The University of Texas at Austin; College of Liberal Arts
Hans C. Boas, Director :: PCL 5.556, 1 University Station S5490 :: Austin, TX 78712 :: 512-471-4566
LRC Links: Home | About | Books Online | EIEOL | IE Doc. Center | IE Lexicon | IE Maps | IE Texts | Pub. Indices | SiteMap

A Reader in Nineteenth Century
Historical Indo-European Linguistics

Winfred P. Lehmann


CHAPTER NINE

C. LOTTNER

EXCEPTIONS TO THE FIRST SOUND SHIFT

Originally "Ausnahmen der ersten Lautverschiebung,"
from Zeitschrift für vergleichende Sprachforschung auf dem Gebiete
des Deutschen, Griechischen und Lateinischen
11.3 (1862), 161-205
Editor's Introduction
Lottner's cataloging article has never been given the acclaim of Grassmann's or Verner's, which it prepared for. Yet Lottner carefully screened the evidence and listed the three large sets of "real" exceptions, after eliminating the apparent ones traceable to "false comparisons", onomatopoeic words and borrowings. In his treatment of the apparent exceptions Lottner reflects the status of historical linguistics shortly after the middle of the nineteenth century; for that reason, and because the first section of his article is his permanent contribution to clarification of the first sound shift, this section is presented rather fully here. The "real exceptions", which Lottner discusses in the remaining sections of his article, may be found in the articles of Grassmann and Verner. After Lottner's publication there was no further confusion about the apparent exceptions. His clarification of these, and his thorough descriptive presentation of the evidence remain the essence of his achievement.
This is not to say that Lottner's article is thoroughly praiseworthy. The tone of it is occasionally intemperate; at the conclusion he added a short paragraph apologizing to Curtius for an unwise phrase in an earlier article. He also misled Grassmann about the "third set of exceptions," which were clarified by Verner, by introducing the glib notion of a Wahlverwandtschaft -- elective affinity -- between the resonants and the mediae. Yet with its faults, the article is an example of the interplay of descriptive and historical linguistics: Lottner's descriptions simplified the explanations of Grassmann and Verner. The article also illustrates that linguistics did not advance by a series of leaps; rather, careful scientific attention to the data led in time to its mastery.
C. Lottner submitted the article from London. To my knowledge he made no further contributions to historical linguistics. From their acknowledgements we conclude that the article was important for both Grassmann and Verner.

Prefatory Note. In general, only those Germanic languages are considered whose consonants have really experienced only the first sound-shift. On the other hand, High German has only then been taken into account when it illuminates the Germanic original form which in Old Norse, Anglo-Saxon, etc., was distorted according to specific sound laws. Words which are only High German are not treated for the time being.

The exceptions to the first sound-shift fall into apparent and real exceptions.

Those words may be treated as the first class among the apparent exceptions with which the entire assumption of a violation of Grimm's law originates simply from the fact that one has compared things that should have remained uncompared. It is of course impossible to list all cases of such erroneous comparison since error and caprice lack definite demarcations. I am therefore content to observe a few examples of this kind where erroneous comparison has enjoyed a certain popularity, whether by virtue of its apparent plausibility or on account of its author's reputation. Of this sort are Gothic: kara 'care' which has nothing to do with Lat. cura, since the latter arose from coira, as is proved by OLat. coerare to which probably belongs Gk -koíranos (cf. Landpfleger, 'prefect'). The Germanic word means primarily 'fear, anxiety, lament' (therefore OHG charôn lamentari, our NHG char-freitag), and it thus goes with the old root GAR 'be heavy', which is present in Skt guru 'heavy' (from garu; comparative garîyas) = Lat. gravis, Gk barús, and from which Goth. kaurs 'difficult' is to be derived. Yet in Germanic the meaning 'to be heavy about something', 'to take care about some thing', must have developed in this root at an early time. For that is the only explanation for ON kaer-r 'dear', the original form of which is KÂRIA and which will have to be understood as 'something worth being cared for' with the same suffix ja which is present in Goth. andanêmja- 'acceptable, pleasant', in Goth. unquêþja- 'impossible to utter, inexpressible', in Lat. exim-ius = eximendus and in the many corresponding formations of Sanskrit (such as bhâr-ya 'ferendus'). This Scandinavian kaerr therefore does not in the least have anything to do with Lat. cârus which, despite the long vowel, is probably to be compared with OIr. caru,1 carimm 'I love'. -- The comparison of Goth. leik with Skt dêha "body" -- the transition from d to l is as indemonstrable for Gothic as the k vs. h is objectionable -- was able to gain a dubious reputation only on account of the great reputation of the founder of comparative grammar. The same is true of his relation of Goth. -leiks 'like' (isolated galeiks = OE lîc, gelîc ON lîkr, glîkr) with Skt. -dṛça, because the regularly corresponding Lith. lygus is much more plausible and because the root dṛç, i.e. dark (Gk dérkō) 'see', from which that Sanskrit word originates, is also regularly represented in our AS torht 'light' = OHG zor(a)ht. Goth. natjan 'moisten' would have hardly been compared with Gk -notîa if one had not dismissed all too easily the connection of the latter with Gk nótos. To compare Goth. raþs 'easy' with Gk hrą́dios, as the otherwise admirable Gabelentz and Loebe do, is a monstrosity, because the latter is contracted, as Homer's Gk hrēídios shows and furthermore, it began with digamma in Lesbian. Our Germanic word belongs to the root RAT, in Skt ratha 'wagon', Lat. rota, OHG rad, Gallic riton (to be deduced from petor-ritum) and therefore means something that 'begins'. The root itself may have been developed from AR (Skt "go" Gk or- etc.). -- Goth. auhns "stove" has often been compared with Skt açna 'fire', but since Aufrecht has related it nicely to Skt açna 'stone' the former comparison must be considered obsolete. Bidjan 'ask for' scarcely has anything to do with Lat. petere, because the basic meaning of the latter is 'to fly at something,' (= Skt pat 'fly, fall'), whereas Goth. badi 'bed' seems to indicate as the original meaning of our word: 'sternere, se prosternere'. Goth. kalds 'cold', or rather its stem-verb ON kala 'be cold' is quite regular with regard to Lat. gelu, gelidus, Skt gala 'cold, coldness, water'. For that reason OSlav. chladŭ 'cold' is either not related at all, or it is borrowed. Lith. szaltas 'cold', szala (3rd p. sg.) 'be cold, become cold' is however quite a different word, which, with respect to the root, Zend çareta 'cold' and Skt çiçira 'cold' resemble. But the two latter words point back to an original root KAR, KAL; whether this root is to be treated as identical with GAL, I do not know. At any rate, the difference occurred already before the language separation and we Germanic people must be absolved from the reproach of an irregular sound-shift ... (166). With these and similar comparisons we lose ourselves completely in a territory where any words are picked up according to their sound-similarity and, with an indubitably blessed, but highly unscientific naïveté, are assumed to be related.

Apparent exceptions to the sound-shift can also stem from the fact that the words under comparison are connected psychologically, but not historically, i.e. that they are imitations of sounds, or they fall into that category which Buschmann designates by the name 'sound of nature'. After the thorough discussion by that scholar it may be considered certain that the consonants T, P, or in other words the syllables ap, pa, at and ta even in non-related languages form the basic elements in the names of the father (more infrequently in that of the mother) because of the identical physiological make-up of the speech organs and because of identical psychological impulses. Although the Indo-European languages possess words from very early times to express these relationships, it must nevertheless be admitted that some of the many parental names have been formed anew only after the time of the language separation. If therefore Goth. atta 'father' corresponds too well to Gk átta OSlav. otĭcĭ (OBohem. ot), then neither relationship nor borrowing is to be assumed here, but the Gothic word has simply sprung from the everflowing fountain of nature sounds. Furthermore, it can be seen from the treatise of the aforementioned scholar that the nasals N and M play the same role in the name of the mother (again, here too, less frequently in the name of the father) as P and T did in the case of the father. It must also be noted that all of these readily pronounceable elements, notably m(a) and p(a), serve at the same time as children's words for 'food' and 'nourishment' and therefore frequently as expressions for the mother's breast. Such words are our pappen, pappe, Engl. pap 'female breast', Lat. papilla, Lat. mamma, mamilla 'mother's breast' but mamma, mammula also meaning 'mother, grandmother'. For our purposes, we must be concerned with still another word for 'mother's breast' (beside the abovementioned apparently very irregular pappen) which staggers just as unsteadily and wildly through the various consonant stages, namely OE titte, NE teat, to which fits NHG Zitze, but also OHG tutta, MHG and NHG tutte, Gk títthē (compare in Dieffenbach a number of related words under Goth. daddjan 'to nurse'). Yet this latter is perhaps no longer the immediate creation from a sound of nature, but rather the reduplication of an old root , which corresponds to Skt dhê 'to nurse' (from which, dhênu 'cow') and also to dhâ (in dhâ-trî 'wet nurse'), compare Gk thêsai, thêsai, tithene, thelus, Lat. femina (?), fellare, Umbr. felio- 'sucking', OIr. dinu 'agna', stem dîna(n)t, Lat. filius (cf. Gk thḗlē, thēlázō; and Lett. dehls 'child'), OSlav. doiti 'to nurse', dê-te 'child.' OHG tila = Gk thḗlē; (cf. OHG taan = Gk thȇsai) is certainly related and must be distinguished from those new formations made of sounds of nature. In all of the latter words the exact sound correspondence stands for a genuine historical relationship. But that the above-noted chaos of sounds in otherwise closely connected words of identical meaning can find a sufficient explanation through the assumption of simple psychological relationship, is most clearly demonstrated by similar sounding words from non-Indo-Germanic languages by Dieffenbach, such as Basque thilia, dithia, titia, Hung. tsets, Eston. tis, all of which mean 'female breast'. To these may be added several onomatopoeic words in which an apparently irregular sound-shift takes place, e.g. ON klaka 'queri (de avibus)' = klökkva 'lament', NE clank, OHG klingan as against Lat. clango Gk klaggḗ, klázō, to which may be joined Goth. hlahjan 'laugh', which originates from a similar formative impulse. In addition, compare NE clatter, clap, NHG klirren, klopfen, klappern. OE cancettan 'laugh' and also ceahhettan do not go with Gk kagkházō, cachinnari. Much more of this kind could be cited, if one wished to include living folk dialects. A peculiar example of this merely psychological relationship is NE lick, OE liccjan = NHG lecken. It is quite impossible to place this with Skt lih, Gk leíkhō, Lat. lingo, OIr. ligum etc., since this root is present with regular shift in Goth. laigon. In addition, there is in Lithuanian and in Slavic a Lith. lak-ti, OSlav. lokati and beside it Lith. laiźyti (OSlav. lizati) which correspond regularly to the Greek-Sanskrit root. If one notes that similar sounds serve also in non-Indo-European languages as a designation for the act of licking (e.g. Hebr. kkl, Finn. lakkia), and that in Germanic itself there is a third form which deviates completely, but is still related in sound: ON sleikja -- Gk láptō, Lat. lambo, labrum, NE lap 'lick', lip = OHG laffan, lefsa, NHG lefze, Lippe (the latter actually Low German) are admittedly more remote but must not be ignored either -- then all this leads one to conclude that only Goth. laigôn is historically connected with Skt lih, whereas the other forms with a seemingly irregular k are new root formations, or if one prefers it this way, that the old root lih is in fact present in them, but is disturbed in its regular sound shift through the influence of sound imitation and thus has been distorted to LIK, LAK (the latter with a quite preposterous a from i). This easy mode of explanation would have arisen a long time ago if the strange hypothesis had not spread in comparative linguistics that root formation cannot possibly have occurred after the Indo-European peoples had separated. I cannot grasp why such a purely exterior event like the disintegration of peoples should have suddenly cut off the capability to create language. It is paramount to denying the writer of this article the capability to create new compounds and derivatives in his mother tongue just because he emigrated to England. I should like to go on record that I have strongly protested against this mechanical, as well as lifeless and unhistoric, interpretation of the aforesaid conception which assumes the existence of a special "root-forming" era.

A similar situation is found in ON gaukr 'cuckoo', also MHG gouch, which just does not fit NHG Kuckuck, NE cuckoo, or Lat. cuculus, Gk kókkuks; all these irregularities can be explained simply by the fact that imitations of the animal's voice have been made over and over again. Only the same assumption will explain the strangely corresponding and deviating names of the crow, partially also those of the raven; compare Lat. corvus, Swed. korp; OHG hraban, ON hrafn, can just barely be related to the Latin word (although our b does not correspond to Lat. v either). Further, there is OE crâve, ON krâka, Gk korṓnē, Lat. cornix and further the verbs NHG krächzen, Gk krázō, Lat. crocito and finally, though applied to the 'rooster', OE crâvan, our krähen. All of these are but bound together by a psychological tie.

A great number of apparent violations of the sound shift must be attributed to borrowings within or from the Germanic languages. Most of their foreign words are clearly of Greek, Latin and recently also of French origin. It is not possible to list all of them; I am content to cite those which appear in Gothic. They are from Latin: akeit 'vinegar', annó 'annual pay', arka, asilus, aurkeis (urceus), faskja, kaisar, kapillon 'shear', karkara, katils (catinus), kavstjo (cautio), kubîtus 'resting place at the table', which is related to anakumbjan 'to lie at the table', laiktjo, lukarn, maimbrana 'parchment, membrane', militon, papa, paurpaura, praitoria, pund, spaikulator, unkja and probably also vein. From Greek they are: aggilus, arkaggilus, aikklesjo, aipiskaupus, aipistaule, aivaggeli, aivlaugia (Gk eulogía), aivxaristia, probably also alev 'oil', anaþaima, apaustulus, azyme, balsan, barbarus, daimonareis, diabaulus, diakaunus, hairaisis, jota (Gk iȏta), nardus, paintekuste, parakletus, paraskaive, paska, pistikeins (Gk pistikós), praizbytarei, praufetus, psalma, saban (Gk sábanon), sabbato, sakkus, satanas, sikls, sinap, skaurpjo, smyrn, spyreida (Gk spurís), synagoge; further ulbandus 'camel', with a changed meaning and strangely enough with regular sound shift from Gk eléphant. - - - I pass over the borrowings of the younger dialects from the two Classical languages, as well as over the few cases where the borrowings take place into them. I also omit the many Germanic words which have gone into the Romance languages, since all of these have been exhaustively treated by Diez. Yet on the whole I must say that it is completely wrong to assume complete isolation of the Germanic peoples of heathen times. Just one word like ulbandus should speak against this, since it must have been in use for a long time in order to be Germanicized in such a form. But we also know that the Germanic heathens took over the seven day week from the Romans -- where else would the pagan names of the weekdays come from? We know from Kirchhoff that the runes originate from the Roman uncials. There are Roman coins in Old Scandinavian graves from the time of Tiberius to Marcus Aurelius; and in the oldest Eddic songs we already have evidence of Roman words: tafla 'tabula', tefla 'to play a game at a board' are to be found in the Vǫlusspâ; ketill 'kettle' appears in the Hymiskvîða and has deeply penetrated the northern heathendom, as is shown by the names Ás-ketill, þôr-ketill 'god's kettle, Thor's kettle'. Even the Roman state affairs and Roman religion have influenced us early; Kemble detected the Old English name Säter not only in Säteres däg (Saturday), but also in the names of places, compare Säteres byrig (Saturn's castle). Saturn must therefore have enjoyed a certain popularity, if not in fact veneration, and Caesar, as is well-known, rose to the honor of being moved into the Old English genealogies as the son of Vôden. Under these circumstances it will be advisable in the future not to be too eager to disregard the possibility of borrowing even for very old Germanic words which have apparently escaped the sound-shift.

It is of course to be expected that there was an early exchange of words between the Germanic peoples and their neighbors to the east and west. As regards the Celts, our connections with them were obviously quite lively in the pagan era. This is proved by the fact that the king of the truly Germanic Marcomannen, for instance, had the decidedly Celtic name Maroboduus. In addition, there are the great number of Gallic names ending in -rîx, -rîg-is and -mârus to which the many Old Germanic names ending in -ricus, -merus correspond very precisely. Compare, for example, Gall. Segomârus with OGmc Sigimerus, a condition which can only be explained through the factor of mutual influence. But since Celtic not only resembles Germanic in lowering the old aspirates to mediae, but also shows beginnings of a sound shift of the mediae to tenues, for these reasons it is in most cases extremely difficult to determine which one of the two languages has borrowed from the other, and often whether borrowing or original relationship exists. Besides, most words of this kind will be discussed below, because the irregularity present in them can best be explained by a comparison with words of the other originally related languages. Therefore I mention here only Gall. bracca 'trousers', from which is borrowed ON brôk, OE brôc, NE breech. Furthermore, the strange Goth. kelikn Gk púrgos which is no doubt identical with celicnon, a word that recently appeared on a Gallic inscription. The fact that the word stands alone in Gothic, as well as its strange suffix and the completely un-Germanic appearance and sound, speaks for borrowing from Celtic.

Since earliest times Germanic has many specific agreements with Slavic and Lettish; much of this is admittedly due to an original kinship, but very early borrowing is not rare either. The Slavic-Lettish languages resemble the Celtic in their consistent lowering of the old aspirates to mediae, and they also coincide otherwise (see below) in an anticipating manner with our sound shift. These circumstances make it also extremely difficult to decide whether borrowing took place and if so, from where ... (174) Although I deal here only with those Slavic-Germanic loanwords which, not being recognized as such, seem to constitute exceptions to the sound shift, I cannot refrain on this occasion from drawing attention to the strange fact that we have indeed borrowed from Slavic a great many expressions dealing with commerce, comforts and amusements (buying, names of coins, debt, market and translator, the beaker and the dance), but the Slavs on the other hand have taken from us the word for ruler (likewise the Finns their kuningas). The historic position of the two peoples corresponds fully to this phenomenon: throughout the Middle Ages the Germans treated the Slavs as servants; therefore our Sklave, earlier in the fifteenth century without k, Slave, NE slave, Swed. slaf. The Scandinavians established for the Slavs their Russian empire. It remains to be seen whether this political position of the two peoples is going to change in the future.

Borrowings have also taken place from one Germanic language into another, i.e. 1) from Low German into High German, 2) from Scandinavian into English, 3) from Low German into New Scandinavian, and 4) from New High German into Low German and New Scandinavian. Yet, in 2) and 3) the borrowing and receiving dialects are at the same stage of the sound shift. This is admittedly not the case with the borrowings of the first and fourth class, and some of these indeed give the appearance of a disturbance of the sound shift (e.g. Swed. dyster, borrowed from NHG düster, does not correspond to OE þŷster, Lith. tamsus, Swed. an-dakt = NHG Andacht, not to Lat. tongere). But a more detailed treatment of these mutual borrowings of the Germanic peoples must be reserved for another time.

A great number of apparent irregularities in the sound-shift also come into being through the irregularities of the related languages. To start from the beginning, it is known that Sanskrit has an entire class of aspirates, namely the voiceless aspirates, which only the Iranian languages share with it. With every example one must first determine whether -- which seems mostly the case -- these voiceless aspirates come from an original tenuis or -- which admittedly occurs in some cases -- originated from an aspirated media. For this Greek will generally guide us safely. Thus everything is in good order for the Sanskrit root path, from which comes panthan 'path', equal to the Germanic root fanþ, which will have to be treated later because of other irregularities, because Gk pátos, patéō witnesses the existence of the original tenuis. But Skt nakha 'fingernail' is likewise quite regularly represented by OE nägel, since Gk ónukh- shows here the age of the aspirate. There are, however, also other cases where the Germanic only seems to be irregular with regard to Sanskrit, for example Goth. hairto 'heart' vs. Skt hṛd, hṛdaya. Since all European languages have here either k or its regular substitute (Gk kardía, Lat. cord-, 0Ir. cride, OSlav. srĭdĭce, Lith. szirdis) there really is no alternative to admitting an irregularity in Sanskrit (and Zend).

Another series of exceptions can be explained when one looks at the history of the sound shift law. It seems to me that Curtius has demonstrated that the sound shift began with the lowering of the aspirates to mediae, which in turn led to the raising of the genuine mediae into tenues, and further the old tenues into aspirates. I should like to point out in anticipation of the conclusion of this examination that there will then be another reason in support of his view. But it would be false to believe that this lowering of the old aspirates owed its origin to a sudden caprice of the Germanic people, for it is very deeply ingrained in our whole language development. Zend already shows b for the old bh and frequently also d for the old dh. In OPers. bh, dh and gh always change to mediae. On European soil this same degeneration is very old in some words; and earlier I have pointed to the conformity of all European lan guages in this respect, which is attested several times as one of the reasons which compel us to assume that there was a lasting association of the Europeans after their separation from the Asiatics. Goth. ik, mikils, -k (suffix in mi-k, þu-k etc.), kinnus all correspond to Gk egṓ, megalo-, ge, génus, to Lat. ego, mag-is, gena (the Celtic, Slavic and Lettish languages prove nothing in this matter since all old aspirates become mediae in them), while Sanskrit offers aham, mahat, ha (Ved. gha) and hanu. The situation is also similar with Skt vṛh 'to grow' 2) 'to work' -- Zend verez- but Gk werg-, Goth. vaurkjan; also with Skt vṛdh 'to grow', but Gk (b)ríza from wridja, hrádeks, hrádamnos, with which goes Goth. vaurts, OE wyrt, and further ON rôt, NE root, while OE rôd 'pertica', NE rood, rod, OHG ruota agree with the Sanskrit sound level. From Gk megalo- it can be affirmed against all doubt that the root had originally gh, for this root is in Sanskrit manh 'crescere, augere' and has maintained Gk kh in mȇkhos, mēkhanḗ, with which in turn Goth. magan agrees. It may also be assumed as proved that Gk ge arose out of older Gk khe only on European soil, in the event that Gk -khi (in hȇkhi), which can not possibly be considered a case suffix, is related to it as Skt hi is to ha (all these little words are enclitic). With all the other examples there remains the slight possibility that originally a g stood here and that the h of Sanskrit is a special irregularity.

Aside from these anticipations of the sound-shift, which pervade all European languages, each language has also specific preludes to the sound shift, as well as some peculiar irregularities of its own. [Lottner goes on to list these pp. 177-182].

(182) Finally, apparent irregularities come into being through dialect peculiarities of the individual Germanic languages, through which the system of their mutes is more or less altered. To this belongs above all the second sound shift of the High German which has affected some of its individual dialects more, others less, but none completely. Therefore determination of the original Germanic form meets with considerable difficulty where words have only been preserved in High German. The second sound shift, as is well-known, has penetrated the dental group most thoroughly, and of these the z (= Goth., OE, ON t). But there is an exception here too, to which insufficient attention has been paid: the groups tr, tl always remained unshifted. This not only explains our treu, OHG triuwi vs. NE true, ON trûa, Goth. trauan etc., but also cases like OHG bittar vs. OE biter, ON bitr, Goth. baitrs, where a vowel was inserted at later times. --- A great number of apparent exceptions to the first sound shift originate in Old Norse and in Old English through the almost consistent change of medial b to f (changed further in English, Low German and Danish to v, in Swedish to fv), whereas Old Saxon has preserved the intermediate grade bh. Thus, OE leof 'dear', ON liufr, OE lufjan 'love' appear to be on the same grade with Skt lubh 'cupere'. Likewise, ON stafr 'staff' OE stäf seem to fit exactly Skt stambh 'fulcire', but HG lieb, Stab, Goth. liubs, stabs show that everything is in order. In most cases of Old Norse medial d becomes þ, ð, and in Anglo-Saxon this may happen under certain conditions which, of course, gives rise to new apparent exceptions. 2) In New Norse (as well as in Low German) th is lost and is replaced by t or d. It is replaced by the former when the English form begins with a hard th, and by the latter when the English pronounce a soft th (thus Swed. du, de, den, dem, desse, än-då = Engl. thou, the, they, them, these, though). For this reason Swed. du 'you', tänka 'think' are apparently irregular when compared with Lat. tu, tongere, but in Old Norse we quite regularly have þu, þenkja. Finally, medial tenues between vowels, and final tenues after a vowel, change in Danish to mediae (rarely also in Swedish). Thus, Dan. bog 'book', vide 'know', aede 'eat' correspond, for example, to Gk phȇgos, wid, ed, but compare Swed. bok, veta, äta, ON bôk, vita, eta. All these special irregularities are, of course, to be revoked and the state of Primitive Germanic sounds to be restored before a comparison is possible, also when the dialect form is apparently more regular than the primitive form arrived at by a comparison of the other Germanic languages. It is, for example, uncritical to cite OE seofon 'seven' for the correct sound shift with regard to septem, because Goth. sibun as well as the High German form demonstrate clearly that here Primitive Germanic had a b.

The Gothic sound conditions of the mutes are identical with Primitive Germanic in by far the most cases, but not always, just as little as the grammar of this dialect does not always have the oldest forms. Some examples of irregular sound shift in which Gothic is corrected by other dialects will be given later. Here I cite but the two peculiar examples in which the Gothic alone has maintained an unshifted d, namely du 'to' = OSlav. do 'to', da Gk hína, OIr. do and, according to Stokes, probably also Lat. -du (in in-du = NE in-to), whereas OE and OHG zuo, za, zi have been shifted; and Goth. dis- = Lat. dis, but OHG zir- which presupposes an earlier regular tis.

After elimination of the apparent exceptions we can now proceed to consideration of the real exceptions.

I. Irregularities of the original tenues:
a) The tenues remain regularly (184-187).
[First Lottner cites the groups sp, st, sk, which he says are well-known; then ht and ft. Apart from these he finds little material.]
b) The old tenuis appears as media (187-197).
[Here he finds the greatest number of exceptions, especially in medial position. He cites, though not coherently, the well-known words, such as "Goth. sibun 'seven' = Skt saptan", "Goth. taihun 'ten' beside -tigus (-zig) Lat. decem". Nor can he account for subsequent changes, such as the devoicing in Goth. hlaifs beside the hlaib- of the oblique cases. Accordingly he is nowhere near a solution. In his final comment on medial mediae instead of expected aspiratae he points to the "elective affinity" between liquids and mediae and to the interchange between "aspiratae" and mediae in the same word, citing for example,

OE veorðan vearð vurdon  
OHG ziohan zôh zugum zogan
NHG leiden     gelitten

He concludes the section with the sentence: "Although to be sure examples occur, in which the older aspirates can no longer be demonstrated, it may not be too daring to presuppose in general the transition of the tenues through aspirates to mediae as a former intermediate stage." It remained to Verner to correct the phonetic statements and associate the phenomenon with the Indo-European accent.]

II. Irregularities of the original mediae (197-202):
[This section was very useful for Grassmann. Although some of Lottner's equations had to be discarded, others are:

Goth. grēdus "hunger, greed" - Skt. gardth "be greedy"
Goth. bindan - Skt. bandh
OHG bodam "floor" - Skt. brudhna, Gk puthmḗn

The ON botn and OE botm perplexed Lottner, as did the Greek tenuis. In this section too he associated the irregularities with the liquids, but noted that there were many fewer than for tenues.]

III. Irregularities of the original aspirates (202-3):
[In this short section Lottner's examples are largely erroneous comparisons, which he himself calls uncertain.
After a brief summary Lottner concludes with a statement on the relative chronology of the sound shift.] (204) "It has been disputed where exactly the sound shift began. Grimm finds boldness in the shift of the mediae to the tenues, and accordingly seems to view this as the starting point; I heard Bopp present the entire shift as a weakening of sound, completely opposite to Grimm, and he put the change of the tenues to aspirates at the beginning. The third assumption, that the aspirates became mediae first of all, Curtius capably demonstrated as the most probable by comparing the originally related languages. Through the observation that the aspirates were shifted with greatest regularity, with somewhat less regularity the mediae and least of all the tenues, this view of Curtius gains new support."

London, 10 November, 1860.

Notes
  1. The older form caru Stokes attests in Félire Oingosso Céli dé -- "á ísu notcaru" -- "O Jesus, I love you." [return to text]
  2. It is peculiar that in English the Old English medial dental mediae often appear as th; thus in together, weather, father, mother, all very common words; OE ät-gädere, vedr, fäder, môdor (the three last as exceptions to the sound shift; see below). Is this Scandinavian influence?