NwG NG EN /e~/: NwG märide, talgide, tawide, wate, NG worte, EN ute; cf. PG */ai/; on swestar, see 6.1.6. PG */S7 (PIE */aj 07 > NwG /ü/ finally (including before */i/ lost in third syllables, see PG */i/): NwG gibü, wrTtu, haglu, EN runu, and a-stems (see 6.1.2); NwG /o/ elsewhere: gen. pi. of all stems (6.1), fem. nom. sg. on-stems (6.1.5), 1st sg. pret. wk. I (6.3); EN /a"/: runäz, siinäz, stabä, brm, baziaz, -barbä; cf. PG */au/. PG */ai/ (PIE */-o-y/) > NwG hantie, -ride, wage, waje, NG -korne; (PG */-aj-i/) NwG wine, WG fabe; (PG */-a-ez/) -arjoste'z. PG */au/,· see */-aw-/; PG */-ei/, see */-ej-/; PG */-eu/, see */-ew-/, 5.2.2. REMARKS: Since */a/ in third syllables is not lost in NwG, e.g. haitinaz, the gen. sg. of the y-, w-, and on-stems must reflect PIE */-es/ > PG */-ez/. There is no attestation of PIE PG final */e/ in second syllables; 25 was . . . ends where the stone is broken off and may have been *wase (3d sg.) or *wasa (1st sg.).NwG ist and EN CKZ « PG */esti, ezi/) lost final */i/ when without sentence stress (i.e. */i/ occurred in second unstressed syllable). The loss of */i/ in third syllables in NwG and in second syllables in the later dialects caused the coalescence of the jy-stems with the cons, stems before their subsequent coalescence with the o-stems (see Antonsen 1969—70, § 1-4; rule 11, p. 59 must read ,*/e/ and */i/ in third syllables are lost', i.e. */a/ should be excluded). That the PG ending for the dat. sg. of y- and w-stems was actual16
ly */-aji ~-eji/ and */-awi ~-ewi/ is seen from Go. -ai and -au (instead of *-a, the reg. reflex of PG */-ai, -au/). Final WG NG /-a/ < PG */-an/ in the masc. ace. and neut. nom., ace. of o-stems indicates that NwG -a, in masc. nom. sg. of ow-stems was nasalized (and therefore also retained in WG and NG), as also was the /-a/ of EN haedra. 5.5 Resonants PIE PG (consonantal) */r, 1, m, n/ > NwG /r, 1, m, n/, except that final PG */n/ «PIE */m, n/) is lost; see Index of Forms 1.1 and OM-stems (6.1.5), ace. sg. of o-, a-, y-, w-stems (6.1), 1st sg. pret. wk I, II, and 1st sg. pres. wk. II (6.3). The reflexes of PIE */y/ = [i, j, ij] and */w/ = [u, w, uw] by Sievers's law > PG */i, j, ij/ and */u, w, uw/ (see Antonsen 1972a, § 4.2) show the expected distribution in alja-, -arjöstez, auja, frawa-, harja, -harjaz, waja-, waje; sawilagaz, tawide, tawidö, tawö; arbij-, arbijanö, ingijön, makija, raunijaz, birbijaz; but irregular distribution (after short syllable) in gudija, harijan, holtijaz «PIE */kldyos/, see Lehmann 1968), talijö, brawijan, -warijaz, and (after long syllable) rannja. In the case of holtijaz, one might consider reformation after the root syllable had become long in PG */hult-/, but as seen from rannja and the five example of /-ij-/ after short syllables, we must be dealing with a morphemic phenomenon in which the original phonological rule is still visible, but no longer productive. Both niujila and niuwila are Gmc. formations based on PIE */newy-/ + */-yl-/ (an impossible sequence in PIE); */ew/ + cons, yields PG */eu/ + cons., but */ew/ + *[j] yields WG */iuw/ (with gemination; cf. OHG OS niuwi, OE niewe), so that niuwila must be a WG form. Confirmation is found in 99 andagast tasauwija, where the loss of */-iz/ in gast is clearly WG (not NG .vocative') and -auwija shows the same gemination « PG */awja-/ with /-ija/ leveled from stems with long root syllables); cf. OHG drouwen < */braujanan/ < PIE *[trowjonom] .threaten', drewit < */prawidi/ (see Antonsen 1969, 205-6). In wiwiö and gutaniö, the omission of -/- indicates it had come to serve merely as a hiatus-breaker; by NG times, all distinction between /-j-/ and /-ij-/ has been lost and the glide is written i. The disappearance of /-ij-/ and /-uw-/ caused PG */j, w/ to revert to the status of allophones of /i/ and /u/ (cf. 71 -uisa, 88 uigiz). The runes j and w became superfluous and were eliminated from the younger fupark. There is no evidence for the Verschärfung of PG */-jj-, -ww-/. (On the loss of initial */j/ in NG, see 4.7). In EN, /n/ is lost before open juncture in u- 120. 5.6 Obstruents PG */p, t, k, b, d, g, f, b, h, s, z/ are retained in NwG, except that PG final */d/ after a nonroot syllable is lost, e.g. dälidun, */-nz/ > */-nn/ > /-n/ (gen. sg. on-stems, 6.1.5), and */-gw-/ >/-w-/, e.g. bewaz. PG */b, d/ are devoiced finally in EN gaf, brytib, while */p/ is voiced intervocally in -d^de. EG -rids and EN -dtfds testify to the neutralization of the voice contrast in fricatives before /s/. PG */h/ is lost before /t/ in NG worie (cf. NwG worhtö), and before /!/ in EN vvu/-. The reverse spellings in EN Afatz = /aftr/ and hidez-, hAidz- = /hiidr-/ indicate that PG */r/ (originally a uvular trill) and PG */z/ have coalesced in an apical trill after apicals. The use of e to designate the epenthetic vowel in hederA, hAiderA = /haedra/ (as in hid^-) also indicates the apical nature of EN /r/ after apicals, whereas initial /r/ was still uvular, äs seen from gin^ninAz with epenthetic A. EN -lAs = -tyss with^he assimilation of */-sz/ >/-ss/, while -lAusz represents the same phonetic form [-l^ss], but with [-ss] analyzed as root /s/ + nom. marker = (apical) /r/, i.e. -tysr. 17
18
u
z
z
O ω
PH U
Ο
l
z z P