XVI.—§ 3. Facts, not Fancies.
As early as 1867 Michel Bréal, in an excellent article (reprinted in M 267 ff.), called attention to the dangers resulting from the general tendency of comparative linguists to “jump intermediate steps in order at once to mount to the earliest stages of the language,” but his warning has not taken effect, so that etymologists in dealing with a word found only in comparatively recent times will often try to reconstruct what might have been its Proto-Aryan form and compare that with some word found in some other language. Thus, Falk and Torp refer G. krieg to an Aryan primitive form *grêigho-, *grîgho-, which is compared with Irish bríg ‘force.’ But the German word is not found in use till the middle period; it is peculiar to German and unknown in related languages (for the Scandinavian and probably also the Dutch words are later loans from Germany). These writers do not take into account how improbable it is that such a word, if it were really an old traditional word for this fundamental idea, should never once have been recorded in any of the old documents of the whole of our family of languages. What should we think of the man who would refer boche, the French nickname for ‘German’ which became current in 1914, and before that time had only been used for a few years and known to a few people only, to a Proto-Aryan root-form? Yet the method in both cases is identical; it presupposes what no one can guarantee, that the words in question are of those which trot along the royal road of language for century after century without a single side-jump, semantic or phonetic. Such words are the favourites of linguists because they have always behaved themselves since the days of Noah; but others are full of the most unexpected pranks, which no scientific ingenuity can discover if we do not happen to know the historical facts. Think of grog, for example. Admiral Vernon, known to sailors by the nickname of “Old Grog” because he wore a cloak of grogram (this, by the way, from Fr. gros grain), in 1740 ordered a mixture of rum and water to be served out instead of pure rum, and the name was transferred from the person to the drink. If it be objected that such leaps are found only in slang, the answer is that slang words very often become recognized after some time, and who knows but that may have been the case with krieg just as well as with many a recent word?
At any rate, facts weigh more than fancies, and whoever wants to establish the etymology of a word must first ascertain all the historical facts available with regard to the place and time of its rise, its earliest signification and syntactic construction, its diffusion, the synonyms it has ousted, etc. Thus, and thus only, can he hope to rise above loose conjectures. Here the great historical dictionaries, above all the Oxford New English Dictionary, render invaluable service. And let me mention one model article outside these dictionaries, in which Hermann Möller has in my opinion given a satisfactory solution of the riddle of G. ganz: he explains it as a loan from Slav konǐcǐ ‘end,’ used especially adverbially (perhaps with a preposition in the form v-konec or v-konc) ‘to the end, completely’; Slav c = G. z, Slav k pronounced essentially as South G. g; the gradual spreading and various significations and derived forms are accounted for with very great learning (Zs. f. D. Alt. 36. 326 ff.). It is curious that this article should have been generally overlooked or neglected, though the writer seems to have met all the legitimate requirements of a scientific etymology.
XVI.—§ 4. Hope.
I have endeavoured to fulfil these requirements in the new explanation I have given of the word hope (Dan. håbe, Swed. hoppas, G. hoffen), now used in all Gothonic tongues in exactly the same signification. Etymologists are at variance about this word. Kluge connects it with the OE. noun hyht, and from that form infers that Gothonic *hopôn stands for *huqôn, from an Aryan root kug; he says that a connexion with Lat. cupio is scarcely possible. Walde likewise rejects connexion between cupio and either hope or Goth. hugjan. To Falk and Torp hope has probably nothing to do with hyht, but probably with cupio, which is derived from a root *kup = kvap, found in Lat. vapor ‘steam,’ and with a secondary form *kub, in hope, and *kvab in Goth. af-hwapjan ‘choke’—a wonderful medley of significations. H. Möller (Indoeur.-Semit. sammenlignende Glossar. 63), in accordance with his usual method, establishes an Aryo-Semitic root *k̑-u̯-, meaning ‘ardere’ and transferred to ‘ardere amore, cupiditate, desiderio,’ the root being extended with b-: p- in hope and cupio, with gh- in Goth. hugs, and with g̑- in OE. hyht. Surely a typical example of the perplexity of our etymologists, who disagree in everything except just in the one thing which seems to me extremely doubtful, that hope with the present spiritual signification goes back to common Aryan. Now, what are the real facts of the matter? Simply these, that the word hope turns up at a comparatively late date in historical times at one particular spot, and from there it gradually spreads to the neighbouring countries. In Denmark (håb, håbe) and in Sweden (hopp, hoppas) it is first found late in the Middle Ages as a religious loan from Low German hope, hopen. High German hoffen is found very rarely about 1150, but does not become common till a hundred years later; it is undoubtedly taken (with sound substitution) from Low German and moves in Germany from north to south. Old Saxon has the subst. tō-hopa, which has probably come from OE., where we have the same form for the subst., tō-hopa. This is pretty common in religious prose, but in poetry it is found only once (Boet.)—a certain indication that the word is recent. The subst. without tō is comparatively late (Ælfric, ab. 1000). The verb is found in rare instances about a hundred years earlier, but does not become common till later. Now, it is important to notice that the verb in the old period never takes a direct object, but is always connected with the preposition tō (compare the subst.), even in modern usage we have to hope to, for, in. Similarly in G., where the phrase was auf etwas hoffen; later the verb took a genitive, then a pronoun in the accusative, and finally an ordinary object; in biblical language we find also zu gott hoffen. Now, I would connect our word with the form hopu, found twice as part of a compound in Beowulf (450 and 764), where ‘refuge’ gives good sense: hopan to, then, is to ‘take one’s refuge to,’ and to-hopa ‘refuge.’ This verb I take to be at first identical with hop (the only OE. instance I know of this is Ælfric, Hom. 1. 202: hoppode ongean his drihten). We have also one instance of a verb onhupian (Cura Past. 441) ‘draw back, recoil,’ which agrees with ON. hopa ‘move backwards’ (to the quotations in Fritzner may be added Laxd. 49, 15, þeir Osvígssynir hopudu undan).[75] The original meaning seems to have been ‘bend, curb, bow, stoop,’ either in order to leap, or to flee, from something bad, or towards something good; cf. the subst. hip, OE. hype, Goth. hups, Dan. hofte, G. hüfte, Lat. cubitus, etc. (Holthausen, Anglia Beibl., 1904, 350, deals with these words, but does not connect them with hop, -hopu, or hope.) The transition from bodily movement to the spiritual ‘hope’ may have been favoured by the existence of the verb OE. hogian ‘think,’ but is not in itself more difficult than with, e.g., Lat. ex(s)ultare ‘leap up, rejoice,’ or Dan. lide på ‘lean to, confide in, trust,’ tillid ‘confidence, reliance’; and a new word for ‘hope’ was required because the old wen (Goth. wens), vb. wenan, had at an early age acquired a more general meaning ‘opinion, probability,’ vb. ‘suppose, imagine.’ The difficulty that the word for ‘hope’ has single or short p (in Swed., however, pp), while hop, OE. hoppian, has double or long p, is no serious hindrance to our etymology, because the gemination may easily be accounted for on the principle mentioned below (Ch. XX § [9]), that is, as giving a more vivid expression of the rapid action.
XVI.—§ 5. Requirements.
It is, of course, impossible to determine once for all by hard-and-fast rules how great the correspondence must be for us to recognize two words as ‘etymologically identical,’ nor to say to which of the two sides, the phonetic and the semantic, we should attach the greater importance. With the rise of historical phonology the tendency has been to require exact correspondence in the former respect, and in semantics to be content with more or less easily found parallels. One example will show how particular many scholars are in matters of sound. The word nut (OE. hnutu, G. nuss, ON. hnot, Dan. nød) is by Paul declared “not related to Lat. nux” and by Kluge “neither originally akin with nor borrowed from Lat. nux,” while the NED does not even mention nux and thus must think it quite impossible to connect it with the English word. We have here in two related languages two words resembling each other not only in sound, but in stem-formation and gender, and possessing exactly the same signification, which is as concrete and definite as possible. And yet we are bidden to keep them asunder! Fortunately I am not the first to protest against such barbarity: H. Pedersen (KZ n.f. 12. 251) explains both words from *dnuk-, which by metathesis has become *knud-, while Falk and Torp as well as Walde think the latter form the original one, which in Latin has been shifted into *dnuk-. Which of these views is correct (both may be wrong) is of less importance than the victory of common sense over phonological pedantry.
There are two explanations which have had very often to do duty where the phonological correspondence is not exact, namely root-variation (root-expansion with determinatives) and apophony (ablaut). Of the former Uhlenbeck (PBB 30. 252) says: “The theory of root determinatives no doubt contains a kernel of truth, but it has only been fatal to etymological science, as it has drawn the attention from real correspondences between well-substantiated words to delusive similarities between hypothetical abstractions.” Apophony inspires more confidence, and in many cases offers fully reliable explanations; but this principle, too, has been often abused, and it is difficult to find its true limitations. Many special applications of it appear questionable; thus, when G. stumm, Dan. stum, is explained as an apophonic form of the adj. stam, Goth. stamms, from which we have the verb stammer, G. stammeln, Dan. stamme: is it really probable that the designation of muteness should be taken from the word for stammering? This appears especially improbable when we consider that at the time when the new word stumm made its appearance there was already another word for ‘mute,’ namely dumm, dumb, the word which has been preserved in English. I therefore propose a new etymology: stumm is a blending of the two synonyms still(e) and dum(b), made up of the beginning of the one and the ending of the other word; through adopting the initial st- the word was also associated with stump, and we get an exact correspondence between dumm, dum, stumm, stum, applied to persons, and dumpf, stumpf, Dan. dump, stump, applied to things. Note that in those languages (G., Dan.) in which the new word stum(m) was used, the unchanged dum(m) was free to develop the new sense ‘stupid’ (or was the creation of stum occasioned by the old word tending already to acquire this secondary meaning?), while dumb in English stuck to the old signification.
XVI.—§ 6. Blendings.
Blendings of synonyms play a much greater rôle in the development of language than is generally recognized. Many instances may be heard in everyday life, most of them being immediately corrected by the speaker (see above, XV § [4]), but these momentary lapses cannot be separated from other instances which are of more permanent value because they are so natural that they will occur over and over again until speakers will hardly feel the blend as anything else than an ordinary word. M. Bloomfield (IF 4. 71) says that he has been many years conscious of an irrepressible desire to assimilate the two verbs quench and squelch in both directions by forming squench and quelch, and he has found the former word in a negro story by Page. The expression ‘irrepressible desire’ struck me on reading this, for I have myself in my Danish speech the same feeling whenever I am to speak of tending a patient, for I nearly always say plasse as a result of wavering between pleje [plaiə] and passe. Many examples may be found in G. A. Bergström, On Blendings of Synonymous or Cognate Expressions in English, Lund, 1906, and Louise Pound, Blends, Their Relation to English Word Formation, Heidelberg, 1914. But neither of these two writers has seen the full extent of this principle of formation, which explains many words of greater importance than those nonce words which are found so plentifully in Miss Pound’s paper. Let me give some examples, some of them new, some already found by others: