[1] See his essay on Herder’s “Ursprung der sprache” in Modern Philology, 5. 117 (1907).

[2] It dates back to Vulcanius, 1597; see Streitberg, IF 35. 182.

[3] I have given a life of Rask and an appraisement of his work in the small volume Rasmus Rask (Copenhagen, Gyldendal, 1918). See also Vilh. Thomsen, Samlede afhandlinger, 1. 47 ff. and 125 ff. A good and full account of Rask’s work is found in Raumer, Gesch.; cf. also Paul, Gr. Recent short appreciations of his genius may be read in Trombetti, Come si fa la critica, 1907, p. 41, Meillet, LI, p. 415, Hirt, Idg, pp. 74 and 578.

[4] Only in one subordinate point did Rask make a mistake (b = b), which is all the more venial as there are extremely few examples of this sound. Bredsdorff (Aarsagerne, 1821, p. 21) evidently had the law from Rask, and gives it in the comprehensive formula which Paul (Gr. 1. 86) misses in Rask and gives as Grimm’s meritorious improvement on Rask. “The Germanic family has most often aspirates where Greek has tenues, tenues where it has mediÆ, and again mediÆ where it has aspirates, e.g. fod, Gr. pous; horn, Gr. keras; ÞrÍr, Gr. treis; padde, Gr. batrakhos; kone, Gr. gune; ti, Gr. deka; bÆrer, Gr. phero; galde, Gr. khole; dØr, Gr. thura.” To the word ‘horn’ was appended a foot-note to the effect that h without doubt here originally was the German ch-sound. This was one year before Grimm stated his law!

[5] The muddling of the negatives is Grimm’s, not the translator’s.

[6] I am therefore surprised to find that in a recent article (Am. Journ. of Philol. 39. 415, 1918) Collitz praises Grimm’s view in preference to Rask’s because he saw “an inherent connexion between the various processes of the shifting,” which were “subdivisions of one great law in which the formula T:A:M may be used to illustrate the shifting (in a single language) of three different groups of consonants and the result of a double or threefold shifting (in three different languages) of a single group of consonants. This great law was unknown to Rask.” Collitz recognizes that “Grimm’s law will hold good only if we accept the term ‘aspirate’ in the broad sense in which it is employed by J. Grimm”—but ‘broad’ here means ‘wrong’ or ‘unscientific.’ There is no kreislauf in the case of initial k = h; only in a few of the nine series do we find three distinct stages (as in tres, three, drei); here we have in Danish three stages, of which the third is a reversal to the first (tre); in E. mother we have five stages: t, Þ, Ð, d, (OE. modor) and again Ð. Is there an “inherent connexion between the various processes of this shifting” too?

[7] Probably under the influence of Humboldt, who wrote to him (September 1826): “Absichtlich grammatisch ist gewiss kein vokalwechsel.”

[8] Humboldt’s relation to Bopp’s general ideas is worth studying; see his letters to Bopp, printed as Nachtrag to S. Lefman’s Franz Bopp, sein leben und seine wissenschaft (Berlin, 1897). He is (p. 5) on the whole of Bopp’s opinion that flexions have arisen through agglutination of syllables, the independent meaning of which was lost; still, he is not certain that all flexion can be explained in that way, and especially doubts it in the case of ‘umlaut,’ under which term he here certainly includes ‘ablaut,’ as seen by his reference (p. 12) to Greek future stalÔ from stÉllo; he adds that “some flexions are at the same time so insignificant and so widely spread in languages that I should be inclined to call them original; for example, our i of the dative and m of the same case, both of which by their sharper sound seem intended to call attention to the peculiar nature of this case, which does not, like the other cases, denote a simple, but a double relation” (repeated p. 10). Humboldt doubts Bopp’s identification of the temporal augment with the a privativum. He says (p. 14) that cases often originate from prepositions, as in American languages and in Basque, and that he has always explained our genitive, as in G. manne-s, as a remnant of aus. This is evidently wrong, as the s of aus is a special High German development from t, while the s of the genitive is also found in languages which do not share in this development of t. But the remark is interesting because, apart from the historical proof to the contrary which we happen to possess in this case, the derivation is no whit worse than many of the explanations resorted to by adherents of the agglutinative theory. But Humboldt goes on to say that in Greek and Latin he is not prepared to maintain that one single case is to be explained in this way. Humboldt probably had some influence on Bopp’s view of the weak preterit, for he is skeptical with regard to the did explanation and inclines to connect the ending with the participle in t.

[9] Humboldt seems to be the inventor of this term (1821; see Streitberg, IF 35. 191).

[10] It has been objected to the use of Aryan in this wide sense that the name is also used in the restricted sense of Indian + Iranic; but no separate name is needed for that small group other than Indo-Iranic.

[11] In Lefmann’s book on Bopp, pp. 292 and 299, there are some interesting quotations on this point.

[12] For example, the correct appreciation of Scandinavian o sounds and especially the recognition of syllables without any vowel, for instance, in G. mittel, schmeicheln, E. heaven, little; this important truth was unnoticed by linguists till Sievers in 1876 called attention to it and Brugmann in 1877 used it in a famous article.

[13] A young German linguist, to whom I sent the pamphlet early in 1886, wrote to me: “Wenn man sich den spass machte und das ding Übersetzte mit der bemerkung, es sei vor vier jahren erschienen, wer wÜrde einem nicht trauen? MerkwÜrdig, dass solche sachen so unbemerkt, ‘dem kleinen veilchen gleich,’ dahinschwinden kÖnnen.” A short time afterwards the pamphlet was reprinted with a short preface by Vilh. Thomsen (Copenhagen, 1886).

[14] In numerous papers in North Am. Review and elsewhere, and finally in the pamphlet Max MÜller and the Science of Language, a Criticism (New York, 1892). MÜller’s reply to the earlier attacks is found in Chips from a German Workshop, vol. iv.

[15] Who was the discoverer of the palatal law? This has been hotly discussed, and as the law was in so far anticipated by other discoveries of the ’seventies as to be “in the air,” it is perhaps futile to try to fix the paternity on any single man. However, it seems now perfectly clear that Vilhelm Thomsen was the first to mention it in his lectures (1875), but unfortunately the full and able paper in which he intended to lay it before the world was delayed for a couple of years and then kept in his drawers when he heard that Johannes Schmidt was preparing a paper on the same subject: it was printed in 1920 in the second volume of his Samlede Afhandlinger (from the original manuscript). Esaias TegnÉr had found the law independently and had printed five sheets of a book De ariska sprÅkens palataler, which he withdrew when he found that Collitz and de Saussure had expressed similar views. Karl Verner, too, had independently arrived at the same results; see his Afhandlinger og Breve, 109 ff., 305.

[16] “Es ist besser, bei solchen versuchen zu irren als gar nicht darÜber nachzudenken,” Curtius, K 145.

[17] In this book the age of a child is indicated by stating the number of years and months completed: 1.6 thus means “in the seventh month of the second year,” etc.

[18] An American child said autonobile [?t?nobi·l] with partial assimilation of m to the point-stop t.

[19] Cf. Beach-la-Mar, below, Ch. XII § 1.

[20] Cf. below on the disappearance of the word son because it sounds like sun (Ch. XV. § 7).

[21] Cf. the fuller treatment of this question in GS ch. ix.

[22] H. G. Wells writes (Soul of a Bishop, 94): “He was lugging things now into speech that so far had been scarcely above the threshold of his conscious thought.” Here we see the wrong interpretation of the preposition over dragging with it the synonym above.

[23]

[24] This is not the place to speak of the way in which prevalent methods of teaching foreign languages can be improved. A slavish copying of the manner in which English children learn English is impracticable, and if it were practicable it would demand more time than anyone can devote to the purpose. One has to make the most of the advantages which the pupils possess over babies, thus, their being able to read, their power of more sustained attention, etc. Phonetic explanation of the new sounds and phonetic transcription have done wonders to overcome difficulties of pronunciation. But in other respects it is possible to some extent to assimilate the teaching of a foreign language to the method pursued by the child in its first years: one should not merely sprinkle the pupil, but plunge him right down into the sea of language and enable him to swim by himself as soon as possible, relying on the fact that a great deal will arrange itself in the brain without the inculcation of too many special rules and explanations. For details I may refer to my book, How to Teach a Foreign Language (London, George Allen and Unwin).

[25] Hence, also, the second or third child in a family will, as a rule, learn to speak more rapidly than the eldest.

[26] I translate this from Ido, see The International Language, May 1912.

[27] I have collected a bibliographical list of such ‘secret languages’ in Nord. Tidsskrift f. Filologi, 4r. vol. 5.

[28] I subjoin a few additional examples. Basque aita ‘father,’ ama ‘mother,’ anaya ‘brother’ (Zeitsch. f. rom. Phil. 17, 146). Manchu ama ‘father,’ eme ‘mother’ (the vowel relation as in haha ‘man,’ hehe ‘woman,’ Gabelentz, S 389). Kutenai pa· ‘brother’s daughter,’ papa ‘grandmother (said by male), grandfather, grandson,’ pat! ‘nephew,’ ma ‘mother,’ nana ‘younger sister’ (of girl), alnana ‘sisters,’ tite ‘mother-in-law,’ titu ‘father’ (of male)—(Boas, Kutenai Tales, Bureau of Am. Ethnol. 59, 1918). Cf. also Sapir, “Kinship Terms of the Kootenay Indians” (Amer. Anthropologist, vol. 20). In the same writer’s Yana Terms of Relationship (Univ. of California, 1918) there seems to be very little from this source.

[29] Tata is also used for ‘a walk’ (to go out for a ta-ta, or to go out ta-tas) and for ‘a hat’—meanings that may very well have developed from the child’s saying these syllables when going out or preparing to go out.

[30] The Swede Bolin says that his child said tatt-tatt, which he interprets as tack, even when handing something to others.

[31] The views advanced in § 8 have some points in contact with the remarks found in Stern’s ch. xix, p. 300, only that I lay more stress on the arbitrary interpretation of the child’s meaningless syllables on the part of the grown-ups, and that I cannot approve his theory of the m syllables as ‘centripetal’ and the p syllables as ‘centrifugal affective-volitional natural sounds.’ Paul (P § 127) says that the nursery-language with its bowwow, papa, mama, etc., “is not the invention of the children; it is handed over to them just as any other language”; he overlooks the share children have themselves in these words, or in some of them; nor are they, as he says, formed by the grown-ups with a purely pedagogical purpose. Nor can I find that Wundt’s chapter “Angebliche worterfindung des kindes” (S 1. 273-287) contains decisive arguments. Curtius (K 88) thinks that Gr. pater was first shortened into p and this then extended into pÁppa—but certainly it is rather the other way round.

[32] The same inconsistency is found in Dauzat, who in 1910 thought that nothing, and in 1912 that nearly everything, was due to imperfect imitation by the child (V 22 ff., Ph 53, cf. 3). Wechssler (L p. 86) quotes passages from Bremer, Passy, Rousselot and WallenskÖld, in which the chief cause of sound changes is attributed to the child; to these might be added Storm (Phonetische Studien, 5. 200) and A. Thomson (IF 24, 1909, p. 9), probably also Grammont (MÉl. linguist. 61). Many writers seem to imagine that the question is settled when they are able to adduce a certain number of parallel changes in the pronunciation of some child and in the historical evolution of languages.

[33] See E. Herzog, Streitfragen der roman. philologie, i. (1904), p. 57—I modify his symbols a little.

[34] In Russian Marfa, Fyodor, etc., we also have f corresponding to original Þ, but in this case it is not a transition within one and the same language, but an imperfect imitation on the part of the (adult!) Russians of a sound in a foreign language (Greek th) which was not found in their own language.

[35] Reduplications and assimilations at a distance, as in Fr. tante from the older ante (whence E. aunt, from Lat. amita) and porpentine (frequent in this and analogous forms in Elizabethan writers) for porcupine (porkepine, porkespine) are different from the ordinary assimilations of neighbouring sounds in occurring much less frequently in the speech of adults than in children; cf., however, below, Ch. XV 4.

[36] Karl SundÉn, in his diligent and painstaking book on Elliptical Words in Modern English (Upsala, 1904) [i.e. clipped proper names, for common names are not treated in the long lists given], mentions only two examples of surnames in which the final part is kept (Bart for Islebart, Piggy for Guineapig, from obscure novels), though he has scores of examples in which the beginning is preserved.

[37] It is often said that stress is decisive of what part is left out in word-clippings, and from an a priori point of view this is what we should expect. But as a matter of fact we find in many instances that syllables with weak stress are preserved, e.g. in Mac(donald), Pen(dennis), the Cri, Vic, Nap, Nat for Nathaniel (orig. pronounced with [t], not [Þ]), Val for Percival, Trix, etc. The middle is never kept as such with omission of the beginning and the ending; Liz (whence Lizzy) has not arisen at one stroke from Elizabeth, but mediately through Eliz. Some of the adults’ clippings originate through abbreviations in writing, thus probably most of the college terms (exam, trig, etc.), thus also journalists’ clippings like ad for advertisement, par for paragraph; cf. also caps for capitals. On stump-words see also below, Ch. XIV, §§ 8 and 9.

[38] See my MEG ii. 5. 6, and my paper on “Subtraktionsdannelser,” in Festskrift til Vilh. Thomsen, 1894, p. 1 ff.

[39] Semantic changes through ambiguous syntactic combinations have recently been studied especially by Carl Collin; see his Semasiologiska studier, 1906, and Le DÉveloppement de Sens du Suffixe -ATA, Lund, 1918, ch. iii and iv. Collin there treats especially of the transition from abstract to concrete nouns; he does not, as I have done above, speak of the rÔle of the younger generation in such changes.

[40] I know perfectly well that in these and in other similar words there were reasons for the original word disappearing as unfit (shortness, possibility of mistakes through similarity with other words, etc.). What interests me here is the fact that the substitute is a word of the nursery.

[41] “Einige namentlich in der Ältern litteratur vorkommende angaben Über kinder, die sich zusammen aufwachsend eine eigene sprache gebildet haben sollen, sind wohl ein fÜr allemal in das gebiet der fabel zu verweisen” (S 1. 286).

[42] Cf. against the assumption of Keltic influence in this instance Meyer-LÜbke, Die Romanischen Sprachen, Kultur der Gegenwart, p. 457, and Ettmayer in Streitberg’s Gesch. 2. 265. H. Mutschmann, Phonology of the North-Eastern Scotch Dialect, 1909, p. 53, thinks that the fronting of u in Scotch is similar to that of Latin u on Gallic territory, and like it is ascribable to the Keltic inhabitants: he forgets, however, that the corresponding fronting is not found in the Keltic spoken in Scotland. Moreover, the complicated Scotch phenomena cannot be compared with the French transition, for the sound of [u] remains in many cases, and [i] generally corresponds to earlier [o], whatever the explanation may be.

[43] Curiously enough, Feist uses this argument himself against Hirt in his earlier paper, PBB 37. 121.

[44] Feist, on the other hand (PBB 36. 329), makes the Kelts responsible for the shift from p to f, because initial p disappears in Keltic: but disappearance is not the same thing as being changed into a spirant, and there is no necessity for assuming that the sound before disappearing had been changed into f. Besides, it is characteristic of the Gothonic shift that it affects all stops equally, without regard to the place of articulation, while the Keltic change affects only the one sound p.

[45] ME. knowleche, stonËs [st?·nes], off, with [wiÞ] become MnE. knowledge, stones [stounz], of [?v, ?v], with [wiÐ], etc.; cf. also possess, discern with [z], exert with [gz], but exercise with [ks]. See my Studier over eng. kasus, 1891, 178 ff., now MEG i. 6. 5 ff., and (for the phonetic explanation) LPh p. 121.

[46] Sharp tenues and aspirated tenues may alternate even in the life of one individual, as I have observed in the case of my own son, who at the age of 1.9 used the sharp French sounds, but five months later substituted strongly aspirated p, t, k, with even stronger aspiration than the usual Danish sounds, which it took him ten or eleven months to learn with perfect certainty.

[47] I use the terms loan-words and borrowed words because they are convenient and firmly established, not because they are exact. There are two essential respects in which linguistic borrowing differs from the borrowing of, say, a knife or money: the lender does not deprive himself of the use of the word any more than if it had not been borrowed by the other party, and the borrower is under no obligation to return the word at any future time. Linguistic ‘borrowing’ is really nothing but imitation, and the only way in which it differs from a child’s imitation of its parents’ speech is that here something is imitated which forms a part of a speech that is not imitated as a whole.

[48] The etymology of this name is rather curious: Portuguese bicho de mar, from bicho ‘worm,’ the name of the sea slug or trepang, which is eaten as a luxury by the Chinese, was in French modified into bÊche de mer, ‘sea-spade’; this by a second popular etymology was made into English beach-la-mar as if a compound of beach.

My sources are H. Schuchardt, KS v. (Wiener Academie, 1883); id. in ESt xiii. 158 ff., 1889; W. Churchill, Beach-la-Mar, the Jargon or Trade Speech of the Western Pacific (Carnegie Institution of Washington, 1911); Jack London, The Cruise of the Snark (Mills & Boon, London, 1911?), G. Landtman in Neuphilologische Mitteilungen (Helsingfors, 1918, p. 62 ff. Landtman calls it “the Pidgin-English of British New Guinea,” where he learnt it, though it really differs from Pidgin-English proper; see below); “The Jargon English of Torres Straits” in Reports of the Cambridge Anthropological Expedition to Torres Straits, vol. iii. p. 251 ff., Cambridge, 1907.

[49] Similarly the missionary G. Brown thought that tobi was a native word of the Duke of York Islands for ‘wash,’ till one day he accidentally discovered that it was their pronunciation of English soap.

[50] There are many specimens in Charles G. Leland, Pidgin-English Sing-Song, or Songs and Stories in the China-English Dialect, with a Vocabulary (5th ed., London, 1900), but they make the impression of being artificially made-up to amuse the readers, and contain a much larger proportion of Chinese words than the rest of my sources would warrant. Besides various articles in newspapers I have used W. Simpson, “China’s Future Place in Philology” (Macmillan’s Magazine, November 1873) and Dr. Legge’s article “Pigeon English” in Chambers’s EncyclopÆdia, 1901 (s.v. China). The chapters devoted to Pidgin in Karl Lentzner’s Dictionary of the Slang-English of Australia and of some Mixed Languages (Halle, 1892) give little else but wholesale reprints of passages from some of the sources mentioned above.

[51] See An International Idiom. A Manual of the Oregon Trade Language, or Chinook Jargon, by Horatio Hale (London, 1890). Besides this I have used a Vocabulary of the Jargon or Trade Language of Oregon [by Lionnet] published by the Smithsonian Institution (1853), and George Gibbs, A Dictionary of the Chinook Jargon (Smithsonian Inst., 1863). Lionnet spells the words according to the French fashion, while Gibbs and Hale spell them in the English way. I have given them with the continental values of the vowels in accordance with the indications in Hale’s glossary.

[52] See Martius, Beitr. zur Ethnogr. und Sprachenkunde Amerikas (Leipzig, 1867), i. 364 ff. and ii. 23 ff.

[53] “Ai is the man’s diphthong, and soundeth full: ei, the woman’s, and soundeth finish [i.e. fineish] in the same both sense, and vse, a woman is deintie, and feinteth soon, the man fainteth not bycause he is nothing daintie.” Thus what is now distinctive of refined as opposed to vulgar pronunciation was then characteristic of the fair sex.

[54] There are great differences with regard to swearing between different nations; but I think that in those countries and in those circles in which swearing is common it is found much more extensively among men than among women: this at any rate is true of Denmark. There is, however, a general social movement against swearing, and now there are many men who never swear. A friend writes to me: “The best English men hardly swear at all.... I imagine some of our fashionable women now swear as much as the men they consort with.”

[55] “OÙ femme y a, silence n’y a.” “Deux femmes font un plaid, trois un grand caquet, quatre un plein marchÉ.” “Due donne e un’ oca fanno una fiera” (Venice). “The tongue is the sword of a woman, and she never lets it become rusty” (China). “The North Sea will sooner be found wanting in water than a woman at a loss for a word” (Jutland).

[56] The uniformity in the speech of the whole Roman Empire during the first centuries of our Christian era was kept up, among other things, through the habit of removing soldiers and officials from one country to the other. This ceased later, each district being left to shift more or less for itself.

[57] “Dass unsere Ältesten vorfahren sich das sprechen erstaunlich unbequem gemacht haben,” DelbrÜck, E 155.

[58] Sometimes appearances may be deceptive: when [nr, mr] become [ndr, mbr], it looks on the paper as if something had been added and as if the transition therefore militated against the principle of ease: in reality, the old and the new combinations require exactly the same amount of muscular activity, and the change simply consists in want of precision in the movement of the velum palati, which comes a fraction of a second too soon. If anything, the new group is a trifle easier than the old. See LPh 5. 6 for explanation and examples (E. thunder from Þunor sb., Þunrian vb.; timber, cf. Goth. timrian, G. zimmer, etc.).

[59] This is rendered most clear by my ‘analphabetic’ notation (a means lips, tip of tongue, d soft palate, velum palati, and e glottis; 0 stands for closed position, 1 for approximation, 3 for open position); the three sound combinations are thus analysed (cf. my Lehrbuch der Phonetik):

[60] The only clear cases of saving of time are those in which long sounds are shortened, and even they must be looked upon as a saving of effort.

[61] In the reprint in Samlede Afhandlinger, ii. 417 (1920), a few lines are added in which Thomsen fully accepts the explanation which I gave as far back as 1886.

[62] The above remarks are condensed from the argument in ChE 38 ff. Note also what is said below (Ch. XIX § 13) on the loss of Lat. final -s in the Romanic languages after it had ceased to be necessary for the grammatical understanding of sentences.

[63] Against this it has been urged that Fr. oncle has not preserved the stem syllable of Lat. avunculus particularly well. But this objection is a little misleading. It is quite true that at the time when the word was first framed the syllable av- contained the main idea and -unculus was only added to impart an endearing modification to that idea (‘dear little uncle’); but after some time the semantic relation was altered; avus itself passed out of use, while avunculus was handed down from generation to generation as a ready-made whole, in which the ordinary speaker was totally unable to suspect that av- was the really significative stem. He consequently treated it exactly as any other polysyllable of the same structure, and avun- (phonetically [awu?, auu?]) was naturally made into one syllable. Nothing, of course, can be protected by a sense of its significance unless it is still felt as significant. That hardly needs saying.

[64] Compare also the results of the same principle seen in writing. In a letter a proper name or technical term when first introduced is probably written in full and very distinctly, while afterwards it is either written carelessly or indicated by a mere initial. Any shorthand-writer knows how to utilize this principle systematically.

[65] “His pronunciation of some words is so distinct that an idea crossed me once that he might be an actor” (Shaw, Cashel Byron’s Profession, 66).

[66] Dickens, D. Cop. 2. 149 neverberrer, 150 I’mafraid you’renorwell (ib. also r for n: Amigoarawaysoo, Goori = Good night). " Our Mut. Fr. 602 lerrers. " Thackeray, Newc. 163 Whas that? " Anstey, Vice V. 328 shupper, I shpose, wharriplease, say tharragain. " Meredith, R. Feverel 272 Nor a bir of it. " Walpole, Duch. of Wrex. 323-4 nonshensh, Wash the matter? " Galsworthy, In Chanc. 17 cursh, unshtood’m. Cf. also Fijn van Draat, ESt 34. 363 ff.

[67] The inconveniences arising from having many homophones in a language are eloquently set forth by Robert Bridges, On English Homophones (S.P.E., Oxford, 1919)—but I would not subscribe to all the Laureate’s views, least of all to his practical suggestions and to his unjustifiable attacks on some very meritorious English phoneticians. He seems also to exaggerate the dangers, e.g. of the two words know and no having the same sound, when he says (p. 22) that unless a vowel like that in law be restored to the negative no, “I should judge that the verb to know is doomed. The third person singular of its present tense is nose, and its past tense is new, and the whole inconvenience is too radical and perpetual to be received all over the world.” But surely the rÔle of these words in connected speech is so different, and is nearly always made so clear by the context, that it is very difficult to imagine real sentences in which there would be any serious change of mistaking know for no, or knows for nose, or knew for new. I repeat: it is not homophony as such—the phenomenon shown in the long lists lexicographers can draw up of words of the same sound—that is decisive, but the chances of mistakes in connected speech. It has been disputed whether the loss of Gr. humeÎs, ‘ye,’ was due to its identity in sound with hemeÎs, ‘we’; Hatzidakis says that the new formation eseÎs is earlier than the falling together of e and u [y] in the sound [i]. But according to Dieterich and C. D. Buck (Classical Philology, 9. 90, 1914) the confusion of u and i or e dates back to the second century. Anyhow, all confusion is now obviated, for both the first and the second persons pl. have new forms which are unambiguous: emeÎs and eseÎs or seÎs.

[68] The NED has not arrived at this explanation; it says: “Peer is not a phonetic development of pire, and cannot, so far as is at present known, be formally identified with that word”; “the verbs keek, peek, and peep are app. closely allied to each other. Kike and pike, as earlier forms of keek and peek, occur in Chaucer; pepe, peep is of later appearance.... The phonetic relations between the forms pike, peek, peak, are as yet unexplained.”

[69] See, for instance, the following strong expressions: “Une langue est sans cesse rongÉe et menacÉe de ruine par l’action des lois phonÉtiques, qui, livrÉes À elles-mÊmes, opÉreraient avec une rÉgularitÉ fatale et dÉsagrÉgeraient le systÈme grammatical.... Heureusement l’analogie (c’est ainsi qu’on dÉsigne la tendance inconsciente À conserver ou recrÉer ce que les lois phonÉtiques menacent ou dÉtruisent) a peu À peu effacÉ ces diffÉrences ... il s’agit d’une perpÉtuelle dÉgradation due aux changements phonÉtiques aveugles, et qui est toujours ou prÉvenue ou rÉparÉe par une rÉorganisation parallÈle du systÈme” (Bally, LV 44 f.).

[70] Some speakers will say [su·] in Susan, supreme, superstition, but will take care to pronounce [sju·] in suit, sue. Others are more consistent one way or the other.

[71] I.e. “With confused and indistinct articulation; also, with a husky or hoarse voice”—NED.

[72] Even in speaking a foreign language one may unconsciously apply phonetic correspondences; a countryman of mine thus told me that he once, in his anger at being charged an exorbitant price for something, exclaimed: “Das sind doch unblaue preise!”—coining in the hurry the word unblaue for the Danish ublu (shameless), because the negative prefix un- corresponds to Dan. u-, and au very often stands in German where Dan. has u (haus = hus, etc.). On hearing his own words, however, he immediately saw his mistake and burst out laughing.

[73] With regard to Lat. signum it should be noted that it is by others explained as coming from Lat. secare and as meaning a notch.

[74] It is, of course, impossible to say how great a proportion of the etymologies given in dictionaries should strictly be classed under each of the following heads: (1) certain, (2) probable, (3) possible, (4) improbable, (5) impossible—but I am afraid the first two classes would be the least numerous. Meillet (Gr 59) has some excellent remarks to the same effect; according to him, “pour une Étymologie sÛre, les dictionnaires en offrent plus de dix qui sont douteuses et dont, en appliquant une mÉthode rigoureuse, on ne saurait faire la preuve.”

[75] Westphalian also has hoppen ‘zurÜckweichen,’ ESt. 54. 88.

[76] Lewis Carrol’s ‘portmanteau words’ are, of course, famous.

[77] Speculation has been rife, but without any generally accepted results, as to the relation between plumbum and words for the same metal in cognate languages: Gr. molibos, molubdos and similar forms, Ir. luaide, E. lead (G. lot, ‘plummet, half an ounce’), Scand. bly, OSlav. olovo, OPruss. alwis; see Curtius, Prellwitz, Boisacq, Hirt Idg. 686, Schrader Sprachvergl. u. Urgesch., 3d. ed., ii. 1. 95; Herm. MÖller, Sml. Glossar 87, says that molibos and plumbum are extensions of the root m-l ‘mollis esse’ and explains the difference between the initial sounds by referring to multum: comp. plus—certainly most ingenious, but not convincing. Some of these words may originally have been echo-words for the plumping plummet.

[78] I have discussed this more in detail and added other m-words of a somewhat related character in Studier tillegnade E. TegnÉr, 1918, p. 49 ff.

[79] Quoted here from John Wilkins, An Essay towards a Real Character and a Philosophical Language, 1668, p. 448: Wilkins there subjects Bacon’s saying to a crushing criticism, laying bare a great many radical deficiencies in Latin to bring out the logical advantages of his own artificial ‘philosophical’ language.

[80] Cf. also what Paul says (P 144) about one point in German grammar (strong and weak forms of adjectives): “But the difficulty of the correct maintenance of the distinction is shown in numerous offences made by writers against the rules of grammar”—of course, not only by writers, but by ordinary speakers as well.

[81] It has often been pointed out how Great Britain has ‘blundered’ into creating her world-wide Empire, and Gretton, in The King’s Government (1914), applies the same view to the development of governmental institutions.

[82] In the realm of significations he sees the ‘humanization’ of language exclusively in the development of abstract terms. An important point of disagreement between Baudouin and myself is in regard to morphology, where he sees only ‘oscillations’ in historical times, in which he is unable to discover a continuous movement in any definite direction, while I maintain that languages here manifest a definite progressive tendency.

[83] On the other hand, it is not, perhaps, fair to count the number of syllables, as these may vary very considerably, and some languages favour syllables with heavy consonant groups unknown in other tongues. The most rational measure of length would be to count the numbers of distinct (not sounds, but) articulations of separate speech organs—but that task is at any rate beyond my powers.

[84] Thus also the corresponding Lat. jecur by ficatum, Fr. foie.

[85] This ungainly repetition is frequent in the Latin of Roman law, e.g. Digest. IV. 5. 2, Qui quÆve ... capite diminuti diminutÆ esse dicentur, in eos easve ... iudicium dabo. " XLIII. 30, Qui quÆve in potestate Lucii Titii est, si is eave apud te est, dolove malo tuo factum est quominus apud te esset, ita eum eamve exhibeas. " XI. 3, Qui servum servam alienum alienam recepisse persuasisseve quid ei dicitur dolo malo, quo eum eam deteriorem faceret, in eum, quanti ea res erit, in duplum iudicium dabo. I owe these and some other Latin examples to my late teacher, Dr. O. Siesbye. From French, Nyrop (Kongruens, p. 12) gives some corresponding examples: tous ceux et toutes celles qui, ayant ÉtÉ orphelins, avaient eu une enfance malheureuse (Philippe), and from Old French: Lors donna congiÉ À ceus et À celes que il avoit rescous (Villehardouin).

[86] If instead of omnium veterum I had chosen, for instance, multorum antiquorum, the meaning of masculine gender would have been rendered four times: for languages, especially the older ones, are not distinguished by consistency.

[87] The change of the initial sound of the reminder belonging to the adjective is explained through composition with a ‘relative particle’ a; au becoming o, and ai, e. The numbers within parentheses refer to the numbers of Bleek’s classes. Similar sentences from Tonga are found in Torrend’s Compar. Gr. p. 6 f.

[88] This protecting consonant was dropped in pronunciation at a later period.

[89] Why so? Did sheep and cows also begin with vowels only, adding b and m afterwards to make up their bah and moo?

[90] The examples taken from Gabelentz’s Grammar and an article in Techmer’s Internat. Zeitschrift I.

[91] I must also mention A. Conrady, Eine indochinesische Causativ-denominativ-bildung (Leipzig, 1896), in which Lepsius’s theory is carried a great step further and it is demonstrated with very great learning that many of the tone relations (a well as modifications of initial sounds) of Chinese and kindred languages find their explanation in the previous existence of prefixes which are now extinct, but which can still be pointed out in Tibetan. Though I ought, therefore, to have spoken of prefixes instead of ‘flexional endings’ above, p. 371, the essence of the contention that prehistoric Chinese must have had a polysyllabic and non-isolating structure is thus borne out by the researches of competent specialists in this field.

[92] Madvig Kl 170, Max MÜller L 1. 271, Whitney OLS 1. 283, G 124, Paul P 1st ed. 181, repeated in the following editions, see 4th, 1909, 350 and 347, 349; Brugmann VG 1889, 2. 1 (but in 2nd ed. this has been struck out in favour of hopeless skepticism), Schuchardt, Anlass d. VolapÜks 11, Gabelentz Spr 189, TegnÉr SM 53, Sweet, New Engl. Gr. § 559, Storm, Engl. Phil. 673, Rozwadowski, Wortbildung u. Wortbed., Uhlenbeck, Karakt. d. bask. Gramm. 24, SÜtterlin WGS 1902, 122, Porzezinski, Spr 1910, 229.

[93] Two explanations of this formative element were given by the old school: according to Schleicher C § 290, it was the root ja of the relative pronoun; according to Curtius and others it was the root i ‘to go,’ Greek fer-o-i-mi being analyzed as ‘I go to bear,’ whence, by an easy (?) transition, ‘I should like to bear,’ etc.

[94] Cf. Sommer, Lat. 528, and on Armenian and Tokharian r forms MSL 18. 10 ff. and Feist KI 455. But it must not be overlooked that H. Pedersen (KZ 40. 166 ff.) has revived and strengthened the old theory that r in Italic and Keltic is an original se.

[95] If s was a definite article, why should it be used only with some stems and not with others? Why should neuters never require a definite article?

[96] While it is difficult to see the relation between a demonstrative pronoun or a deictic particle and genitival function, it would be easy enough to understand the latter if we started from a possessive pronoun (ejus, suus), and, curiously enough, we find this very sound s used as a sign for the genitive in two independent languages, starting from that notion. In Indo-Portuguese we have gobernadors casa ‘governor’s house,’ from gobernador su casa (above, Ch. XI § 12, p. 213), and in the South-African ‘Taal’ the usual expression for the genitive is by means of syn, which is generally shortened into se (s) and glued enclitically to the substantive, even to feminines and plurals: Marie-se boek ‘Maria’s book,’ di gowweneur se hond ‘the governor’s dog’ (H. Meyer, Die Sprache der Buren, 1901, p. 40, where also the confusion with the adjective ending -s, in Dutch spelt -sch, is mentioned. For the construction compare G. dem vater sein hut and others from various languages; cf. the appendix on E. Bill Stumps his mark in ChE 182 f.).

[97] Cf. Lloyd George’s speech at Dundee (The Times, July 6, 1917): “The Government will not permit the burdens of the country to be increased by what is called ‘profiteering.’ Although I have been criticized for using that word, I believe on the whole it is a rather good one. It is profit-eer-ing as distinguished from profit-ing. Profiting is fair recompense for services rendered, either in production or distribution; profiteering is an extravagant recompense given for services rendered. I believe that unfair in peace. In war it is an outrage.”

[98] Bleek is here thinking of classes like those of the Bantu languages, which have nothing to do with sex.

[99] For bibliography and criticism see Wheeler in Journ. of Germ. Philol. 2. 528 ff., and especially Josselin de Jong in Tijdschr. v. Ned. Taal- en Letterk. 29. 21 ff., and the same writer’s thesis De Waardeeringsonderscheiding van levend en levenloos in het Indogermaansch vergel. m. hetzelfde verschijnsel in Algonkin-talen (Leiden, 1913). Cf. also Hirt GDS 45 ff.

[100] “Inner and essential connexion between idea and word ... there is none, in any language upon earth,” says Whitney L 32.

[101] I have learnt very little from the discussion which followed Wundt’s remarks on the subject (S 1. 312-347); see DelbrÜck Grfr 78 ff., SÜtterlin WSG 29 ff., Hilmer Sch 10 ff.

[102] Schuchardt, KS 5. 12, Zs. f. rom. Phil. 33. 458, Churchill B 53, Sandfeld-Jensen, NationalfØlelsen 14, Lentzner, Col. 87, Simonyi US 157, The Outlook, January 1910, New Quarterly Mag., July 1879.

[103] F, for instance, in fop, foozy, fogy, fogram (old), all of them more or less variants of fool.

[104] The preceding paragraphs on the symbolic value of i are an abstract of a paper which will be printed in Philologica, vol. i.

[105] Benfey Gesch 791, Misteli 539, Wundt S 1. 331 (but his examples from out-of-the-way languages must be used with caution, and curiously enough he thinks that the phenomenon is limited to primitive languages and is not found in Semitic or Aryan languages), GRM 1. 638, Simonyi US 255, Meinhof, Ham 20.

[106] I must confess that I find nothing symbolical in glas and very little in fouet (though the verb fouetter has something of the force of E. whip). On the whole, much of what people ‘hear’ in a word appears to me fanciful and apt to discredit reasonable attempts at gaining an insight into the essence of sound symbolism; thus E. Lerch’s ridiculous remark on G. loch in GRM 7. 101: “loch malt die bewegung, die der anblick eines solchen im beschauer auslÖst, durch eine entsprechende bewegung der sprachwerkzeuge, beginnend mit der liquida zur bezeichnung der rundung und endend mit dem gutturalen ch tief hinten in der gurgel.”

[107] It may not be superfluous expressly to point out that there is no contradiction between what is said here on the disappearance of tones and the remarks made above (Ch. XIX § 4) on Chinese tones. There the change wrought in the meaning of a word by a mere change of tone was explained on the principle that the difference of meaning was at an earlier stage expressed by affixes, the tone that is now concentrated on one syllable belonging formerly to two syllables or perhaps more. But this evidently presupposes that each syllable had already some tone of its own—and that is what in this chapter is taken to be the primitive state. Word-tones were originally frequent, but meaningless; afterwards they were dropped in some languages, while in others they were utilized for sense-distinguishing purposes.

[108] On the lack of abstract and general terms in savage languages, see also Ginneken LP 108 and the works there quoted.

[109] Of course, if instead of look upon and outcome we had taken the corresponding terms of Latin root, consider and result, the metaphors would have been still more dead to the natural linguistic instinct.

[110] From the experience I had with my previous book, Progress, from which this chapter has, with some alterations and amplifications, passed into this volume, I feel impelled here to warn those critics who do me the honour to mention my theory of the origin of language, not to look upon it as if it were contained simply in my remarks on primitive love-songs, etc., and as if it were based on a priori considerations, like the older speculative theories. What I may perhaps claim as my original contribution to the solution of this question is the inductive method based on the three sources of information indicated on p. 416, and especially on the ‘backward’ consideration of the history of language. Some critics think they have demolished my view by simply representing it as a romantic dream of a primitive golden age in which men had no occupation but courting and singing. I have never believed in a far-off golden age, but rather incline to believe in a progressive movement from a very raw and barbarous age to something better, though it must be said that our own age, with its national wars, world wars and class wars, makes one sometimes ashamed to think how little progress our so-called civilization has made. But primitive ages were probably still worse, and the only thing I have felt bold enough to maintain is that in those days there were some moments consecrated to youthful hilarity, and that this gave rise, among other merriment, to vocal play of such a character as closely to resemble what we may infer from the known facts of linguistic history to have been a stage of language earlier than any of those accessible to us. There is no ‘romanticism’ (in a bad sense) in such a theory, and it can only be refuted by showing that the view of language and its development on which it is based is erroneous from beginning to end.