With the discovery of the importance of Sanskrit (q.v.) in linguistic investigation, and the rise of the science of comparative linguistics (see Pirli.m.00v), etymoh)gy was placed on a sci entific foundation. Its history is connected in separably with the branch of learning of which it forms a part, but its method may be briefly outlined. First and foremost there must be a strict adherence in all etymological investigation to the principles of phonetic law (q.v.). The etymology which fails to conform to these laws must receive overwhelming confirmation from other quarters before it can be regarded as even possible. In the case of loan words phonetic law is apparently violated, and it will frequently happen that a language will have two or more words derived from a single word, one being the regular phonetic development and the other a bor rowed form. In this ease the latter form, known by the French term mot savant, is usually dif ferentiated in meaning from the former. Thus we have in French and English such words as royal and regal, both from the Latin regalis, kingly; the form regal being borrowed directly from the Latin, while royal (cf. French roi, king, from the Latin accusative regent) is the phonetically cor rect form. Loan words may also undergo the regular sound-changes of the language into which they have been adopted. Thus, Latin pondus, pound, appears in Gothic and Anglo-Saxon as peed, with unchanged consonants. but in Old High German it is subject to the action of Grimm's law (q.v.) and becomes Aunt, It is therefore evident that in etymology attention most be paid to the history of words, and some times to the records of the tribes speaking them. Thus, the English wise is akin to the Gothic ne wels, unwise, Old High German iris, New High German weise: but wise is also a doublet of guise, which is the form assumed by iris in the Romance languages, which borrowed the word from their Germanic invaders. If it is true that the same word may assume different. forms in the same language, it is equally true that different. words may become identical in form in a given language. The large class of homonyms in every language is sufficient proof of this. An excellent English example of this phenomenon is sound, which is a conglomerate of four originally dis tinct words — namely, Anglo - Saxon gesund, hearty, Anglo-Saxon send, a body of water. Latin sonus, noise, and Latin subundare, to dive be neath the waves. It is probable that many in stances in which a word shows extraordinary di versity of meanings are to be traced to this pro cess of conglomeration rather than to semasinlogi cal developments. (See SEMASIOLOGY.) Tt is. however, in the tracing of words back through an entire group of cognate languages to a hypo thetical original form, denoted conventionally by an asterisk (*), that etymology finds its prin cipal application. The older etymologists made wild guesses in their primitive investigations, and such etymologies are still made by untrained minds. Thus. Latin delis. god, Old Latin del MR, akin to Sanskrit dere, god. has been connected with English devil, from Greek dafticAos, slanderer, and English god, in addition to the old stock comparison with good, with which the word has no etymological relation, has been equat ed with Sanskrit girt/ha, hidden. It is true that many etymologies which are perfectly sound seem at first sight impossible to those who are not acquainted with phonetic laws and the prin ciples of word-formation. It is also true that many etymologies which are very plausible to students of comparative linguistics are in reality doubtful and accepted only provisionally. Such etymologies may ultimately be discarded. just as the provisional assumptions often accepted by in vestigators in the exact sciences are discarded, if further research shows them to be false.
Etymology may be confined to a specific group of languages or dialects. We may thus speak of Romance etymology, where words in the Ro mance languages are traced back for the most part to folk-Latin- originals (as French nu'ine, self. Old French mcisme, Provencal medesme, Old Provencal smetessme, Spanish »iisino, Italian wedcsimo, from folk-Latin semetipsissimum), Germanic, Celtic, Indo-Iranian etymology, and the like. All these are combined in Indo-Ger manic etymology. Similarly, we may have Se mitic, Dravidian, 1:m10-Altaic, or Polynesian ety mologies, but Indo-Germanic is the most thor oughly systematized of all, and serves as a model for the rest. It must be borne in mind, however,
that accidental resemblance of sound is no proof of etymological kinship. It is, consequently, un scientific to compare, as some have done, Semitic or Dravidian with Indo-Germanic words. The fact, for example, that Latin taurus sounds like Arabic thau•, both meaning bull, or English sheriff (Anglo-Saxon scir-gerefa, shire-reeve) resembles in sound the Arabic sharif, exalted, also used of an official of a city, implies no relationship. Within a language group the same statement holds true. Sanskrit sum:, broth, has no connec tion with English soup, nor are the English verbs drag and draw akin. As an example of etymo logical procedure, we may take the word for ten in the Indo-Germanic languages. Thus we have English ten, Anglo-Saxon tyre, Old Saxon Wan, Icelandic Mr, Gothic taihun, Old High German zchan, New High German zehn, Old Irish deieh, Irish deag, Gaelic deug, Cornish (irk. Breton ,icc, Latin decent (whence the Romance group, Italian died, Spanish diet-, Old French dis, French dix etc.), 'Umbrian dcsen-duf for * deceni•ditf, twelve (ten - two), Greek oltca, Old Church Slavic dcscti, Czech dcsat,i1, Polish dr:icsi2ty, Russian desyati, Lithuanian di.s.-.:itatis, Lettish dcsmit, Old Prussian dessfints, Armenian tom Albanian ojet E, Avesta, dasa, New Persian dah, Afghan /as, Shighni 5is, lis, Sanskrit daiai,. Prakrit, Pali, dam llindi drys, Ma rat hi daloi. A comparison of all these forms, and more which might be added to the list, results in the postulation of a pre-Indo-Germanic form *lickip, to which, in ae• cordance with the sound-laws governing the vari ous divisions of the Indo-Germanic languages, and with reference to the principles of word-for mation (as in the formation in Old Church Slavic, Czech, Polish, Russian, Lithuanian, Let tish, and Old Prussian in the example quoted) the various forms of the numeral ten are referred as to a convenient formula. (See l'unsmoov.) The scope of etymology has liven immensely wid ened by the theories of and root-extensions, and by the doctrine of the dissyl labic base or root (see which have rendered possible the explanation of many words whose derivation had before been unknown.
Consult: Pott, Etymologisrhe P'orsrhangen auf dem Gebiete der indogermanisehen ,`)'praehen ( Det mold, 1859-74) ; Fick, Vergleichendes 11•6rter 1)101 der indogerinanischen Spraehen (3d ed., Gottingen, 1374-76; 4th ed., incomplete, 1890 94) ; Gray, lado-Iranian Phonology (New York, 1902) : Uhlenbeck, Kur.gefasstcs etymologisches der altindische» Spindler? (Amster dam, 1898-99) ; Bartholomai, Aitiranischcs Woe- terbach (Strassburg, 1903) : Hiibschmann, Ety mologie and Loutle/re dcr ossetischen Spraehe (Strassburg. ; Horn, Orundriss dcr neuper sisehen Etymologie (Strassburg, IS93) ; Meyer, Etymologisches lVdrtcrbueh der afboeresisvhen ,yprache ( Strassburg, 1891) ; Hfibschtnann, A rine aischc Grammatik (I. Then, Leipzig, 1895) ; Cur tins, G•andz-iige der yriechisehen Etunioloqie (5th ed., Leipzig, 1879) : Prellwitz, Etymologisehes ll oirterbiich der giuuclmschm Sirache (Gottingen, 1892) : Meyer, Ilaadhuch der grieehisehen Ety iologle (Leipzig. 1901 et seq.) ; Vaniiek, Ety mologisches Wui•te•baeh der latcunisehen Sprache (2d ed., Leipzig, 1881) ; Breal and Rainy, Die• tionnaire etymologique lot in (Paris, 18S5) ; Mik losich, Etymoloyisc1>es 1Vortcrbnch der slaivise/ren (Vienna, 1SS6) ; Uhlenbeck, Kurzge fasstes etymologisehes Wo•terbuch der gotisehen ,S'praehe (2d ed., Amsterdam, 1900) ; Kluge, Ety mologisches Worterbuch der deutschen Nprachc (6th ed., Strassburg, 1399) : Franck, Etymo logisch woordenboek der nederlandsche taal (The Hague, 1884-92) ; Tamm. Etymologisk srensk o•dbok (Stockholm, 1891) : Dietz, Etymologisehes Martel-bitch der romanischen Sprarhen (2d ed., Bomi, 1S61-62) : Forting, Lateinisrh-romanisches it iirterbueh (2d ed., Paderborn, 1901) ; Coelho, Diecionario mannel et ymologieo da lingua port u gesa (Lisbon, IS90) : Brachet, Dietionnaire nwlogique de la langue froneaise (2d yd., Paris, n. d.) ; Scheler, Dietionnaire dYtymologie fran ca ise (Park, 1880 ) ; iil ler, Etymologisehes 1Vorterbuch dcr cnglisehen (2(1 ed., 2 vols., Cothen, 1878) ; Skent, Principles of Eng lish Ehrmology (2 vols., Oxford, ISS7-91) ; Ety mological Dictionary of the English Language (3d ed., Oxford, 189S) : Palmer, Folk-Et yin olog y (London, 1882) Osthoff, Etymologische Parerga (Leipzig, 1901).