Back to Table of Contents

14. The Semi-Vowels

The Semi-Vowels, j w, are essentially vowels as to articulation, since the air passage is, in general, somewhat narrowed, but not suffiiently obstructed to characterize them as spirants. As reduced forms of diphthongs, they appear as pure vowels,i u (39); on the other hand, they tend to be narrowed to spirants or even stops, as in *jugóm 'yoke', Gk. ζυγόν = [dzugon]; Sk. dváyōs 'of two', tueggia; Gmc. *trewa-, tryggr (33 d). The bilabial semi-vowel (E. w) tends to become a labio-dental spirant, v, (cf. E. water: Ger. Wasser). To indicate this double character, the letters i̯ u̯ are frequently used, but there are advantages in the use, in reconstructed IE forms, of j w for true consonantic function, and of i u for pure vowels or diphthongal glides. (However, j v w are used in this book as everywhere, if historical spelling requires it.)

• j: L. iugum, Gk. ζυγόν, Sk. yugám, Go. juk 'yoke'

• w: L. vestis 'dress', Gk. (verb) 'εννυμι < ϝεσνυμι, Sk. vastra, Go. (verb) wasjan 'dress'.

THE GERMANIC CONSONANTS

A. The Consonant Shift

Back to Table of Contents

15. Lautverschiebung

Back to Table of Contents

a

'Sound Shift' was the term coined by Jacob Grimm in the second edition of his Deutsche Grammatik (1822) to designate a very large complex of interrelated phonetic changes through which the Germanic system of stops and spirants developed from the corresponding IE consonants. It is probably the most comprehensive group of sound changes that has been observed in the history of any language. But the individual changes are of such homogeneous character that finally the Germanic consonants constitute practically the same 'phonetic pattern' as the Indo-European consonants, especially if traditional bh dh gh are interpreted as voiceless spirants. Every single genuine consonant (cf. 13, 14) had altered its mode of articulation, but the fundamental types of the Indo-European and the Germanic consonants are the same.

Back to Table of Contents

b

Back to Table of Contents

c

(1) the second or High German shifting proceeds in general on the same lines as the first or common Germanic shifting;

(2) one and the same general formula is applicable to the various sets of consonants, whether they be labials or dentals or gutturals; (3) the shifting proves to imply a fixed sequence of the principal forms of the shifting, based on the arrangement of the three classes of consonants involved in the order of media, tenuis, aspirata....

Such then is Grimm's Law: not merely a set of observations on consonant changes resulting from the first Germanic shifting, but rather a general theory as to the mutual relation between certain consonant changes, occurring as a rule in combination with each other. However much Grimm may be indebted for details to his predecessors, the law remains his own, and is something very different from what others had noticed beforehand.

Back to Table of Contents

16. The Fundamental Principle.

Back to Table of Contents

a

The articulation of consonants is conditioned by two counteracting factors: The current of air issuing from the lungs, and the tension of the vocal cords and the muscles of the mouth (tongue, lips, velum, cheeks) intercepting the flow. In the case of stops, the breath is completely checked in the mouth-- by the lips in the case of labials (p b), by the tongue in the case of dentals and 'gutturals' (t d; k g). In the case of voiced sounds it is checked in the glottis, by the close approach of the vibrating vocal cords. In some languages, e.g. French, the voiceless stops, p t k, are frequently pronounced with simultaneous closing of the glottis (glottal stop), so that the audible articulation is due to the release of the muscular tension of tongue or lips. We cannot know whether the articulation of IE voiceless stops was originally the same in all parts of the Indo-European territory. It is quite possible that some of the dialects had the glottal stop articulation, but we cannot postulate this for general Indo-European, in spite of Meillet, Caractères généraux des langues germaniques 36 f. For Germanic, we must assume open-glottis articulation, and it matters little whether it represented there an original type or a later development.

Back to Table of Contents

b

Back to Table of Contents

c

Back to Table of Contents

d

All Indo-European stops and spirants pass through a consistent system of changes during the early history of the Germanic languages. In concrete terms, defining merely the actual result, its general course may be described as follows:

In the case of stops, the breath is released.

In the case of spirants, the breath is checked.

Both release and check follow the direction of the breath passing from the lungs, i.e., either process takes place first in the glottis, then in the mouth, as far as this is possible. An occlusion is opened in the case of stops, and a passage is closed in the case of spirants, but of course an open glottis cannot become opened, and a closed mouth cannot become closed. For instance: t is a voiceless stop. The glottis is open. Consequently the breath is released between tongue and teeth, and þ results. This being a voiceless spirant, the open glottis is closed, so that þ changes to ð. This voiced spirant requires an occlusion in the mouth, thus changing to d, a voiced stop, which then, by release in the glottis, becomes t. Doubtless we must assume certain intermediate steps as phonetic probabilities. Thus, t was probably in Pre-Germanic, as elsewhere in Indo-European, a pure stop, and at first the air pressure increased so that the aspirate t' resulted as a transition between t and þ; possibly there existed for a certain time, and in certain positions, a further transitional step, the 'affricate' tþ. þ was probably originally a fortis, being articulated with that force of expiration and counter acting muscle tension that had led to its formation. But this double maximum of intensity was transitory; the sound became lenis. Between d and t we are bound to assume as a phonetic interpolation the voiceless lenis ḍ. Therefore, the complete series of these changes is the following:

t > t' > þ > ð > d > ḍ > t

Back to Table of Contents

e

Grimmm's description of the process is generally represented in this form

M = media (b d g)

T = tenuis (p t k)

A = aspirata; in this term Grimm includes both the Gmc. spirants f þ ɧ and the Gk. 'aspirates', as representatives of what are now termed IE bh dh gh. This nomenclature, while phonetically inaccurate, is natural enough in a chapter that is inscribed 'Lehre von den Buchstaben', and in point of fact there is a certain practical simplicity in the application of this diagram.

Back to Table of Contents

f

The following diagram may, in a sense, be called an adaptation of Grimm's diagram to phonetic facts, but it is correct only if the IE 'sonant aspirates' are assumed to have been voiceless spirants in lenis articulation (the arrows denote the direction of the changes; the figures correspond to the arrangement of the following table):

Anticipating sections 18-21 and 24-27, we obtain on the basis of this diagram, the following chronology of the consonant shift, examples being taken from the dentals, since their shift is carried farthest:

I. Pre-Germanic: θ (= dh) > ð > Gk. θίρα: Gmc. *ðurā- (E. door)

II. Germanic: B.C., t > t' > þ— L. tres: Go þreis, E. three

III. Germanic: A.D., 'Verner's Law'— þ > ð— Gk. ματήρ: Go. faðar (spelled fadar)

IV. Germanic: A.D., d > t— L. edo: Go. itan, E. eat

V. Intermediate: ð > d— Gmc. *ðurā- > Go. daur; Go. faðar: fæder

VI. High German: t > z— Go. itan: OHG. ezzan

VII. High German: d > ḍ (> t)—Go. daur: OHG. dor, tor

VIII. 'German: þ > ð— Go. þreis: OHG. dhrī Isidor)

IX. 'German': ð > d— OHG. dhrī > drī.

Back to Table of Contents

17.

Back to Table of Contents

a

The Cause and Time of the Consonant Shift are moot questions but it is probable that they bear a relation to each other. It is hardly by mere accident that this strikingly comprehensive and homogeneous group of phonetic changes is contemporaneous with what may justly be called the most momentous national movement in history: the Germanic Migrations ('Völkerwanderung'), which began at least as early as the second century B.C., perhaps much earlier, and ended during the sixth century of our Era, so that its whole duration was not much less than a millennium.

Back to Table of Contents

b

The estimates of the Time of the consonant shift show a wide difference of opinion. Kluge represents one extreme (Urgermanisch 522) : the second pre-Christian millennium, and Much the other (Btr. 17. 63) : the third century B.C. Between these two estimates stands that of Bethge, Altg. Diai. 176: between 1000 and 400 B.C. This view is approved by H. Meyer, ZfdA. 45. 126, and Kossinna, IF 4. 49, who fix the date at about 400. The only concrete arguments consist in loan words and proper names, but the dating of the former is generally uncertain, and the possibility of sound substitution exists in both types of words. E.g., names like Cimbri, Teutones, with L. c t for Germanic spirants, do not indicate that the consonant shift had not yet taken place at the time of the Cimbrian migration; rather, these consonants are either Roman or Celtic substitutions. Cf. Hirt, Indogermanen, 2. 616; Kluge, i. c. 51 f.; Hirt, HU 1. 101 ff. Apparently the best attempt at a solution is a pragmatic hypothesis that is in keeping with phonetic as well as historical facts, but any such construction must necessarily be uncertain. In that sense, an analysis of the chronology above leads to the following conclusions:

Back to Table of Contents

c

Step I: IE bh dh gh (= φ θ χ) became b d g in all IE languages except Italic, Hellenic,' and Indie. In Germanic, where the transition is incomplete, an intermediate stage ƀ ð ȝ is certain, in the other languages that have b d g, such a phonetic interpolation is very probable. This transition is such a frequent and obvious physiological process that it does not entitle us to consider the b- d- g-languages as a prehistoric unit. The change occurred independently, but it is not a part of the Germanic consonant shift proper, although it must be included in its consideration. It must be ascribed to 'Pre-Germanic' times, between the separation of the Germanic from the Italic group and the actual consonant shift, and therefore, at a random estimate, may be said to have taken place around 1000 B.C.

Steps II and IV : The changes p t k > f þ ɧ and b d g > p t k represent a surplus of expiration over the counteracting muscle tension, either in the mouth, or in the glottis. They seem to be connected, both as an effect and as a contributing cause, with the Pre-Germanic accent change: the IE accent was musical (pitch accent), and the Gmc. accent was dynamic (stress accent) (43). This view was first expressed by Hirt, Indogermanen, 2. 616, and repeated by R. C. Boer, Neophilologus 1. 103, and Oerg. Hb. 136 f. The change of p t k to f þ ɧ cannot have been a sudden transition, but implied a gradual strengthening of expiration that extended over generations, perhaps over centuries. If we try to indicate, not its beginning, but its completion, Much's date, the third century B.C., would seem acceptable. The unvoicing of the voiced stops must have been still later; here at least we have fairly definite evidence from loan words: words like Go. Krēks (L. Graecus), Go. paida (Gk. βαίτη), Go. kanaps (L. cannabis), whatever the way of borrowing may have been, indicate that the transition was not completed before the end of the pre-Christian Era. Cf. Hirt, HU 1. 102.

Step III : Verner's Law (20) must have been almost contemporaneous with Step IV and may be ascribed to the first generations of the Christian Era-— an entirely tentative estimate.

Steps V to IX occurred during the individual history of the several Germanic languages, although at least Step V started in Germanic times. In these changes the following facts of relative chronology become cletfrly evident:

(1)The 'Intermediate Shift', i.e., the change from Prim. Gmc. ƀ ð ȝ to b d g, took place sooner in initial than in medial or final position. This change implies an increase in muscular tension, which normally is greater initially than medially. Some Romance languages exhibit a corresponding change in the opposite direction, from stop to spirant. Go. [siƀun] is the earlier form, and OHG. sibun the later one, but L.debēre became [deƀer] in Spanish, devoir in French. Prim. Gmc. ƀerana-yielded Go. bairan, and b in L. bonus remained b in Standard French, Spanish, Italian. In both groups, then, the stop is more apt to prevail in initial position, the spirant medially, although the Romance languages change the medial stop to a spirant, Germanic the medial spirant to a stop.

The recognition of the Germanic voiced spirants is due to Braune, Btr. 1. 1 ff. (Zur Kenntnis des Fränkischen und zur hochdeutschen Lautverschiebung). The view of Sweet, Meillet, Huchon, that Germanic had voiced stops, which became spirants medially and finally, is clearly wrong.

(2) In the 'High German Shift', p t k became spirants in medial or final position, but affricates (pf ts kh) in initial or post-consonantal position: E. open, eat, make =offen, essen, machen, but pound, ten, corn = Pfund, zehn [tse:n], Early UG khorn (Standard German Korn). The surplus of expiration over tension overcame the occlusion more completely after vowels, where the tension was weaker.

(3) In all phases of the Intermediate and High German Shifts dentals change first and most completely, labials later and less, and velars last and least. This is particularly marked in the High German Shift of b d g to p t k. Everywhere in Upper German, d became t, and this change extended also over the larger part of Middle German; the changes b > p and g > k occurred only in Upper German, but the former was much more far-reaching than the latter. Phonetically, this seems to mean that the ratio of change is proportional to the relative agility of the articulating organs: the tip of the tongue is the most agile, the back of the tongue and the velum are the most inert of the three articulating organs. It is not quite impossible that the relative frequency of sounds may also have played a part, at least in the case of the dentals, which are the most frequent among consonants, but on the other hand, velars are more frequent than labials, which speaks against that assumption.

Back to Table of Contents

d

Residuary Forms ('Restwörter'). In general, phonetic laws are carried through 'without exceptions' (cf. 9), but phonetic laws are limited in their duration. For instance, in Greek intervocalic s disappeared: γενεσος > γένους. Later, intervocalic t became s under certain conditions, and this new s remained: *λυτις > λύσις. In Germanic as well as in other languages it happened occasionally that a phonetic law came to a standstill before it had covered the entire vocabulary of the language. Especially connectives and other words of relatively neutral semantic contents appear as such linguistic residue. Thus in Gothic, the shift from d to t did not affect the prepositidu 'to' and the prefix dis- 'asunder', while *dwōi, *dek^ṃ(t) had become twai, taíhun. In Middle Franconian, postvocalic t became regularly zz, as in *watar > wazzar; but in pronominal forms and endings -t remained: that, et, wat, allet.

The very existence of such residuary forms ('Restwörter' ; cf. Prokosch, Mod. Phil. 26. 459 f.) seems to confirm the theory that the consonant shift had its origin in the emphatic articulation of stressed words and syllables, and this is in keeping with the order described above, in which the consonants changed according to their place of articulation and their position in the word. The 'Germanic Consonant Shift' took place in pre-historic times and was virtually completed at the time to which our earliest records belong. But the 'Intermediate' and the 'High German' shifts belong to a period of expansion and migration, the main events of which are sufficiently well known to establish a parallelism with the linguistic changes of that period, and it is fairly safe to assume that the pre-historic developments would show a similar parallelism if the details could be ascertained.

Back to Table of Contents

e

The Cause of the consonant shift is even more problematic than its time. Jacob Grimm gave a purely psychological explanation which, in spite of the Romantic exuberance of its wording, may fundamentally contain a good deal of truth. He sees in the consonant shift (the Ger manic as well as the High German) an expression of the impetuous character of the Germanic tribes during their early history. H. Meyer(-Benfey) advanced the theory that at least a large part of it was due to an increase in the force of expiration, brought about by life in mountainous districts: the Scandinavian highlands for the Germanic, the Alps for the High-German Shift. His theory has found considerable approval, expressed, for instance, by Osthoff in his lectures, and by Collitz, Language 2. 181. Lately, the 'ethnological theory' seems to be the leading view. According to this, the Germanic as well as the High German Shifts are due to language mixture between Pre-Germanic or Germanic, and Celtic, Finnic, Rhaeto-Etruscan, or the speech of an autochthonous population of North-Central Europe.

In the author's opinion, H. Meyer's view is not without a measure of intrinsic probability. But his factual arguments, showing similar phonetic processes among other mountain tribes, are scattered and insufficient. Moreover, his two chief postulates are highly problematic : If the original Germanic home was on the Scandinavian peninsula, it was not upon the high fjällar, but in comparatively level southern Sweden. This speaks against Meyer's explanation of the Germanic shift. As to the High German shift, Braune's assumption that it started in the Alps (see below) is a mere hypothesis without historical foundation. Should it happen to be correct, it would be difficult to understand why the Romance languages of the Western Alps, or the languages of the Caucasus or the Himalaya do not show any similar phenomena. Meyer's physiological explanation of certain parts of the consonant shift as the result of increased expiration is a valuable supple ment to Grimm and Raumer, but this increase cannot have been due to life in the mountains.

The varieties of the ethnological theory are too numerous to discuss in detail. In spite of some brilliant observations (especially by Kauffmann and Naumann) all of them are weak in principle. On the one hand, voiceless stops are supposed to change to spirants under the influence of absorbed languages (none of which, as far as is known, possessed such sounds to a similar extent), and on the other, for the same reason, voiced stops became voiceless. Moreover, the very multitude of guesses is self-defeating. The phonetic basis of Finnic, e.g., differs radically from that of Celtic; about Rhaeto-Etruscan (if such a language group existed) we know very little, and about the mythical autochthonous language of North-Central Europe nothing at all. Finally, the thorough consistency of this wide complex of phonetic changes is a decisive argument against foreign influence as a determining cause. Any one of the individual steps might, by some stretch of imagination, be ascribed to such— say, the spirantization of occlusives to Celtic,7 the unvoicing of b d g to Finnic or Rhaetian; but these influences would necessarily have acted independently of each other, and it is quite incredible that the phonetic laws resulting from these separate influences would accidentally have fallen into such a homogeneous pattern as is described in the 'Fundamental Principle', 16.

Back to Table of Contents

f

The following attempt at a solution of the problem is far from adeuate, since our knowledge of the historical foundations is incomplete; but it tries to avoid the worst fallacies of the geographical and ethnological theories; the historical details (based essentially on Lamprecht's Deutsche Geschichte) are given by Prokosch, Die deutsche Lautverschiebung und die Völkerwanderung, JEGPh 16. 1 ff.

(1) The Consonant Shift is one continuous process, the duration of which is approximately parallel to that of the Germanic Migration in the wider sense: it began several centuries B.C. and ended about 500 A.D. The Second or High German Shift continues or duplicates in a consistent way the phonetic processes of the Germanic Shift.

(2) The basic trend of the consonant shift developed in and near the Germanic homeland: Scadinavia, i.e. the West Baltic Basin, and Northern Germany, between Elbe and Oder. In accordance with 7, Chronology and Drift, it ceased in each tribe soon after the final settlement in the new home. Accordingly, as the following sections will show, the scope of the shift in each dialect corresponds to the relative period of colonization.

(3) These two statements are based on facts and are merely descriptive of what actually took place. But what was the cause of this 'Basic Trend'? Here we can offer only a hypothetical construction. Linguistic change is largely due to imitation. Every individual departs at times or always from the average norm of speech. To an extent these personal deviations are the result of character and circumstances. Leading personalities are apt to be imitated in their manners, their dress, their speech: they set fashions.

The Germanic Migration was the result of overpopulation of comparatively infertile land, aggravated in the northwest by inundations, and possibly in the east by pressure from Slavs and other nations. Emigration brought temporary relief for some districts, but in general the necessity of emigration continued in the homeland through these centuries. Every expedition required years of preparation and organiszation, and such continued storm and stress could not fail to develop leaders whose personalities influenced the 'behavior' of their followers. It can easily be imagined that in personalities of this type will and contents predominated over reflection and form: the influence of a Boiorix or Ariovistus upon speech as a part of behavior differed from that of a Petronius or Marcus Aurelius. Not only the consonant shift, but also the accent shift, the vowel shift, the Germanic verb system, all reflected a predominance of elements of contents over elements of form.

At any rate, the consonant shift appears to stand in chronological and causal relation to the social, economic, and emotional background of the Völkerwanderung. These chronological facts are certain (cf. 7): East and North Germanic, the oldest branches from the point of view of settlement in new homes, show only steps I-IV and a small part of step V. West Germanic north of the Benrather Linie (6 e) continues step V. High German, particularly Upper German, in the most recent Germanic settlements, completes step V and adds steps VI and VII. The parallelism is too close to be accidental.

The Germanic Shift

Back to Table of Contents

18.

The IE Spirants, φ θ χ conventional bh dh gh, appear in Germanic as voiced spirants, ƀ ð ȝ: Step I. For their origin and transcription, cf. 10 (4). It is generally asserted that the shift ph th kh > f þ ɧ and that of bh dh gh > ƀ ð ȝ are perfect phonetic parallels. This is by no means the case, in spite of the deceptive appearance of the Schriftbild. In Germanic, voiced and unvoiced sounds move in opposite directions, although the underlying 'drift' is the same. If IE bh dh gh had existed and had developed in accordance with the general trend, they would have become ph th kh— a transition that is generally, although perhaps wrongly, assumed for Hellenic and Italic, but which certainly did not take place in Germanic.

If we accept φ θ χ as tentative reconstructions, we find complete consistency. In accordance with the 'Fundamental Principle' (16) the breath is checked in the glottis, and Gmc. ƀ ð ȝ result, exactly as later in step III. This stage is generally assumed for Primitive Germanic ('Urgermanisch'); cf. especially Braune, Btr. 1. 1 ff. But even there the transition to stops had begun, at least in gemination and after nasals: theoretical ƀƀ ðð ȝȝ, mƀ nð ŋȝ (24) appear everywhere as bb dd gg, mb nd ng. The reason is obvious: gemination implies increased tension, and nasals are essentially stops, so that assimilation results. Probably also initial b d are to be ascribed to the Germanic period, although possibly some very early Runic inscriptions indicate spirantic pronunciation: ƀarutʀ, Bjorketorp = brȳtr; ðaȝaʀ , Einag = dagr; (Noreen, Aisl. Gramm. 143).

--

Examples (cf . 10, 4):

φ = bh:*bher-, Go. bairan, bera, OE OS OHG. beran

*bhātēr, Go. brōþar, brōþer, brōþor, OS brōðar, OHG. bruoder

*nebh-, nifl- 'night', nifol 'dark', OS rubal, OHG nebvi 'fog'

θ = dh:*dhē-, Go. ga-dēþs, dāþ, dǣd, OS dād, OHG. tāt 'deed'

*roudho-, Go. rauða-, rauþr, rēad, OS rōd, OHG. rōt

*medhjo-, Go. miðjis, miþr, mid, OS middi, OHG. mitti

χ = ĝh:*ĝhḷtó, Go. gulþ, gull, gold, OS OHG. gold, OSi. zlato 'gold'

*ĝheu(d)-, Go. giutan, giōta, gēotan, OS giotan, OHG. giozan

ĝheŋgh-, Go. gaggan, ganga, gongan, OS OHG. gangan

*weĝh-, Go. -wigan, vega, OE OHG. wegan

χ = gh:*ghosti-, Go. gasts, gestr, giest, OS OHG. gasts

*ghorto-, Go. gards, garþr, geard, OS gard, OHG gart (garto)

*steigh-, Go. steigan, stīga, OE OS OHG. stīgan

χw = ghw: *ghwṇtjā 'killing', gunnr, OHG. gund- 'fight'.

Note: The pronunciation of b d g in the various Germanic languages is discussed in 24. In the Gothic examples given above, ð is used for medial d in Gothic spelling, since its spirantic character is quite certain. After this, d will be used (fadar, instead of faðar).

Back to Table of Contents

19.

The IE Voiceless Stops, p t k, became Gmc. f þ ɧ, with p' t' k' as intermediate stage: Step II. According to 16 c these spirants were originally fortes, but in historical times they are, relatively, lenes. This is apparent from the following facts: In Gothic they remain voiceless in all positions (broþar, wulfis 'wolf's', taihun 'ten'). In Old Norse and Old English, f and þ in voiced surroundings become voiced, which presupposes lenis articulation; in both languages, the letter f is used for f and v, and þ and ð are used indiscriminately for the voiceless and voiced spirants; cf. 27 b: brōþor, brōþer = brōðor, brōðer are pronounced with ð, regardless of the spelling. The same is true for OS th (brōthar has the sound ð), but the voiced labial is usually spelled ƀ in Old Saxon : neƀo 'nephew'. In Old High German, medial Gmc. f is spelled f or v (nefo, nevo 'nephew') and was probably voiceless, but lenis, sharply distinguished from OHG. f < Gmc. p, which is generally spelled ff: offan 'open' (but ovan 'oven'). As to the development of þ in Old High German, see 27 b. For the Gmc. voiceless velar spirant this book uses the symbol ɧ, a modified h: it had originally the value of ch (probably in both pronunciations, according to the preceding or following vowel), but became h or disappeared in most positions. Roman historians of the older period express it by ch: Chariowaldus, Cherusci, or c: Cimbri. In the historical dialects it is weakened to the glottal spirant in initial position (Go. harjis 'Heer'); medially and finally it tends to disappear: Go. taihun, OHG. zehan, but ON tīo, tīen 'ten'. For details, see 27 d. It had become voiced ȝ in Prim. Gmc. under Verner's Law (20), but in historical times this process was not repeated.

Examples (cf. 10, 1):

p: *pətēr, Go. fadar, faþer, fæder, OS fadar, OHG. fater

*pōd-, Go. fōtus, fōtr, OE OS fōt, OHG. fuoz

*penkwe, Go. fimf, fimm, OE OS fīf, OHG. fimf

t:*treies, Go. þreis, þrīr, þrīe, OS thrie, OHG. drī

*tū, Go. ON þū OS thu, OHG. dū

tod, Go. þata, þat, þæt, OS that, OHG. daz

k^:*k^mtóm, Go. hund, hund(raþ), OE OS hund, OHG. hunt

*dek^ṃ, Go. taihun, tiō, tien, OS tehan, OHG. zehan

*deik^-, Go. teihan, tiō, tēon, OS -tīhan, OHG. zīhan

k:*kən-/kōn-, Go. hana, hane, hana, OS OHG. hano 'rooster', L. canō, ci-cōnia

*kāro-, Go. hōrs, ON OE OS hōr, OHG. huor 'adulterer, -y', L. cārus

*leuk-, Go. liuhaþ, leoht, OS OHG. lioht 'light', L. lūc-s

kw:*kwod, Go. ƕa, huat, hwæt, OS hwat, OHG. hwaz

*sekw-, Go. saiƕan, siā, sēon, OS OHG. sehan

*leikw-, Go. leiƕan, liā, -lēon, OS OHG. līhan 'lend', L. līnquō.

There is no Germanic evidence for IE aspirated voiceless stops.

Note: After voiceless spirants, i.e., Gmc. f þ ɧ and s, this shift did not take place, apparently because the surplus expiration was absorbed by these sounds. In a sense, we can speak of dissimilation: L. -spiciō, OHG. spehōn 'spy', L. stella, OHG. stern(o) 'star', L. scindō, Go. skaidan 'separate', L. neptis, OHG. niftila, NHG (borrowing from Low German) Nichte 'niece', L. octō, Go. ahtau 'eight'. In some cases, this Gmc. spirant goes back to IE b bh, g gh before a voiceless consonant, e.g. L. regō— rictus, Go. raihts 'right'; IE ĝhoŋghti-, Go. gāhts 'walk', IE məghti-, Go. mahts 'might'. — For words like IE keudh-tó-, Go. huzd 'hoard' cf. 28.

Back to Table of Contents

20.

Back to Table of Contents

a

Verner's Law

After Step II of the Consonant Shift, Primitive Germanic, like Primitive Indo-European, had four voiceless spirants, namely, f þ ɧ < IE p t k, and s = IE s. In voiced surroundings, these tend to become voiced, ƀ ð ȝ z. The same trend is frequently found in other languages. Thus IE bh dh are voiceless in Umbr. tefe, rufru, Osc. mefiai, but voiced in L. tibi, rubro, mediae; IE ĝenesā > Italic genezā > L. genera. Gothic has a voiceless spirant in brōþar, wairþan, but the corresponding Norse and West-Germanic words show voicing: ON [brōðer, verða], OE [brōðor, weorðan]. In many instances, however, the voicing dates back to Primitive Germanic times: Go. sibun [siƀun], fadar [faðar], tigjus [tiȝjus], seofon, fæder, -tig. The divergence itself had been recognized by Grimm, who coined the term 'Grammatischer Wechsel' (see below) for its most characteristic type. W. Braune, Btr. 1. 513-27 (1873), systematized this group of forms ('Ueber den grammatischen Wechsel in der deutschen Verbalflexion'), without, however, offering any explanation of the apparent irregularity. The reason was approximately recognized by E. Sievers in 1874, who stated it in a letter to Braune (published by H. Osthoff, Die neueste Sprachforschung und die Erklarung des indogermanischen Ablauts 13, Heidelberg 1886), but did not publish it. In 1877, the Danish linguist Karl Verner found the cause of the divergence independently and published it in KZ 23. 97 ff., under the title 'Eine Ausnahme der germanischen Lautverschiebung'. He showed that the Germanic strong verb has voiceless spirants in medial position in those forms where in Sanskrit the accent is on the root, but voiced spirants, where Sanskrit has the accent on the ending; Sk. vártati 'I turn', vavárta 'I have turned', vavṛtimá 'we have turned', vartāná- 'turned' (past part.) —Gmc. werþið, warþ: wurðum, wurðan-.

Starting with this observation, Verner formulated the following law: The Germanic voiceless spirants remained voiceless, if the preceding syllable had the IE accent, but became voiced in voiced surroundings. if the preceding syllable had been unstressed in IE times. (Sievers: 'Im Nachlaut der indogermanisch unbetonten Silbe'). Paul (Btr. 6. 538) gave the following wording: 'Die nach Vollzug der germanischen Lautverschiebung vorhandenen vier harten Reibelaute h þ f s sind ausser in den Verbindungen ht hs ft fs sk st sp erweicht, wenn der nachst vorhergehende Sonant nicht nach der idg. Betonung den Hauptton trug.'1

-—This is Step III of the Consonant Shift.

We find the same sound change in Modern Englist in such forms as exámine, extért (with gz) as compared éxit,éxercise (with ks). In spite of this well established parallel, the phonetic explanation is difficult. Probably we have to assume the following: In medial and final position, the Germanic spirants were relatively weaker than initially; their articulation may have approached that of IE φ θ χ (bh dh gh), but, as in E. exit, it was relatively stronger at the end of a stressed syllable than after an unstressed syllable, as in E. exert. In the latter position, the lenis spirant was readily affected by the vocal vibrations of the surrounding voiced sounds, and thereby became voiced. In a later period, this voicing of spirants in voiced surroundings became general, regardless of accent. In spite of the title of Verner's article, Verner's Law is not an exception to the Consonant Shift, but an acceleration of it. (Collitz, Language 2. 177: 'The instances covered by Verner's Law constitute an exception to Grimm's Law only in the sense of an accelerated action of the latter, caused by the IE accent.') The change of voiceless to voiced spirants follows the general trend of the Shift in accordance with the Fundamental Principle given in 16 d. Spirants being open consonants, an occlusion must take place; the glottis being open, this occlusion must be formed in the glottis. After an originally unstressed syllable, the lenis development, and accordingly the voicing, took place sooner than in the relatively strong position in the transition from a stressed syllable. But gradually this voicing becomes general. Thus, Verner's Law is a typical instance of the very frequent gradual spread of a phonetic law. It starts under the most favorable conditions, but spreads until it has covered the whole field. In the second Shift and in Early English (partly late Old English, partly early Middle English) we find an apparently different, but really analogous process in the treatment of th. It became voiced (in voiced surroundings) in medial position without exception (clothes, bathing), but initially only in words which are relatively unstressed in a sentence, or which have little semantic function: the, that, this, then, there, thou, thee, thine. Here the whole word is 'lenis', and therefore the weak spirant is voiced; in the case of Verner's Law, the spirant was lenis because it continued the degree of strength of the preceding syllable.

-—Scandinavian shows a similar development, beginning in the thirteenth century: Initial þ, appears regularly as t (þing > ting), but as d in the same type of words that have ð in English (de, dem, den, det, der, du, dig, din, etc.).

Note: R. C. Boer, Neophil. 1. 110, and Oerg. Hb. 123 f., ascribes Verner's Law to double accentuation in Primitive Germanic. He assumes that the syllable that had been stressed in IE times, but became unetressed through the Germanic accent change (48), retained higher pitch. For instance, IE pətēr > Gmc. faþǣr > faðǣr would, in Germanic, have a dynamic accent of the first syllable, but higher pitch on the second syllable. Inherently, there is nothing impossible in this; we do find similar conditions in modern Swedish and Serbian. But the explanation falls short of explaining Verner's Law in the case of final consonants, as well as the analogous processes of later periods in English, Scandinavian, and Old High German (see 27).

A similar hypothesis had been advanced by R. Gauthiot, MSL 11. 193-7; Prokosch, JEGPh. 11. 1 ff., tries to show that Verner's Law can be explained on the basis of dynamic accent. Cf . also Kip, MLN 20. 16 f.

---

Back to Table of Contents

b. Chronology

Unless Boer's theory be accepted, Verner's Law must have preceded the Germanic accent shift (43). This would make the first or second century A.D. the most probable time. At that period, the separation of the Goths from the western and northern Germanic tribes was under way. Now, in Gothic, Verner's Law is much more rare than in the other. Germanic languages (see below). It is commonly assumed that it had existed there to the same extent, but was 'leveled out'. But it is more probable that Hirt, HU 1.148 (cf. also 155) is right when he says 'Ich halte es fur unmöglich, alles dies auf Ausgleichung zurückzuführen.' He assumes that Gothic had carried out the accent shift sooner than the other Germanic languages, so that only a comparatively small group of rather isolated words were still subject to the law. That is quite possible, and I have held a similar view for many years. The accent shift, like other phonetic laws, was a gradual process. There must have been a time when individual speakers would, in the same word, use sometimes the Germanic, sometimes the Indo-European accent. This was favorable to the development of a semiconscious differentiation between the two accent types, resulting in Verner's Law.But at approximately that time the phonetic drift in Gothic was retarded through its separation from the main stock (cf. 7 and 9). In general, the spirant remained voiceless in words where the preceding syllable was stressed in some forms, unstressed in others. But it was voiced where the preceding syllable was unstressed throughout. Thus, Gothic has wairþan, warþ, waurþum, waurþans, although the root was originally unstressed in the last two forms. But Verner's Law did take effect in words like sibun < *septḿ, fadar < *pətēr, where the suffix accent had become fixed in IE or very early Gmc. times. Of course, even so we may speak of a sort of leveling, but only in the sense that the analogy of other forms of the same word prevented the change of waurþun to *waurdun. —The instances of Verner's Law in Gothic are listed by F. A. Wood, Verner's Law in Gothic, 1895.

Back to Table of Contents

c. Instances

Due to later consonant changes and to peculiarities of spelling, Verner's Law is not equally traceable in the several Germanic dialects. The table below shows that, in general, only Gothic, Old Saxon, and Old High German are apt to give evidence of V.L. in all four places of articulation. ON distinguished neither the labial nor the dental spirants; OE did not distinguish the labials. However, the table does not take into account some special cases, such as the change of nþ to nn (finna: Go. finþan) and lþ > ll, but lð > ld (ellre: aldenn) ; Heusler, Aisl. Elb. §159. In ON and f denotes both the voiceless and the voiced spirant, according to position, and the letters þ and ð are used interchangeably, also denoting either the voiceless or the voiced spirant, according to position (27 b, c). The table applies to medial or final position only.

Gmc.

f

=

Go.

f

ON

f

OE

f

OS

f

OHG

f

"

ƀ

"

b

"

f

"

f

"

ƀ

"

b

"

þ

"

þ

"

þ ð

"

þ ð

"

th

"

d

"

ð

"

d

"

þ ð

"

þ ð

"

d

"

t

"

ɧ

"

h

"

-

"

-, h

"

h

"

h

"

ȝ

"

g

"

g

"

g

"

g

"

g

"

s

"

s

"

s

"

s

"

s

"

s

"

z

"

z

"

r

"

r

"

r

"

r

In the following instances leveling, including the assumed retention of voiceless spirants in Gothic, is indicated by brackets. For a more complete list, see Streitberg, UG 127 ff., and Noreen, UL 124 ff. Cf. also 63.

	Non-Gmc.		Go.	ON	OE	OS	OHG
f > ƀ	Sk.	tarpáyati	þaurban	þurfa	þurfan	thurban	durfan
	Gk.	επτά	sibun	siau1	seofon	sibun	sibun
	L.	caput	haubiþ	hǫfoþ	hēafod	hōbid	houbit
þ > ð	Gk.	εκατόν	hund	hund	hund	hund	hunt
	L.	altus	alds (noun)	aldenn	eald	ald	alt
	Gk.	πατήρ	fadar	faþer	fæder	fadar	fater
ɧ > ȝ	Gk.	δεκάς	tigus	tiger	-tig	-tig	-zug
	L.	dūco	[taúhans]	togenn	togen	gitogan	gizogan
	L.	cunctārī	[hāhan]	hanga	hangian	hangon	hangēn
s > z	L.	auris	[ausō]	eyra	ēare	ōra	ōra
	Osc.	mais	maiza	meire	māra	mēro	mēro
	L.	-ōrum	þizē	þeira	þāra	thero	dero

Back to Table of Contents

d.

Grammatical Change is primarily the effect of the application or non-application of Verner's Law in different inflectional forms of the same word, due to the movable accent of Indo-European (43), but in a wider sense the term is often also applied to the alternation of Gmc. voiceless and voiced spirants in different words of like formation, or in the same word in different Gmc. languages. To the first class belongs, e.g., OE pres. cēosan 'choose', past part. coren; to the second class, on the one hand, Go. fadar < *pətēr against brōþar < *bhrātēr, and on the other hand, Go. dauþa- against dēad 'dead'.

The following types of Grammatical Change are especially important:

(1)For the principal parts of the strong verb the standard view assumes that in all seven classes the first two forms (pres. and singular of the pret.) had in IE root accent, the other two (plural of the pret. and past part.) suffix accent. As shown in 63, this is probably true only for the first three classes, but through analogical transfer Grammatical Change does occur, more or less sporadically, also in the fifth, sixth, and seventh classes. Gothic retains (or restores?) the voiceless spirant in the regular strong verb, but Grammatical Change is found in some forms of the preterit-presents.

Instances (analogical forms are given in brackets):

I.	Go.	leiþan	laiþ	[liþum]	'go'
	ON	līþa	leiþ	liþom
	OE	līþan	lāþ	lidon
	OS	līthan	lēth	lidun
	OHG	līdan	leid	litum
	Go.	sneiþan	snaiþ	[sniþum]	'cut'
	ON	snīþa	sneiþ	sniþom
	OE	snīþan	snāþ	snidon
	OS	snīthan	snēth	snidun
	OHG	snīdan	sneid	snitum

II.	Go.	kiusan	kaus	[kusum]	'choose'
	ON	kiōsa	kaus	kørom
	OE	cēosan	cēas	curon
	OS	kiosan	kōs	kurun
	OHG	kiosan	kōs	kurum
	Go.	tiuhan	tauh	[tauhum]	'pull'
	ON	tiōa	-	-
	OE	tēon	tēah	tugon
	OS	tiohan	tōh	tugun
	OHG	ziohan	zōh	zugum

III.	Go.	wairþan	warþ	[waruþum]	[waurþans]	'become'
	ON	verþa	varþ	urþom	orþenn
	OE	weorþan	wearþ	wurdon	worden
	OS	werthan	warth	wurdun	wordan
	OHG	werdan	ward	wurtum	wortan
	Go.	finþan	fanþ	[funþum]	[funþans]	'find'
	ON	finna	fann	fundom	fundenn
	OE	[findan]	[fand]	fundon	fundenn
	OS	fīthan	[fand]	fundun	fundan
	OHG	findan	fand	funtum	funtan

V.	Go.	wisan	was	wēsun	-	'be'
	ON	vesa	vas	[vǫrom]	veret
	OE	wesan	was	[wǣron]	-
	OS	wesan	was	[wārun]	-
	OHG	wesan	was	[wārum]	-
	Go.	qiþan	qaþ	qēþum	[qiþans]	'speak'
	ON	kueþa	kuaþ	kuǫþom	kueþenn
	OE	cweþan	cwæþ	[cwǣdon]	cweden
	OS	quethan	quath	[quādun]	giquedan
	OHG	quedan	quad	[quātum]	giquetan

VI.	Go.	[slahan]	slōh	slōhum	[slahans]	'strike'
	ON	[slā]	slō	[slōgom]	slegenn
	OE	[slēan]	slōh	[slōgon]	slagen
	OS	[slahan]	[slōg]	[slōgun]	gislagen
	OHG	[slahan]	sluoh	[sluogum]	gislagan
	Go.	[hafjan]	hōf	hōfum	[hafans]	'seize'
	ON	hefia	hōf	hōfom	hafenn
	OE	hebban	hōf	hōfon	hæfen
	OS	hebbian	hōf	[hōƀun]	gihaƀan
	OHG	[heffen]	[huob]	[huobum]	gihaban

VII.	Go	[fāhan]	faifāh	faifāhum	[fāhans]	'catch'
	ON	[fā]	[fekk]	[fengom]	fengenn
	OE	[fōn]	[fēng]	[fēngom]	fangen
	OS	[fāhan]	[feng]	[fengum]	gifangan
	OHG	[fāhan]	[fieng]	[fiengum]	gifangan

Preterit Presents:

I.	Go.	aih	aigum	infin.	aigan	'own'
	ON	ā	eigom	"	eiga
	OE	āh	āgon	"	āgan
	OS	*ēh	ēgun	"	ēgan
	OHG	*ēh	eigun	noun	eigan

III.	Go.	þarf	þaurbum	infin.	þaurban	'need'
	ON	þarf	þurfom	"	þurfa
	OE	þearf	þurfon	"	þurfan
	OS	tharf	thurƀun	"	thurƀan
	OHG	darf	[durfum]	"	[durfan]

(2)There existed certain present types with IE suffix accent (63 a), notably the aorist presents, the verbs with n-infix, and the causatives. In spite of a great deal of leveling, these still give in some cases evidence of Grammatical Change:

Aorist Presents:

I. IE v)et(n)k-/wi(n)k* (L. rtncd): Go. weihan 'fight'—ON vega, v& < *waih, vfgom, vegenn; OE vflgan; OHG vbar-wehan 'conquer'; OE OS uHgand, OHG uHgant 'fighter' II. IE bh£uk-/bhuh. (Lith. buklits 'sly') and bhhig-/bhvg'- (L. fugit): Go. biugan 'bend', ON bogenn (past part.), OE bUgan, OS 'bOgan, OHG biogan, but OHG buhU 'hill' VI. IE kbp-JJap- (L. capid): Go. hafjan, OE hebban, etc. (see above).

Verbs with n-Infix: IE std-/st& (L. stare—status) : Go. standan, stop, stdpum, standans; ON standa, stdp, stdPom, stapenn; OE standan, [stdd], [stddon], standen; OS standan, [stdd], [stddun], astandan; OHG stantan, [stuont], [stuon-tum], gistantan. IE trenkr.: Go. preihan (transferred to the first class) 'throng', ON pryngua, OE pringan, OS thringan, OHG dringan (Gothic general izes h, the other dialects 5 through all forms).

Causatives: IE w6rt-e-ti 'he turns', intr., wort-eje-ti 'he makes turn, he turns', trans.; cf. Sk. vdrtati —vartdyati: Go. frawairpan 'spoil', intr., frawardjan 'spoil', trans. Go. leipan, OE Upon, etc. (see above) 'go' —Go. *laidjan 'lead', OE Ixdan, OS ledian, OHG leiten. Go. ganisan, OE genesan, OS OHG ginesan 'recover' —Go. nasjan, OE nerigan, OS nerian, OHG nerien 'save'. Go. lisan 'know' —Go. laisjan, OE Ueran, OS lerian, OHG ISran 'make know, teach'.

(3) Often the same dialect offers different word formations with original accent variation and, therefore, Grammatical Change:

OHG. durfan 'need' -—darbēn (Go. gaþarban) 'abstain'

OHG. hof 'court'—- hūbesch 'courtly, pretty'

OHG. fāhan 'catch'— fuogen 'join'

dēaþ, OS dōth, OHG. tōd 'death'— dēad, OS dōd, OHG. tōt 'dead'.

(4)In other cases, we find differences of the same type between dialects; in such cases, generally, but by no means always, Gothic shows the voiceless spirant: Go. ufar 'over' indicates the accent of Sk. úpa, OHG. ubir that of Sk. upári, Gk. 'υπέρ; Go. alþeis 'old' shows root accent, ON >i>aldenn, eald, OHG. alt, suffix accent (*altós); the stem of the adjective 'dead' is dauþa- in Gothic, but *dauða- in West Germanic: dead, OS dōd, OHG. tōt (dauþr could, theoretically, go with either of the two forms, but doubtless has to be classed with the WGmc. forms). Likewise Go. ganauha 'sufficiency', ganōhs 'enough', but OS ginōg, OHG. ginuog; also Go. fraþi 'sense', gabaurþi- 'birth', gaqumþi- 'meeting', hūhrus 'hunger', and many others have cognates with voiced spirants in West Germanic and Norse. In such cases, the assumption of leveling is not sufficient; we must assume, with Hirt, l. c., that in these words Gothic had root accent sooner than the other Germanic languages. -—Variations between the other Germanic lan guages are less frequent: OHG. haso, but hara, here 'hare'.

Note: According to Bugge, Btr. 12. 309 ff., 13. 167 ff. and 311 ff., Verner's Law, under certain conditions, also affected initial spirants. Probably, this can be accepted only for the second parts of compounds, such as OHG. mezzirahs (but also mezzisahs) 'eating-knife', Messer < *mati-sahs, OS mezas < *metsahs, meteseax. However, the equation of Gmc. ga-, gi- with L. co(n) seems to find justification in the assumption of Verner's Law initially, in the same sense in which we may apply the term to E. the, that, Norse de, det, etc. (see above). Unstressed Gmc. ɧa- had lenis pronunciation at an earlier period than accented syllables with initial ɧ, and thereby became subject to the operation of Verner's Law.

Back to Table of Contents

21.

The IE Voiced Stops, b d g, became voiceless: p t k— Step IV. The intermediate stage was doubtless the voiceless lenis, ḅ ḍ .g. Aside from the obvious phonetic probability, this is shown by the parallelism of the second shift (26). In general p t k are aspirated fortes in the Germanic languages, but exceptions exist in Dutch and High German.

It is not known why b was so extremely rare in Indo-European, but it can hardly have been entirely accidental. Balto-Germanic cognates with IE b are fairly numerous, and Slavo-Germanic cognates almost equally so. But there are extremely few reliable etymologies of this kind that can be called 'Indo-European'. Perhaps a consideration like this may, in the course of time, lead to an understanding of the problem: 'Indo-European' is just as little a fixed linguistic system as is 'Germanic', or 'Slavic', or 'Indic'. It had been in flux for an indefinite period before the time of our reconstruction, which is necessarily fixed in an arbitrary way. There are indications that it was going through a pre-historic process similar to the Germanic Consonant Shift. According to 24, the stop b < ƀ was rare at a given period in early Germanic, on account of the chronological difference in the development of the three places of articulation. Perhaps something similar had taken place in early Indo-European. The voiced dental spirants, precursors of d, had perhaps already completed their transition; the 'gutturals' may still have been in the spirantic stage, so that IE χ(gh) would represent an earlier chronological stage than θ(dh), just as k does not correspond to t, which is a later development (k — Gmc. k, IE g; t — Gmc. d, IE θ). The IE labials may have been approaching the occluding stage at the time of the dialect separation. All of this is admittedly vague and open to some obvious objections, but we can hardly get any farther until new methods are devised.

Instances: (Cf. 19, 3)

b:IE skəb-/skāb-, L. scabō, Go. skapjan, skape, scieppan, OS -scapan, OHG. scepfen 'shape' (but also the root doublet IE skəbh-, Go. skaban, OHG. scaban 'scrape'

IE terb-/trb-, L. turba 'troup', Go. þaurp 'field', þorp, þorp, OS thorp, OHG. dorf 'village'

IE dheub-/dheup-, lith. dibùs, Go. diups, diūpr, dēop, OS diop, OHG. tiof 'deep' (but also dūfa, dȳfan 'dive')

d: IE deˆkṃ, L. decem, Go. taihun, tiō, tīen, OS tehan, OHG. zehan 'ten'

IE dwōi (dwōu), L. duo, Go. twai, tveir, twā, OS twā, OHG. zwei 'two'

IE dom-, L. domus, Go. timrjan, timbra, timbrian, OS timbrōn, OHG. zimbarōn 'build'

ĝ: IE ĝeus-, L. gustāre, Go. kiusan, kiōsa, cēosan, OS kiosan, OHG. kiosan 'choose'

IE ĝen-, L. genus, Go. kuni, kyn, cynn, OS kunni, OHG. kunni 'kin'

IE ĝneu-, L. genu, Go. kniu, kni, cnēo, OHG. kniu 'knee'

g: IE aug-, L. augeō, Go. aukan, auka, ēacian, OS ōkian, OHG. ouhhon 'increase'

IE gel-, L. gelidus, Go. kalds, kalþr, ceald, OS kald, OHG. kalt 'cold'

IE jugo-, L. iugum, Go. juk, ok, geoc, OS juk, OHG. joch 'yoke'

gw: IE gwem-, L. veniō, Go. qiman, kona, cuman, OS kuman, OHG. koman 'come'

IE gwīno-, Gk. γυνή, Go. qinō, kona, cwene, OS OHG. quena 'woman'

IE gwīwo-, L. vīvus, Go. qius, kuikr, cwic(u), OS quick, OHG. queck 'alive'.

Back to Table of Contents

22.

The Germanic Consonant Lengthening.

Indo-European possessed no original 'double' or 'long' consonants (geminates), but doubling resulted frequently from composition or derivation: IE wid-to, wit-to- 'known' = OHG. giwisso. The Germanic development of such groups is treated in 29, but one type is so closely connected with Steps III and IV of the Consonant Shift that it is anticipated here. This is the treatment of Germanic stops and spirants followed by n. Such groups occurred chiefly in certain derivative verbs, such as Gk δάκνω, Sk. krīṇāmi, and in the 'weak cases' of the n-declension, such as Gk. 'αρνός, gen. sg. of 'αρήν 'ram'. There are so many irregularities and so few non-Germanic equations in these forms that as yet complete agreement has not been reached, but the following must be considered the standard view: IE p t k under Verner's Law became identical with IE φ θ χ (bh dh gh), developing to Gmc. ƀ ð ȝ. These voiced spirants became stops, b d g, after nasals (24 b), and apparently also before n. Thus, before an accented n-suffix the three series p t k, φ θ χ (bh dh gh), and b d g became identical, namely, bn dn gn. Through assimilation of n to the preceding voiced stops, bb dd gg resulted, which became pp tt kk in Step IV. It is clear that this process occurred after Verner's Law and before the unvoicing of the voiced stops, unless we assume that the latter preceded Verner's Law, and that bb dd gg became voiceless through a second process of unvoicing; that is possible but unlikely.

Instances:

(The IE equations are in most cases mere formulae of reconstruction. This was essentially a Germanic process, whether the standard view be accepted in principle or not.)

• pn' > pp: IE kup-ná- (OSL. kypēti) Go. *huppōn, hoppa, hūpfen
• tn' > tt: IE snit-nó- (Go. sneiþan, etc.) snitzen 'whittle' (intensive formation)
• dn' > tt: IE (s)tud-nó- (L. tundō, Sk. tudáti, Go. stautan, etc.) stutzen 'be startled'
• kn' > kk: IE ḷk-nā (L. laciō) lokka, geloccian, OHG. locchōn 'entice'

IE feminine suffix -nī (-nyā), as in Sk. patnī, Gk. πότνια 'mistress', Ricke, fem. of Reh (Gmc. ráiha-; *rignjá > riggja)

• gn' > kk: IE lug-nó- (Lith. lùgnas 'pliable', Gk. λύγος 'twig', λύγινος 'pleated') lokkr (but lykna 'bend the knees'), loc(c), OHG. loc 'lock, curl'
• IE bhəgnó- (Gk. φώγω) OHG. bacchan 'bake'
• ghn' > kk: IE ligh-ná- (Gk. λείκω 'lick', λιχνεύω'taste', L. linguo, Go. bilaigon 'lick') liccian, OHG. lecchōn 'lick'.

But we find Gk. 'ιπνός Go. aúhns 'oven' (ogn and ofn, ofen, OHG. ofan); Gk. τέκνον 'child' : OHG. degan 'warrior', in spite of the Greek root accent (the etymology is very uncertain). In the numerous n-derivatives of verbs (Gothic fourth weak class) we should expect suffix accent and therefore assimilation of n, but at least in Gothic, and partly in Norse, n is preserved: Go. -waknan, waknōda. Norse vakna, vaknada. Gothic has only two words with tt : atta 'father', skatts 'hoard', both uncertain as to etymology; two borrowed words with kk: sakkus 'sack' and smakka 'fig'; and no words with pp.

After a long syllable these geminates were simplified:

The Germanic gemination explains numerous double forms, such as schneiden and schnitzen, stossen and stutzen, Reh and Ricke (also Ziege and Zicke), given above. Simplification after a long syllable produces a different type of parallels. Gmc. knuppa- > Knopf 'button', but Gmc. knauppa-, knaupa- > knōp, Knauf 'hilt'. knübel 'knuckle' shows that the labial was not IE b, but either φ (bh), or p under Verner's Law. In High German, -p became f, but pp became pf. It is not always possible to decide whether such double forms are due to the Germanic or to the West Germanic Consonant Lengthening. Cf. 30.

Probably very few cases of Germanic Consonant Lengthening actually go back to Indo-European forms. A few of these, like IE lugnós, may have established a pattern on which many new forms were modeled in the several Germanic languages. Once geminates had been established by assimilation, they could easily become the instrument of sound symbolism; through this, 'bildete schon das Urgermanische partielle Neuschopfungen durch Dehnung des wurzelauslautenden Konsonanten eines schon vorhandenen Wortes zur Bezeichnung der Intensität, ähnlich wie zu gleichem Zweck das Semitische durch Dehnung des mittleren der drei Wurzelkonsonanten.' (Loewe, GS 77 f.; similarly Braune, Ahd. Gramm. §95 Anm. 2.)

Back to Table of Contents

23.

The Labiovelars.

Back to Table of Contents

a

The most important development of these a sounds in other Indo-European languages have been given in 11. Their Germanic development is shown in the following table:

The treatment of the labiovelars in Germanic is similar to that in Latin. Cf. for kw, sequor—secutus—socius; for gw, veniō < *gwṃjō, vivus < *gwīwos ; for ghw, ninguit—nīcs = nīx—nīvem.

for gw, venid < *gwrnjd, vivus < *gwiwos ; for ghw, ninguit—nics = nil—nivem.

Back to Table of Contents

b

The general principle is this: The velar element (k g gh) goes through the regular shift, including Verner's Law (indicated where needed by ´ for stressed, ` for unstressed syllable preceding). The labial element is retained under certain conditions, lost under others. But Gmc. ȝw is a 'phonetic interpolation' —an intermediate stage that must be assumed theoretically, though it is not preserved historically. It appears either as ȝ or as w; only after n (ŋ), Gothic and Norse have gw. This is exactly like the Latin treatment of IE χw (ghw, Brugmann's q*h): IE neiχw- (Gk. νείφει), ninχw- > nīg-, nīw-, ningw-.

As in Latin, the labial element disappears before back vowels (IE u ū o ō—therefore secūlus, but sequitur; sequor is analogical formation for *secor, like equus for *ecos). It also disappears before consonants (cf. socius, secta), and doubtless too in final position (L. que = -ɧ in Go. sah, nih; Go. saƕ, laiƕ) are analogical forms).

Back to Table of Contents

c

Instances:

• IE kw > Gmc. ɧw/ɧ:
• IE kwe-/kwo- (interrogative pronoun), Go. ƕis, OHG. hwes 'whose' — OSw. har, but also analogical hwar = Go. ƕas 'who'
• IE kwel-/kwol- 'turn' (Gk. τέλος 'goal', πόλος 'axis', OSL. kolo 'wheel', L. collum < *kwolsom), huel 'wheel' (hweogol *hweȝla- < IE kwe-kwlóm, reduplicated form similar to Gk. κύ-κλος)— Go. hals < *kwolsos.
• IE kwei-/kwoi- (L. quies), Go. ƕeila 'while'— haims 'home'.

Note: Gmc. ɧw can also come from IE ^kw, as Sk. śvētas, Go. ƕeits 'white', Sk. aśvas, L. equus, Go. aíha- 'horse' (iōr < *ekwos > *eɧwaz > *joɧaz > *johr, jōr). In a few instances, it apparently goes back to a plain velar followed by w, instead of a real labio-velar, in which the lip-rounding extends over both elements. Our transcription does not indicate this; in Brugmann's transcription, Go. afhapjan 'expire' — Lith. kvėpti 'breathe' would be *qwep-, while Go. hals is represented by *q*olso-. It does not seem expedient to complicate the whole system of transcription for the sake of a very few words.

Through Verner's Law:

Gmc.	seɧwana-	saɧw	sǣɧwum	seȝwana-		-ena-
Go.	saiƕan	saƕ	sēƕum	[saiƕans]
ON	siā	sā	sǫm	[sēnn]
OE	sēon	seah	[sāwon]	sewen
OS	sehan	sah	[sāwun]	gisewan
OHG	sehan	sah	sāhum	[gisehan]	geseuuen (Notker)

Go. siuns, siōn, sȳn 'sight' < *sekwnís > *seȝwnis.

Note: The OE forms are Saxon; Anglian has sēgon ( = sǣgon), gesegen. According to the present standard view (cf . Streitberg, UG 116, 113; Hirt, HU 107 f.), we should expect g in the pret. pl., before Gmc. u, and in the past part. g, if the IE ending was -ono-, w, if it was -eno-. In either case, there are complications. Sievers, Btr. 5. 149, assumed that Gmc. ȝw from either source became w if the accent followed. This would simplify the explanation of the forms with w, but add to the difficulty presented by the forms with g. The view is now accepted by few (e.g. by Trautmann, Germanische Lautgesetze 57).

IE kwetwōres (Sk. catvāras, Gk. τέσσαρες, L. quattuōr, Lith. keturì) 'four', Go. fidwōr; the other Gmc. forms go back to a *kwekwōres, with medial kw either through the influence of the initial, or by analogy with *penkwe 'five' : fiōrer, fēower, OS fiuwar, fior, OHG. feor, fior. The w of these forms must be due either to the initial labial, or to a suffix form -ēres instead of -ōres; Old Norse, however, has the neuter fiogor.

Before consonants:

IE sokwjós (L. socius) 'follower', secg, seggr 'companion, man'

IE sekwti- 'sight', sicht, Eng. sight

IE ghwren- (Gk. φρήν 'diaphragm, mind'), grunr 'suspicion'.

In some instances, we find f ƀ for Gmc. ɧw ȝw. The fact is certain, but the exact conditions are not known; sometimes it seems to be due to assimilation to a preceding w or u. Cf. Noreen, UL 147, Hirt, HU 108 f.:

IE wḷkwos 'wolf', Sk. vrkas, OSL. vehъkъ, Lith. vilkas: Go. wulfs, ulfr, wulf, OHG. wolf; with Verner's Law— IE wḷkwī (Sk. vṛkī), ylgr, wülpe 'she-wolf'

IE kwekwóres 'four' (see above), Go. fidwōr, etc.; the f can also be explained through the influence of Go. fimf < *penkwe.

IE ukwnos 'fire, stove' (Gk. 'ιπνός), Go. auhns, but ofn, OHG. ofan

IE penkwe 'five' (Sk. pañca, Gk. πέντε), Go. fimf, fimm, OE OS fīf, OHG. fimf, finf (f may be assimilation to the initial labial, as inversely L. quinque shows assimilation of the initial to the medial consonant).

Noreen, l.c., quotes fifty-odd instances, but most of them are quite uncertain.

The Intermediate Shift (Step V)

The term 'Intermediate' is not entirely apt. The change did not occur between the Germanic and the High German Shifts, but began in early Germanic times —probably during the first or second century A.D. —, and continued, in High German, contemporaneously with the earlier stages of the High German Shift. In a sense this Shift might be called 'Overlapping' rather than 'Intermediate'.

Back to Table of Contents

24

Back to Table of Contents

a

The Germanic Voiced Spirants,ƀ ð ȝ, tended to become voiced stops, in accordance with the 'Fundamental Principle' (16). This is virtually the opposite of 'Lenition', a process of common occurrence in Celtic, Romance languages, and elsewhere (e.g. L. debēre > Fr. devoir, Span. [deƀer]), since lenition consists in the change of stops to spirants. But the two processes have at least this in common that a spirant is more apt to be retained or developed medially than initially.

The development of ƀ ð ȝ to b d g was completed in Upper German, probably also in East Franconian (which by some scholars, e.g. Behaghel, is classed with Upper German). In the other Germanic dialects, the change appears in complete agreement, on the one hand with the chronology of the formation of the dialect units, and on the other with the place of articulation of these sounds and of their position in the words. That is to say: (1) In Gothic and Norse there is least evidence of these changes, more in Old English and Low German, still more in Middle German, while in Upper German the change is complete. (2) Dentals show the widest, velars the narrowest scope of the change. (3) The change appears sooner, and, therefore, more widely, in initial than in medial or final position.

Back to Table of Contents

b

Germanic, Gothic, Norse:

(1) Gmc. ƀ ð ȝ after nasals and in gemination are stops in all dialects; therefore, we ascribe this change to Primitive Germanic: Go. ON OE OS OHG. lamb, Go. gen. sg. lambis; Go. OE OS bindan, binda, OHG. bintan, Go. pret. band; Go. laggs = [laŋgs], langr, OE OS OHG. lang, Go. gen. sg. masc. laggis. If b d had been spirants, Gothic would have changed them to f þ when final as in hlaiƀis —hlaif, rēdan — rairōþ, but not *lamf *banþ; for the velar, Gothic spelling would not indicate the difference; cf. 31. Instances of Gmc.i>ƀƀ ðð ȝȝ > bb dd gg have been given in 22.

(2) Initial ƀ and ð appear as stops everywhere in Germanic, with the exception only of a few isolated forms in early Runic inscriptions, (ƀarutR = brȳtr 'breaks', Björketorp; ðohtriR = døtr 'daughters', Tune; cf. 18), for which, however, other explanations might be given. Cf. 18 and Noreen, Aisi. Gramm. 143: Go. bairan, bera, OE OS OHG. beran; Go. dauhtar, dōtter, dohtor, OS dohtar, OHG. tohter.

(3) Gmc. ƀ ð became stops in Gothic after l r, probably also after other consonants, and in Norse ð became d after l: Go. haldan, halda, healdan, OS haldan, OHG. haltan; Go. gards, geard, OS gard, OHG. gart, but garþr; Go. halbis, gen. sg., halb, nom. sg. neut., OHG. halb, but halfr, healf, OS half.

The treatment of Gmc. ȝ in Gothic and Norse must still be considered a moot question. It is most frequently assumed that it was a spirant medially, but a stop initially; for final Goth. g, cf. 31. But for Gothic the evidence is extremely slight (see especially Dieter, AD 194); for Norse there is no early evidence at all (cf. Noreen, Aisi. Gr. 143, Asw. Gr. 175), and the testimony of the modern languages is conflicting. It is best in keeping with the general trend to assume the following: In Gothic and early Norse, Gmc. ȝ was a spirant in all positions, except after nasals and in gemination; before back vowels and consonants it was the velar spirant ȝ (g in North German Tage), before front vowels, the palatal spirant (j): ȝ in Go. gards, garðr, j in Go. giban, gefa. This condition remained in Swedish and Norwegian, while in Danish initial ȝ became g. But it must be admitted that there is no direct evidence for either Gothic or Old Norse.

Streitberg, GE (5th and 6th ed.) 63, claims stop pronunciation in all positions, not only for g, but even for b and d at the time of Wulfila; in his opinion, based on Sievers's Schallanalyse, 'fordert die Intonation fur die Zeit Wulfilas unzweideutig die stimmhaften Verschlusslaute'. In view of the alternation between final f þ and medial b d, the argument does not seem convincing.

Back to Table of Contents

c

West Germanic:

(1) Gmc. ð became d in all positions: fæder, OS fadar, OHG. fater; the change to ð in NE father, due to assimilation to the following r (as also in weather) took place around 1500.

(2) Medial and final ƀ remained a spirant (generally voiceless when final) throughout the North: English, Frisian, Saxon, Low and Middle Franconian; it became a stop in the other German dialects: giefan, OFris. ieva (geva, jeva), OS geban, but OHG. geban. The Germanic spirant ƀ had been bilabial, and at least in Gothic and Old Saxon must doubtless still be considered as such. Later it became labio-dental [v] everywhere where it had not become a stop (bilabial in Go. sibun, labio-dental in E. seven). But since the end of the Middle Ages, medial b has been a bilabial spirant in Bavarian, Alsatian, and considerable parts of Middle German, particularly in Rhine Franconian.

(3)The treatment of Gmc. ȝ is not altogether certain. For Old English, it is certain that medial and final ȝ were spirants, and initial ȝ is generally admitted to have been a spirant at least in West Saxon in the earliest period. This is clearly evident from the fact that it alliterates with the IE palatal spirant, as in the first line of Beowulf: gār < Gmc. ȝaiza- and gēar < Gmc. jǣra-, IE jēro-. But later in the Old English period, initial ȝ before consonants and back vowels (where it had velar articulation) became a stop, as in NE grass, good. Before front vowels (except those that came from the modified vowel ö and ü) it remained a spirant with palatal articulation : NE yard, yellow, yield. But this view is hard to reconcile with such forms as NE give, get, guest; since Kluge, Anz. z. Anglia 5. 83, the initial g in these words is generally ascribed to Scandinavian influence, but, as said above, there is no real evidence for stop pronunciation of ȝ- in Old Norse. It is more likely that the stop forms are of Anglian, the spirant forms of Frisio-Saxon origin.

This is indicated primarily by spelling tradition. In West Saxon the spirantic (palatal) pronunciation of initial ȝ is shown by a following i or e, which are 'silent', having the function of diacritical marks: ȝiefan, ȝeaf, ȝēafon — [jevan, jæf, jǣfon]; in early MSS forms with eae = eæ, instead of ea, occur, but apparently there developed a tradition against the writing of three consecutive vowels. The macron of ȝēafon, when used at all, is a mechanical transfer from the diphthong ēa < au. The ē is silent, and the a stands for æ. Anglian spelling, as far as it is not influenced by West Saxon tradition, has gefan, gæf, gǣfon, with initial stop'. (The distinction between two runes for Gmc. ȝ in the Ruthwell Cross inscription has nothing to do with this; it refers to palatal versus velar articulation, not to spirant versus stop.)

Modern English dialects offer conflicting evidence. There seems to be a tendency in favor of the stop in the North and in favor of the spirant in the South, but dialect mixture has greatly interfered.

If Anglian did have initial g for Gmc. ȝ, the shift may date back to continental times. This is suggested by the parallel development in the essentially Anglian territory east and north of the lower Elbe; see below. Apparently, the Saxon and Frisian element of Southern England brought the initial spirant from their continental homes, the Anglians the initial stop, and in both cases that early pronunciation has in a large measure prevailed to this day.

Also in the continental West Germanic dialects, the evidence of spelling is not always sufficient to decide whether Gmc. ȝ remained a spirant or became a stop. Often the somewhat uncertain conditions of modern dialects are our only guide. In general, the distribution is as follows:

Initial ȝ remained a spirant as far south as Middle Franconian, including the northern part of that dialect, but south of that district became a stop, so that its treatment is essentially the same as that of medial ƀ: L. jeven,geben. The spirant is usually palatal (j), but velar ȝ is found in some districts, especially in Low Franconian (Dutch). Medial ȝ remained a spirant (usually voiced, but partly voiceless) everywhere except in Upper German; cf. North German [la:ȝan li:jan]. However, a detached territory in the Northeast has the stop g for Gmc. ȝ both initially and medially: Initially, g is spoken in Schleswig Holstein, Mecklenburg-Schwerin, Pommern; medially, in Mecklenburg-Schwerin (Behaghel, GddS 212).

This coincides with the initial stop in neighboring Danish. It is difficult to say whether there is any common cause for this. Might it have anything to do with the proximity of this territory to the Germanic homeland, that the phonetic trend continued somewhat longer in this district than in the older colonial lands? Perhaps it is not quite without importance that at least a part of this territory —chiefly Holstein —is identical with that Scatenauge Albiae fluvii ripa (= Scadinavia) from which the Langobards emigrated in the second half of the fourth century (Zeuss, Die Deutschen und die Nachbarstamme 471 f.). The consonantism of the Langobards had gone through the High German Shift (cf. Bruckner, Die Sprache der Langobarden 18 ff.; Behaghel, GddS 16), which is entirely in keeping with the theory presented in this book. However, initial þ probably did not become d, as in German, but t, as in Scandinavian.

The High German Shift

Back to Table of Contents

25. Chronology and Spread.

Back to Table of Contents

a

The changes described in the preceding paragraphs belong essentially to the period before the geographical separation of the Germanic ethnic unit. Aside from the incompleteness of the 'Intermediate' Shift, these are 'Germanic' changes, affecting more or less all Germanic dialects. After the formation of the East and West Germanic groups, approximately at the beginning of the Christian Era, all three groups possessed the following consonants, in addition to the nasals, liquids, and semi-vowels:

ƀ ð ȝ; p t k; f þ ɧ; s z.

The consonant changes after the separation show a marked divergence. In Gothic, and especially in Norse, new tendencies set in that led to consonant changes in new directions. In West Germanic, the Germanic Drift, that is, the physiological principle that had dominated the Germanic Consonant Shift, continued for a longer or shorter period, according to the time of emigration; cf. 17, end. The West Germanic expansion was directed mainly to the east and south, surrounding the East-Elbian homeland like the opening of a fan (cf. 6). In the north-western branch, comprising Frisian, Saxon, and Low Franconian, the old trend affected only the voiced spirants in the limited scope described in the preceding section. These dialects, therefore, differ from Gothic and Norse, as far as the Consonant Shift is concerned, only in the change of medial and final ð to d.

If the trend of the Germanic Shift had continued unabated, the result would obviously have been:

Gmc. p t k > Ger. f þ ɧ

Gmc. ƀ ð ȝ > Ger. p t k (through b d g)

But these changes took place only in the territory of relatively late expansion south of the 'Benrather Linie', that is, in the High German dialects, and even there with limitations and modifications. For this reason we speak of a 'High German' or 'Second Consonant Shift' in contrast to the 'First' or 'Germanic Shift'. In general, it followed the same trend, but in certain instances the phonetic direction was deflected, so as to lead to slightly different results. Moreover, in certain cases secondary developments counteracted changes that had actually taken place.

The historical order and, therefore, the geographical spread of the Second Shift follows the Fundamental Principle stated in 17. In the Intermediate Shift, as shown above, the initial or medial (final) position of the consonants was an important factor; it may have been that, temporarily, in the First Shift too, but we have no historical evidence for this. In the Second Shift, particularly in the case of Gmc. p t k, it is of prime importance: the formation of spirants had a wider sway after vowels than initially, after consonants, and in gemination. Since in the Intermediate Shift the development progresses from spirants to stops, but in the Second Shift from stops to spirants, it follows logically that here initial position is unfavorable to the change, while there it was favorable to it.

Back to Table of Contents

b

The maps on the following page (which include the area of later colonization east of the Elbe and Saale) outline the spread of the High German Consonant Shift; the shaded areas indicate the territories in which the respective changes took place.

This distribution of the High German Shift is entirely in keeping with the chronological sequence of the West Germanic expansion outlined in connection with the Fundamental Principle of the Sound Shift (16). The Germanization of northwestern Germany took place soon after the separation of the East and North Germanic groups from the Central stock. Accordingly, we find that Frisian, Saxon, and Low Franconian continued the shift very slightly beyond the stage reached in the former groups. In Central Germany, the Roman limes halted the expansion for several centuries, and the Germans that had reached the middle Rhine before the time of Caesar received strong reinforcements from the central region during that time. This explains the great contrast between Low German and High German. The formation of the Upper German dialects is marked by the final occupation of Southwestern.

Germany by the Alemannii, and of Vindelicia by the Baiuvarii, during the fifth and sixth centuries. The parallelism between the tribal consolidations and the stages of the Second Shift can be indi cated as follows:

Occupation of the North Sea coast, last centuries B.C. —ð > d, ƀ- > b.

Concentration at the limes, A.D. to 400—gradual spread of the shift of p t k, continuation of the 'Intermediate Shift'.

Occupation of Upper Germany, 400 to about 500 — b > p, g > k; k- > kh.

A detailed description of the process would go far beyond the limits of this book.

Back to Table of Contents

26. Instances and Limitations.

The following instances represent the High German Shift in its extreme form, termed 'Streng-Althoch-deutsch' by Jacob Grimm. For the sake of convenience, post-vocalic position is indicated by -p -t -k, while p- t- k- indicate initial position, position after consonants, and gemination.

(1) Voiceless stops after vowels become voiceless spirants: -p -t -k > ff zz hh. These new spirants are not identical with Gmc. f þ ɧ, these had in the course of time become lenes, while the new ones were fortes. This is generally indicated by double spelling, but after long vowels and in final position simple consonants are commonly used: Go. skipis (gen. sg.), skip, slēpan — OHG. skiffes, skif, slāfan (slāffan) Go. itan, ētum, þata — OHG. ezzan, āzum, daz (thaz) Go. mikils, bdka, ik—OHG mihhil, michil, puoh, ih.

ff (f) was probably labio-dental from the very beginning, since this articulation is better adapted to fortis pronunciation than the bilabial one.

zz (z) is used consistently in OHG and MHG MSS (the Isidor MS uses zss, zs); grammatical works generally use z, to distinguish the spirant from the affricate z = ts. Its pronunciation is not known, but it was clearly distinguished from s. Perhaps it was originally a slit spirant, a fortis þ, which changed to a rill spirant, s, towards the end of the MHG time. In most modern High German dialects, the distinction is merely one between fortis and lenis: lassen — lesen; voiced -s- is a borrowing from Low German.

For hh (h), ch is generally used since the middle of the ninth century. Like ȝ, it is assimilated to the preceding vowel, leading to the distinction between the ich- and the ach-sound. Alemannian and a part of Southern Bavarian have only the ach-sound.

2) Voiceless stops initially, after consonants, and in gemination change to the corresponding affricates: p- t- k- > pf ts (z) kh (kch):

• Go. paida, E. stamp, E. dapper —OHG. pfeit, stampfōn, tapfer (but Go. hilpan, OHG. helpfan, since the tenth century helfan)
• Go. twai, hairtō, satjan - WGmc. sattjan —OHG. zwei, herza, sezzan, setan
• Go. kaurn, -wakjan, drigkan —OHG (UG) khorn, wekhan, trinkhan.
• As in the First Shift, this change is prevented by a preceding voiceless spirant: stein, spil, fisk, naht, luft. tr likewise remains unshifted: Go. trudan, OHG. tretan

(3) Voiced stops tend to become voiceless stops; to what extent these were lenes or fortes is entirely uncertain; at least for the OHG period, fortes are probable in Upper German. Medial b became p in Bavarian only (Bav. kepan, Al. keban). b d g > p t k:

• Go. briggan, haban, gaf — OHG (Bav.) princan, hapen, kap
• Go. dags, fadar, bindan — OHG. tac, fater, bintan
• Go. guþ, fugls, dags — OHG (UG) cot, focal, tac (tak).

These changes are carried through to the following extent:

(a) The shift of -p -t -k and t- covers the whole High German territory, but Middle Franconian preserves dat, it, wat, the ending -et, and, partly, the prepositiup; ('Restwörter' cf. 17 d). Also d > t may be termed general High German, but the voiceless stop was (and is) a lenis in Middle and Rhine Franconian, a fortis (originally) in East Franconian and Upper German.

(b) p- > pf belongs essentially to East Franconian and Upper German; in Middle Franconian the change did not take place at all, in Rhine Franconian only after l and r (lpf, rpf, later generally lf, rf) : MFr. RFr. Pund, Palz, MFr. helpen.

(c) k- > kɧ is general Upper German in early OHG times, but now the affricate is found only in some southern dialects of Switzerland; in other Swiss dialects and in the Tyrol the spirant ch is pronounced: chalt.

(d) The shift of b g to p k belongs to Upper German of the earlier period; the details are not of importance for this brief treatment.

B. Other Consonant Changes

Back to Table of Contents

27.

Back to Table of Contents

a

The Germanic Voiceless Spirants, f þ ɧ, had partly become voiced under the conditions of Verner's Law. Originally, they were doubtless fortes, since they had developed in the same way as the later High German spirants, but since prehistoric times they showed a progressive tendency towards a weakening in articulation. This is a natural trend, since the relatively unchecked current of expiration gradually lessens the resistance of the muscles of the lips or tongue, and then, in turn, loses its own force. Their later development in the several Germanic languages varied according to their place of articulation and their position.

The dental þ, followed the tendency of the Consonant Shift consistently. Unless it was preserved as a voiceless spirant (see below), it became ð, later in part d. The labial f, was perhaps originally bilabial, but there is no certain evidence for this. (Assimilations like Go. OHG. fimf < *penkwe prove nothing. The m may have been labio-dental itself; besides, such pronunciations are common even in modern German, which certainly has labio-dental f.) At any rate, its clearness of articulation was enhanced by labio-dental articulation, which we now find in all Germanic languages; in certain positions and in certain languages it became voiced. The velar, for which this book uses the letter ɧ, (modified h, since h is the usual spelling in all Germanic languages) had the least stability: being articulated by the relatively inert velum and back of the tongue, it lacked the articulatory intensity that would have favored a further shift; under Verner's Law it had become ȝ at a time when it was still pronounced with considerable friction although a lenis. Where it remained voiceless, its physiological character led to further relaxation, resulting in its reduction to actual h, the glottal (instead of velar) spirant, and sometimes in complete disappearance. The following is in part a summing-up of statements occurring in former sections.

Back to Table of Contents

b

Gmc. þ was preserved in Gothic in all positions: þu, wairþan, brōþar, warþ. In Old Norse and Old English it remained þ initially, but became ð in voiced surroundings: þing—brōðor, þing—brōðor; spelling, however, does not indicate this, since the letters þ and ð are used interchangeably in these two languages. In Old Saxon and Old High German we find þ (mostly spelled th) in the earliest records in all positions, but gradually this changed to ð and soon to d, first medially and in habitually unaccented words, e.g., in pronouns and in the definite article. In Bavarian, the process began as early as the eighth century. It gradually spread north, reaching the Low German territory during the late Middle Ages. In Middle English, perhaps even in Old English, þ- in pronominal forms and similar words of little semantic function became ð, and an analogous process took place in Danish, Swedish, and Norwegian where, however, stops resulted: Sw. ting, tānka, but du, De. English and Icelandic preserved þ initially, English also in strictly final1 position (cf. cloth—clothe; in the latter word th was medial in Middle English).

Back to Table of Contents

c

Gmc. f was also preserved in Gothic in all positions: fadar, afar, uf. Like þ, it became voiced in voiced surroundings in Norse and English, thus coinciding with IE bh and `p. In Low German (Old Saxon), too, it became voiced. The usual spelling was ƀ, but we also find u: wulƀos, fiui. In High German it generally remained voiceless, but has been a lenis since OHG times, as its frequent rendering by u v in medial position indicates: nevo, nefor, hevig.

Back to Table of Contents

d

Gmc. ɧ was clearly the ach-/ich-sound, as is indicated by Latin spellings like Cherusci, Chatti, Gk. Χέρουσκοι, Χάττοι. In all Germanic languages it is preserved initially as h, but in other positions it must be considered a velar (or palatal) spirant wherever preserved at all : h in Go. haban, hafa, etc.; [χ] (or [ç]) in Go. nahts, raihts, taih. However, final h in Gothic, while still a velar spirant, must have been decidedly a lenis, while final ɧ, which became unvoiced like final ð and ƀ, was a fortis: dag, dags probably had a velar spirant with stronger friction than taih, comparable to the difference in the articulation of ch in Bach and nach. In Norse, h disappeared everywhere except initially, in Old English it disappeared between voiced sounds: Go. himins, ƕa: himinn, huat, heofon, hwæt; Go. teihan, saiƕan, slahan, aƕa, filhan: tiā, siā, slā, ā, fela, tēon, sēon, slēan, ēa, feolan. In both Low and High German final and intervocalic h is consistently preserved in the older sources, but gradually disappeared, first in Bavarian; cf. sehen, sah = [ze:ən, za:]. Before consonants it remained a velar or palatal spirant: Nacht, Sicht. —Initially it disappeared before l r n w in Norse as well as in West Germanic; only English, at least American English, has preserved hw: Go. hlahjan, hlǣja, hliehhan, OHG. (h)lahhan— Sw. le, E. laugh, Ger. lachen; Go. hneiwan, hniga, OE OS OHG. hnīgan; neigen; Go. hrains, hreinn, OS hrēni, OHG. (h)reini: rein; Go. ƕa, hvat, hwæt, OS hwat, OHG. hwaz: Sw. hvad = [vad], E. what - Brit. [wɔt], L. wat, was.

Back to Table of Contents

28. Rhotacism

IE s was preserved intact in Germanic where it had not changed to z according to Verner's Law. This z remained in Gothic, but in North and West Germanic it was intensified to r, as in Latin (genera < *genezā). This is called Rhotacism from rho, the Greek name of the letter r. In earliest Norse this r is distinct from old r: It is expressed by a different rune (transcribed by ʀ in our grammars) and causes i-mutation (Go. kas 'vessel', ker). Early Norse loan words in Finnic still show s (rengas, kuningas). In later Norse and in West Germanic there is no distinction between old and new r. —For s in final syllables, see 49 d.

Instances:

• Go. maiza: metre, māra, OS OHG. mēro
• Go. kiusan, kaus, analogical kusum, kusans: kiōsa, kaus, kørom, kørenn (r-mutation), cēosan, cēas, curon, coren, OS kiosan, kōs, kurun, gikoran, OHG. kiosan, kōs, kurum, gikoran
• Go. hausjan —heyra, hīeran, OS hōrian, OHG. hōr(r)en
• Go. huzd —OE OS hord, OHG. hort.

Note: Go. huzd goes back to IE kudh-tó- > kud-dhó-, kudˢdhó-, through 'Bartholomā's Law'; it is related to Gk. κύσθος 'concealed place, female organ', L. custos 'guardian': similarly >*ghədh-tó-, L. hasta, Go. gazds, OS gard, OHG. gart 'spike'. IE z goes back to s before a voiced stop. In such cases, through regular consonant shift, Gmc. sp st sk result, as in *ni-zdo-, L. nīdus, OHG. nest, *o-zdo-, Go. Go. zm, OHG. ast.

For Gmc. zm, see below.

Back to Table of Contents

29. Consonant Groups