If non-specialists know anything about historical linguistics, it is Grimm's law. The history of views on the consonant shift is virtually a history of linguistic theory until 1875; subsequently it is equivalent to the theory of historical linguistics, from the neogrammarian position (that each consonant should be treated individually) to that propounded today (that the entire shift be viewed as a whole). Yet our first reaction on looking at Grimm's celebrated statement may be surprise, He is groping through the consonants; his remarks on the liquids show great uncertainty. The vowels are quite obscure for him. And combined with the treatment are peripheral remarks about speech -- comments on the purpose of vowels -- which we would not welcome in any treatise today. Yet this formulation of the Germanic consonant shift has indeed had "momentous consequences for the history of language." Subsequent discussion is voluminous; few Germanists, Indo-Europeanists or even general linguists have failed to comment.
It was Grimm's conception of the shift as a unit which made such an impact on linguistics. Although his formulation lacks the neatness we might expect, he did account consistently for a large segment of the set of Indo-European and Germanic consonants. His consistent account was so overwhelming that no one doubted its validity. The items unaccounted for were considered exceptions and were made the object of research for the next half century.
Yet we may be even more surprised that there is no mention of a law. Grimm has given nine rules, relating the consonants of Germanic with those of Greek and Latin, less commonly with Sanskrit and other Indo-European languages. Instead of rule, Regel might equally well be translated correspondence. If we did use this translation, Grimm's formulation might be quite contemporary. He stated the evidence fully, including exceptions, posited the relevant correspondences, and indicated their relationships to one another. The statement is a classic example of the formulation of a problem in linguistics, and of its solution within the sphere of language.
Possibly an attempt at explanation is implicit, though even this is not certain. By viewing the shift as non-organic, Grimm apparently saw in it a deviation from the organism developed by the speakers of Indo-European languages. Just as inflection, in contrast with agglutination, seemed appropriate to the Indo-European languages, so did the system of obstruents in Greek and Latin. But we see none of the fanciful attempts at explanation which our handbooks summarize -- a shift due to change of geography, or climate and so on -- nor even the more sober attempts which seem appropriate to us, such as a general shift in keeping with one type of phonetic reshaping or with the modification of distinctive features. Grimm's concentration on taxonomy spared him all such ventures.
He was also fortunate in his ignorance of phonetics, which permitted him to class together consonants which were quite different in articulation, and to produce a statement which passes beyond details to the system. Examination of details, as by Raumer, Grassmann, Verner, clarified exceptions, but it also for a time undermined the unity which Grimm saw in the shift, and which a structural approach has restored.
The translation has been deliberately kept stark to illustrate Grimm's pioneering. We might well interpret "guttural" to mean velar, as it often does even among linguists who should be better informed; but that it meant "throat-sound" to Grimm is clear from his German equivalent "Kehllaut". Though we may pride ourselves on superior terminology, our estimate of the capabilities of Grimm's contemporaries is not diminished by the ease with which they were to identify examples as Greek or Latin, with no special indication.
As we update Grimm's terminology, we may wonder at terms that have not been discarded. Grimm speaks of consonant gradation. We no longer do, but our entire treatment of the Indo-European vowels is based on the assumption of gradation. Grimm viewed vowels as virtually hopeless, but brought order into the consonant system by his use of grades. Subsequent linguists brought order into the Indo-European vowel system by using grades. In maintaining their terminology, are we also maintaining an antiquated framework for the vowels?
Though we consider Jacob Grimm (1785-1863) one of the greatest contributors to linguistics, his name is a household word for other achievements as well. The veneration in which he is held by scholars may be indicated by the retention of the page numbers of his original text of the grammar, which are maintained here, as in subsequent editions. His Germanic Grammar is still the most complete one we have. The large German dictionary, recently completed, was inaugurated by him and his brother, Wilhelm (1786-1859). His work in other fields: medieval literature, law, mythology, folklore, is as fundamental as his work in linguistics. After studying law at Marburg he held small government posts, which brought him at various times to Paris and Vienna. In 1817 he was appointed professor and librarian at Göttingen. Here he lectured in his areas of interest until 1837, when with six other professors he protested against the King of Hanover's abrogation of the constitution, and was dismissed. His political action at this time illustrates that his greatness was not confined to academic matters. After returning to Cassel for a few years he and his brother were invited to professorships at Berlin in 1840, and to memberships in the Prussian Academy of Sciences. Acclaim did not hinder his work, which involved all areas of linguistics from phonology to the painstaking activities of a lexicographer.
The above survey informs us that the vowel relationships are uncertain and subject to various influences, but that their distribution and alternation are not arbitrary, rather, resulting from deeply established laws that have not yet been disclosed. The law of the ablauts will spread more light on this. One may view the vowels as the necessary coloring or animation of all words, as the breath without which they would not even exist. The real individuality of the word rests on the vowel sound; it affords the finest relationships.
The form, if I may say so, on the other hand the specification is established by the consonantism. Here the relationships appear far more certain and lasting; dialects, whose vowels for the most part deviate, often maintain the same consonants.
The four liquids are unchangeable; their fluid element preserves them intact during the most powerful upheavals. They undergo only occasional permutations, transpositions, losses or geminations, in spite of which their essential significance remains the same; i.e., although, for example, chilche occasionally appears as chirche, r and 1 remain fundamentally different in all other cases. To be noted:
1) On the one hand l and r are closely related, on the other m and n. When an exchange takes place, m is the earlier and more delicate, n later and coarser (cf. p. 386, 387). [These references are to Grimm's Grammar.] Conversely, the harder r may be older, the softer l younger. M stands in a special relationship with the labials, n with the linguals (cf. p. 536). Thus the OHG au, ou before m and labials, ô before n and linguals (p. 100); l and r are associated as readily with labials, linguals and gutturals. -- L and r disintegrate occasionally into u and i (and could therefore be called semivowels); never m and n, yet the influence of a lost n on the preceding vowel might be compared (gâs, for gans).
2) In the important association of r with s, of the combination rd with dd and sd (Goth. zd) r, rd appear as the younger forms which have gradually developed from s, sd (cf. p. 64, 65, 121, 167, 210, 244, 305, 317, 343, 387, 416).
Like the liquids, the three spirants v, h, s remain essentially unchanged throughout all the Germanic dialects. I deduce their inner relationship in part from the ê and ô which appear in front of them rather than ei and au (p. 91, 94), in part from the changes between h and v, w (p. 148, 403), h and s (p. 318, 416), and the association of the aspiration with the assibilation (th, ts, z); no direct exchange between v, w, and s; h and v, (the softest of all consonants,) disappear occasionally without replacement, even initially and particularly before liquids. (v. addendum)
Relationships are completely otherwise with the remaining consonants; a notable contrast between High German and all the other dialects becomes obvious. In the labial, lingual and guttural sounds, the Gothic (Saxon, Frisian, Northern) tenues correspond to the High German aspirates; the Gothic mediae to the High German tenues; and the Gothic aspirates to the High German mediae. The particulars may be expressed as follows:
A change has taken place by means of which each of these nine consonants in High German shifted similarly from its position.1 There is no doubt that the High German situation must be viewed here as the later, the changed, and the Gothic (Saxon, Frisian, Northern) as the earlier. This has been proved by analysis of the Old High German letters on various grounds. Observations:
1) The lingual series indicates the relationship most clearly; in Gothic táins, dal, þaúrnus are as necessarily distinguished as in Old High German Zein, Tal and Dorn.
2) The labial order also fits as soon as one acknowledges the second aspirate bh for the HG v in initial position and admits this instead of the closely related real media. For f, p, v, the erroneous designations ph, b, f were introduced, or occasionally others. Compare Goth. pund, baíran, filu with the HG funt, përan, vilo (also written phunt, bëran, filo). The older arrangement had visible effect in the inconsistent writing system; the strictest High German pronunciation, in which përan, pein, përag were completely current, did not even rise to the pure media bilo for filo, vilo. Even hard, upper German folk dialects do not know and cultivate such a b for f (certainly, however, many b for the spirant w). This all applies however for the initial position; in medial position the media frequently seems to me to stand in proper position, for example, in ëbar (aper), ëban (aequalis) etc. (cf. below, p. 589, fn. b.).
3) For the series of throat sounds the aspiration is lacking in Gothic, etc.: in High German all three gradations are found, but how are the High German k and g (ch assumed for Goth. k) to be organically divided into Gothic k? This could hardly be answered from the German language; the uncertainty of the Old High German writing system not only confuses k and g, but also k and ch with one another. At the same time however some clarification is provided by this that the OHG k which alternates with g never goes over to ch, and conversely, the k which alternates with ch never to g. So for example, gunni may not stand for chunni (genus) and chans never for gans (anser); for both, however, kunni and kans. Since in addition medial ch may not be exchanged with k (no sprëkan for sprëchan), then HG k for ch would be completely objectionable and of the two sounds, g and k, one would be superfluous and indeed theoretically this would be g. The High German language would thus not actually have any more throat sounds than the Gothic; ch would correspond to Goth. k, k however to g. Yet it appears to me that there is a third case where OHG g of necessity stands, i.e., where it cannot be replaced by k or ch; this is none other than the varying relationship between h and g (p. 427). Here the Goth. g plays a double role; in þragjan (currere), and guma (vir) a different g appears from that in áugô (oculus), and tagram (acrimis). This can become clear only through comparison of further originally related languages.
With such comparisons, which here cannot by any means be thoroughly pursued, but rather only are intended to put our Germanic sound-relationships into proper perspective one proceeds best from the consonants. If a thoroughly grounded statement is ascertained and accepted for these, then perhaps some insights might also be gained into the history of the vowels.
First we encounter the important principle: liquids and spirants agree in all essential relationships with the manner and arrangement of the German tongue. It seems that where the branches of the Germanic languages do not deviate from one another, Latin, Greek and Indic will not deviate. Sanskrit expressly recognizes the r and 1 as vowels, and uses r this way often, l more rarely. The weakening of the older m into later n is common everywhere; a large group of words with m in Sanskrit and Latin receive n in Greek (see addendum); exactly as the final MHG n becomes m again in medial position (lein, leimes; arn, armes, p. 386), so en is related to emen (Lat. eram, eramus; compare néon with novum). Analogous modifications of the s to r are also easily found; especially Latin preferred the r, which is however always to be understood as the younger form. Alternation of the spirants v (digamma), s and h is demonstrated by hespéra, vespera; heptá, for septem; hus; sus; hérpō, serpo; hekurós, socer; hupó, sub; sas, sâ (Skt. isea), Gk ho, hē, Goth. sa, sô; háls, sal; sasa, (Skt. lepus), haso etc.; also the initial spirant disappeared completely, e.g., Lat. anser is found for hanser (Skt. hamsa, cignus), odium for hodium (Goth. hatis), éar Lat. ver, and the Gk ídmen (Skt. vidmas, Lat. videmus, Goth. vitum) earlier had a digamma before it. V and s alternate the least, cf. sinister with winster.
Yet more astounding than the accord of the liquids and the spirants is the variation of the lip, tongue and throat sounds, not only from the Gothic, but also the Old High German arrangement. For just as Old High German has sunk one step down from the Gothic in all three grades, Gothic itself had already deviated by one step from the Latin (Greek, Sanskrit). [See supplement.] Gothic is related to Latin exactly as is Old High German to Gothic. The entire twofold sound shift, which has momentous consequences for the history of language and the rigor of etymology, can be so expressed in a table:
|OHG||B. (V)||F.||P. (V)|||||D.||Z.||T|||||G.||CH.||K.|
or otherwise conceived:
From this we see now how the Goth fills the gap arising from the departure of the throat aspirate: he uses the spiritus h initially rather than ch, and h occasionally also medially and finally, but frequently also the media g. In Old High German the g would appear here consistently everywhere and would be analogous to the b and d of the other series; it may however be a remnant of the earlier sound arrangements that the Gothic initial h has also been carried over to Old High German because it was taken for a spirant and not as aspirate. Only occasionally does g appear beside it. This use of the h for ch is also remarkably found precisely in initial position in Latin so that the gutturals, more precisely determined, show up as follows:
|κ||c||h, g||h, g|
The necessary examples for the proposed nine comparisons are:
I. (P. F. B, V.) 1) Initial position: pax, pacis, pacatus; Goth. fahêds (gaudium, quies), ON feginn (contentus, laetus) --- pes, pedis; Gk poȗs; podós Skt padas; Goth. fôtus; OHG vuoz --- piscis, fisks, visk --- porca (sulcus), OHG vuriha --- porcus, OHG varah --- Gk póros (iter, via), Goth. faran (ire) ---- pater, Gk patḗr, Goth. fadrs, OHG vatar --- patis (Skt conjux), Lith. pats, Gk posis (? Dor. Gk pótis), Goth. brûd-faþs (sponsus) --- Gk pȗr, OHG viur --- Gk polá OHG vilo, Goth. filu --- Gk pléos, Goth. fulls, OHG vol --- Gk prōí, OHG vruo --- pecus, Goth. faíhu, OHG vihu --- pulex, OHG vlôh --- plecto, OHG vlihtu --- Gk pérdō, Lith. perdziu, Swed. fjerter, OHG vërzu --- Gk palámē, Lat. palma, AS folma, OHG volma Gk ptéron (for petéron, like Gk petáō for Gk ptáō) ON fiödhur, OHG vëdar --- Gk peúkē, picea, HG vihta --- pellis, Goth. fill, OHG vël --- pullus, Goth. fula, OHG volo --- pauci, Goth. favai, OHG vaohê --- primus, Goth. frumists, OHG vromist. --- 2) Medial position (The Gothic medial b for f is less precise than Northern and Saxon f. bh) Gk kápros, caper, ON hafr --- Gk loipós (reliquus), ON leifar (reliquiae), Goth. láibôs --- svapa (Skt somnus), Gk húpnos, ON svefn, OS suëbhan --- septem, AS sëofon, Goth. sibun --- aper, ON iöfor, AS ëbar OHG ëbar --- Gk hupér, super, Goth. ufar, ON yfir, OHG ubar --- rapina, AS reáf, OHG roub.
II. (B, P, F) 1) For inital position, I know no example to support my view that the Germanic words with initial p, HG f (ph) are lacking (above p. 55, 131, 212, 247, 397, 462). 2) Medial position: Gk kánnabis, cannabis, ON hanpr, OHG hanaf; should turba be compared with Goth. þaúrp, OHG dorof; stabulum with ON stöpull, OHG staphol; labi with hláupan, loufan?
III. (PH, B, P) The aspirate of the older languages itself still requires closer attention; Sanskrit recognizes both ph and bh, which appear mixed in Gk ph, Lat. f and b. 1) Initial position: The Indic root bhu, the Gk phu, the Lat. fu in the verb 'to be', compare with the AS bëon, OHG pim (sum) --- Gk phēgós, fagus, ON beyki, OHG puocha --- forare, ON bora, OHG poren --- frangere, fregi; Goth. brikan, OHG prëchan --- frui, fructus; Goth. brûkôn, OHG prûchôn --- frater, brôþar, pruoder --- flare, blasan, plasan, --- fero (in Skt the root bhr), Goth. baíra, OHG piru --- Gk phúllon, folium, ON blad, OHG plat --- Gk ophrús, ON brâ, OHG prawa. --- 2) Medial position: Gk eléphas, antos, Goth. ulbandus, OHG olpenta --- Gk kephalḗ, haubiþ, houpit --- Gk nephlélē, nebula, Goth. nibls?, OHG nëpal --- Gk gráphein, Goth graban, OHG grapan. These medials vacillate toward the first class; like: caput, AS heáfod, OHG haubit, cf. the ON nifl to which an OHG nëbal would correspond.
IV. (T. TH. D.) 1) Initial position: tauta (lett. gens, regio) Goth. þiuda, OHG diot --- tu, Goth. þu, OHG dû --- tenuis, tener, ON þunnr, OHG dunni --- Gk teínein, tendere; Goth. þanjan, OHG denen --- Gk treȋs, tres; þreis; drî --- tergere, ON þërra --- Gk térsein (arefacere) Goth. þaursis (aridus) torridus, OHG durri --- tacere, Goth. þahan, OHG dagen --- Gk trékhein, Goth. þragjan --- Gk talận, tlận, tolerare, Goth. þulan, OHG dolen --- tectum, Goth. þak, OHG dach --- Gk taũros, ON þiôr --- tad (Skt id), Gk to (for tad), Goth. þat, OHG daz --- talis, ON þvîlîkr. --- 2) Medial position: ratio, raþjô, redja --- frater, brôþar, pruoder --- Gk metá, Goth. miþ --- dantas (dens, dentis) tunþus, zand --- rota, ON hradhr (celer) OHG hrad (rota) --- iterum, Goth. viþra, OHG widar --- Gk héteros, anþar, andar --- perhaps Gk étēs, hetaȋros (socius) may be compared with OS gesith, OHG sindeo --- étos (annus) with the obscure Goth. ataþni (i.e. at-þni, OHG az-adani?).
V. (D. T. Z.) 1) Initial position: dingua, tuggo, zunga (cf. above p. 152) --- deus, divus, Lith. diéwas; Gk dís, díos (for theós is Cretan) ON tŷr; OHG ziu (cf. above p. 150, 151) --- dantas (Skt) Gk odoús odoú odóntos; dens, dentis; Goth. tunþus, OHG zand --- Gk dia-, Lat. dis-, Saxon to-, OHG zi- --- Gk damą̑n, domare, Goth. tamjan, OHG zemen --- Gk drȗs, Goth. triu --- digitus, cf. with the Saxon têkan (signum) OHG zeichan --- Gk deiknúein, deíkein, indicare, Saxon tôgjan, HG zeigen --- Gk dólos, dolus, ON tâl, OHG zâla --- ducere, Goth. tiuhan, OHG ziohan --- Gk dúo, duo, Goth. tva, OHG zuei. --- Gk dákru, Goth. tagr, OHG zahar --- Gk deksiá, dextra, Goth. taíhsvô, OHG zësawa. --- 2) Medial position: Gk hēdú, Goth. suti, OHG suozi --- ad, Goth. at, OHG az --- Gk hédos, sedes; sedere, Goth. sitan, OHG sizan --- Gk édein, edere; itan, ëzan --- Gk eídein, eidénai, videre, Goth. tva, OHG wizen --- odium, Goth. hatis, OHG haz --- claudere, OHG sliozan --- laedere, HG letzen, --- radix, ON rôt --- Gk húdōr, Goth. vatô, OHG wazar --- Gk hidrōs, sudor, sveiti, sueiz --- pedes, fôtjus, vuozi.
VI. (TH. D. T.) The Latins have no th (except in foreign words), but often the Gk th has become the labial aspirate f of the same grade just as in Greek itself the Aeolic dialect shows ph for th (cf. Gk thumós, spiritus, animus, with fumus, Gk phúmos; thúein with fire, suffire) both remind one of the intersection of Goth. þl with fl indicated on p. 66, 67. 1) Initial position: Gk thugátēr, Goth. daúhtar, OHG tohtar --- Gk thúra, Lat. pl. fores, Goth. daúr, OHG tor --- Gk thḗr, Aeol. phḗr, Lat. fera, ON dŷr, OHG tior --- Gk tharréein (audere) Goth. ga-daúran, OHG turran, cf. the preterite ga-daúrsta, getorsta with Gk thárros, thársos, thrasús. --- Gk thénar (vola manus) OHG tënar --- Medial position: Gk méthu, AS mëdo, OHG mëtu --- Gk éthos, AS sido, OHG situ.
VII. (K. H,G. H,G.) In the second grade the Goth. h is found for ch; in the third the OHG h for g. 1) Initial position: claudus, halts, halz --- Gk kánnabis, ON hanpr, OHG hanaf --- canere cf. with hano (gallus, as this with ON kalla, OHG challôn, clamare, fari) --- caput, háuþip, houbit --- Gk kardía, cor, haírtô, hërza --- Gk kúōn, canis, hunþs, hund --- Gk koȋlos, hol --- celare, hilan, hëln --- Gk kálamos, calamus, halam, helm --- Gk kártos, karterós, hardus, hart --- cornu, haúrn, horn --- collum, hals --- Gk krumós, (gelu), ON hrîm --- Gk klaíein, Goth. hlahan --- Gk krázein, crocitare, Goth. hrukjan --- Gk kléptēs, Goth. hliftus. --- 2) Medial position: Gk ókos, oculus, áugo, ouga --- acies, OHG egga --- lux (lucs) liuhad, lioht, cf. Gk leukós with liuhadeins --- Gk oȋkos, Goth. veihs --- lacus, AS lagu --- acus, aceris, OHG ahan, agan, --- Gk dákru, tagr, zahar --- tacere, þahan, dagen --- pecus, faíhu, vitro --- Gk hekurós, socer, Goth. svaíhra, HG schwager, schwieger --- Gk mḗkōn (papaver), OHG mâgan, NHG mohn (? Goth. mêhan). Medially this sometimes corresponds to Skt sh: e.g. dasha, Gk déka, Lat. decem, Goth. taíhun, Lith. deszimts.
VIII. (G. K. CH.) 1) Initial position: granum, ON korn, OHG chorn --- Gk génos, genus; kuni; chunni --- Gk génus, gene, ON kinn, OHG chinni --- Gk gonu, ON knê, OHG chnio --- Gk gunḗ, ON kona, OHG chona --- gelu (frigus) Goth. kalds, OHG chalt --- gula (guttur) OHG chëla --- gustare, kiusan, chiosan --- gau (Skt vacca), ON kû, OHG chua --- 2) Medial position: Gk egṓ, ego, Goth. ïk, OHG ih --- vigil, OHG wachar --- Gk agrós, agere, Goth. akrs, OHG achar --- Gk ágein, agere, ON aka --- Gk mégas, mégalos; mikils; michil --- rex, regis, regnum; reiks; rîchi --- jugum, juk, joch --- augere, áukan, auchôn --- Gk amélgein, mulgere, ON miólka, OHG mëlchan.
IX. (CH, H. G. K.) In Latin h is here equivalent to ch. cf. Schneider, Lateinische Grammatik, p. 202: Gk kheimṓn, hiems; Gk kheír, Lat. hir; Gk khḗr, herinaceus. Frequently however, OHG g to k, which I carry out here only in theory. 1) Initial position: Gk khḗn, anser (for hanser) Goth. gans, OHG kans --- Gk khéō (fundo), Gk khutós (fusus) Goth. giutan, OHG kiozan --- Gk kholḗ, ON gall, OHG kalla --- Gk khthés, heri, hesternus, Goth. gistra, OHG këstar --- Gk khórtos, hortus, gards, OHG karto --- hostis (peregrinus) gasts, kast --- homo, Goth. guma, OHG komo --- Gk khthṓn like khthés for khés for khṓn and this for khóm, cf. khamái, humi, humus; to be compared with Goth. gauï, OHG kouwi, kou --- 2) Medial position: Gk ékhein, Goth. áigan, OHG eikan --- Gk trékhein, Doric trákhein, Goth. þragjan --- Gk lékhos, Goth. ligrs, OHG lëkar --- Gk leíkhō, líkhō (lambo) Goth. laígo, OHG lêkôn --- Gk lukhą̑n (insidiari), (Goth. lêgôn ?), OHG lâkôn. --- (see supplement).
Notes on this comparison of consonants:
1) Even if certain of the cited samples still appear to be dubious or uncertain, the majority may be considered as clearly demonstrated because of the analogy of the gradation; the correctness of the rules in general is unmistakable. Words in which two consonants agree are doubly certain (Gk trékhein, þragjan; pódes, fôtjus); those in which one consonant agrees, another deviates, are suspicious; even more suspicious, those whose consonants showed essential equivalence in the three languages without gradation. In this case, relationship is either entirely lacking (e.g. AS pädh, padhas and Gk páthos, dolor) or the one language has borrowed from the other (e.g. scrîban is scribere itself, fruht is fructus, hence not Germanic; the same is true for OS sicor, Lat. securus).
2) In the investigations of the words, likeness or resemblance of consonants which are in general related is less important than observation of the historical course of gradation, which does not become disturbed or reversed. A High German word with p, which shows b in Gothic and f in Latin is originally related in these three languages: each possesses it unborrowed. If, however, we were to find an f in a High German word, b in Gothic and p in Latin, then the relationship would be nonsensical, even though in the abstract exactly the same relationships of the letters are present. The Gk t requires a Goth. þ, the Gothic t however no Gk th but rather d, and so the identity throughout is based on the external difference.
3) Words, which the one or other language does not possess, could readily be posited for the nine consonant relationships, but not in the elements of vowels, liquids and spirants. All hypothesizing accordingly remains unprofitable; we might at most claim that for example Gk dáphnē would have to have t-b in Gothic and z-p in High German; Gk phutón b-þ in Gothic and p-d in High German. These nine rules are only touchstones for words which are available. Analogy is generally not sufficient for new creations, for everything alive is incalculable and merges the laws of theory with the exceptions found in reality.
4) Such exceptions, i.e. instances, where the proposed comparisons fail, appear:
a) in the transition of the tenues, mediae or aspiratae, to tenues, mediae or aspiratae of another series. How often do the members of one series exchange with one another: p, t, (Gk taȏs, pavo; Gk pénte, Aeol. pémpe; Gk poȋos, Ion. koȋos) b, d, g (Gk obelós, Dor. odelós; Gk gȇ Dor. dȇ: cf. above p. 445, 446) ph, th, ch (examples above p. 587).
b) because of the imperfection of the aspirations in most languages and the mixing with the related spirants and mediae which arises out of it. Sanskrit has aspiration of the mediae and tenues of each organ, so that bh, ph; dh, th; gh, and kh are found. Jumbled relics of these appear in the other languages. The Greek speaker has ph, th, and ch; the Latin only the first (and then it is modified; his f is close to thebh); th becomes f for him; ch becomes h. Also the Lithuanian and Latvian languages both lack f, th and ch (yes even the simple spirant h); Gothic etc. lack ch, which they replace with h and g. In other Germanic dialects, distinct traces of the bh, dh and gh, which can probably be found more clearly in the future than could happen in my presentation. The lack of initial Goth. p, HG ph (f), appears less striking with this point of view. Since in Greek and Latin the labials fluctuate, e.g. Gk kephalḗ, caput; Gk néphos, nephélē, nubes, nebula; thus each of the Germanic forms is justified, Goth. háubiþ beside gibla and the Saxon heafod; and it must in general remain undecided whether OHG houbit or houpit, nëpal or nëbal deserves preference. The Latins loved medial mediae (habeo, nobilis, mobilis, fabula, cibus, hebes, scabies etc.; origin in v is obvious in novisse, movere, etc.).
c) The sound shift takes place in the mass, but never neatly in individual items; words remain in the relationship of the old arrangements-the stream of innovation has passed them by. Connection with the unchangeable liquids and spirants has usually (not always) preserved them. Thus, α) some words of the Gothic etc. languages still have the stamp of the Latin and Greek order, e.g. du, dis (cited p. 152), compare with Saxon tô and OHG zuo, zi, zër; daddjan (Dan. dîe) was erroneously cited, which is related according to the sixth comparison with Gk tháein, and has nothing to do with the AS tit. Further examples are OS sëdel, instead of sëtel (p. 217), the ON pt instead of ft (p. 314). The relationship between dies, days, däg-dagr may not be interpreted otherwise. β) some of the Old High German words have the stamp of the Gothic etc., as in the words enumerated p. 154, 155, 394. γ) some Gothic and Old High German (the latter accordingly unscathed through two sound shifts) agree with the Latin and Greek e.g. the cited AS tit, Eng. teat, OHG tutto (p. 155), Gk títthē. Further: longus, laggs, lângr; angustus, aggvus, engi; gramen, gras etc.2 δ) of two consonants in a word one may be shifted, the other retained, e.g. in tunga, zunga, lingua, the g remained, while d (dingua) underwent gradation; the lingual does not check in prudentia, Goth. frôdei and Lith. protas; gaudere too may be closely related to Goth. gatjan (facere ut aliquis obtineat restituere, from gitan, like nasjan from nisan) and MHG ergetzen, and for the stricter form katjan (ON kâtr, laetus, beside gëta acquirere and gëtaz acquiescere) go to the MH erchetzen. This possibly misleading sentence should not be misused by the etymologists.
5) I have presented the Old High German sound shift (p. 127, 151, 177) as something non-organic, and admittedly it is a visible deviation from an earlier organism which is still present in vestiges. One must also consider Gothic in contrast with Greek and Latin as equally non-organic. The similarities of both changes puts them right in the proper light. They are great events in the history of our language and neither is without inner necessity.3 It is also not to be overlooked that each gradation fills ever smaller circles. The peculiarity of the latter does not extend beyond the High German dialect. The earlier one encompassed Gothic, Saxon, Northern still; accordingly it had a more significant extent. And how restricted this appears when contrasted with the still older situation, which we must recognize for the Latin, Greek and Indic languages, and to which in general also the Slavic and Latvian tribes adhere, perhaps with some modifications. For example, since aspiration is lacking for the Latvians, Prussians and Lithuanians, they are accustomed to use a media for it or sibilants (see addendum). But they possess the unmodified (Latin and Greek) tenues and mediae, cf. the Lith. pilnas (plenus) (see addendum) pirmas (primus) pakájus (pax, pacis) piemů (poimḗn) peda (vestigium) tris (tres), tu (tu) traukti (trahere) kampas (campus) kas (quis) kélas (kéleuthos) akis (oculus) ratas (rota) dantis (dens) antras (Goth. anþar) wertas (Goth. vaírþs) derwà (ON tiara, NHG zehr) trokszti (NHG dürsten) du (duo) sedeti sedere etc. Similarly in Slavic pasti (pascere) vepr (aper) piti (píein) pokoj (pax) mater (mater) sjekati (secare) videti (videre) dom (domus) smrt (mors, mortis) ptak (pterón) etc. For this reason, the Slavic and Lettish languages are without doubt closer to the Latin and Greek than the Gothic, and this is closer than the Old High German.
6) The result of the sound shift brings it about that HG z (for t) fully takes the place of the th, as HG ph for p, and ch for c. This High German equating of the z (ts) with th is even more remarkable, in part because in no monument known to me is an actual exchange between z and th apparent (no trace of an OHG thiman, thein for ziman, zein), and in part because in the High German dialect the pure spirant h is strongly favored and never is exchanged with the spirant s. This exchange prevails precisely in the Slavic and Lettish languages, in which so many of the original gutturals appear assibilated, cf., cor, cordis, hërza with the Lith. szirdis (pronounce schirdis), Bohemian srdce; canis, hund with the Lith. szů; centum, hundert with the Lith. z̦̓ (pronounce sh or dsh) answers to the Gk kh, Lat. h as: ziema (kheȋma, hiems), zeme (humus, cf. humilis and khthamalós, khamalós), zmogus, (homo, pl. zmones, homines; OPruss. smunents, homo), zasis (khḗn, anser); zengti, zengimas is the Germanic gangan, gang. One should compare, however, the AS sceort, Eng. short for cëort and even the OHG scurz for churz (above p. 175) as well as the hissing pronunciation of the Frisian, English and Swedish initials c, k ch.
The relationships of the consonants accordingly provide adequate proof of original relationship of the compared languages. Might not also, based on this, at the same time contacts between the vowels be detected? -- the analogy between the High German and Gothic vowel situation not lead to the conclusion that Latin vowels too must be connected with Gothic? The connection will be even more uncertain and disrupted for this reason because in the Germanic dialects with the same consonantal gradation we meet such varying and manifold vowels. Nonetheless there are still unmistakable similarities like those given below.
(580-581) The relationship of the semi-vowels v and j (p. 9) to the spirants v,s, h, (p. 10) is still obscure. First of all, the lingual order has no semi-vowel at all. Secondly, the gutturals have a semi-vowel j which is distinct from the spirant h. Finally, the question arises whether the semi-vowel v falls together with the spirant v? I have already touched on this puzzle on p. 187. It is to be noted that semi-vowels (i.e. vowels with consonantal value) only develop from i and u, not from a, obviously not from the nonoriginal e and o. And since further l and r can develop to u and i, they are semi-vocalic in a reverse sense, i.e. consonants with vocalic value. Is it related to the richer endownment of the throat sound series that the aspirate is occasionally withdrawn from it?
(583, 32) madidus, mador, Goth. natjan, OHG naz.
(584, 15) If one also assumed a fourth grade, then the sound would return to the first grade. Isolated items might possibly be put there, such as the ch and z in châpi, hagestolz cited in the addenda to p. 185 and 526; these however are non-organic exceptions. Never does such a thing show up in an established regular series.
(585-588) Some more examples are added here for the nine comparisons.
I, 1. pallidus, Lith. palwas, ON fölr, OHG valêr; Slav. post (jejunium) OHG vasta; Lith. pauksztis (avis) Goth. fugls; Slav. plst (coactile) OHG vilz; Slav. pjast (pugnus) OHG vûst; Gk péras, Goth. fêra.
I, 2. nepos, OHG nëvo; Gk kȇpos, OHG hof, hoves; copia, hûfo; hoplḗ, ON hôfr, OHG huof, huoves.
II, 2. Lith. obolys, Russ. jabloko, ON epli, OHG epfili, Russ. obezjana (simia) Bohemia opice, ON api, OHG affo.
IV, 1. trituro, AS þërsce, OHG driscu; tonitru, AS þaërsce, OHG driscu; tonitru, AS þunor, OHG donar; Slav. trn, tern (spina) Goth. þúrnus OHG dorn.
V, 2. kardía, con, cordis, haírtô, hërza; radix, ON rôt; hoedus, ON geit, OHG keiz; madidus, OHG naz; kónis, kônidos, ON nit, OHG niz (instead of knit, hniz); nidus, Slav. gniezdo, AS nëst, OHG nëst; possibly nodus, Goth. nati (consisting of knots) OHG nezi.
VII, 1. Gk kȇpos, hof; copia, hûfo; crinis, hâr; cerebrum, hirni.
VII, 2. pulex (pulec-s) Slav. blocha, OHG vlôh.
VIII, 1. Slav. gnjetu (premere, depsere) OHG chnëtan).
VIII, 2. Lith. nogas (nudus) ON naktr, OHG nacchot.
IX, 1. hoedus (= hoidus) ON geit.
(591, 22) In the Slavic initial position the media of the second or third grade occasionally prevails, especially in the combinations bl, br, gn, gr, e.g. blocha (pulex) brat (frater) (bronja) (lorica, Dobr. p. 115) OHG prunja; gnida (kónis, kónidos Dobr. 195); graditi (cingere, Goth. gaúrdan) etc.; to the Germanic hl, hu correspond chl, chv, e.g. chvila (more) hvîla; chlev, hleip and many others.
(591, 24) pilnas, plenus, Slav. pln, poln.