Function )f pitch. LANGUAGES.] It is obvious that through these rules the existence of I, u, ~, I, n cannot be explained, and yet they do exist. Osthoff has suggested the explanation that they represent intermediate stages of shorten ing between the full diphthongs and the short i, u, &c., which were sometimes kept under the influence of a sort of half-stress. 1 They may just as well be subsequent lengthenings of the shorts due to some reason as yet unknown ; but this whole chapter is still very obscure, and it may be doubted if the point will ever be sufficiently elucidated. The principle of explanation by presence or absence of stress in " roots " is also applicable to derivative or inflexional syllables. It s evident that forms like the Greek irbXeis (for *7r6e/es) TroXis, or yXvKfTs (for *yvKtFes) yXu/cvs, or irar^pa, Trartpes Trarpda-i warpOiv follow the same rule as XeiTrw XnTeiv, <pevyu <pvyeiv, 5ep/co/zcu HdpaKov, &c. But analogy and change of stress from one syllable to another (which even in root -syllables have often somewhat obscured the original state of things) have done much to render the working of the old laws indistinct, so that no more than this short hint can be given here. There are yet other interchanges of vowels in Aryan, quite as important as those which find their explanation in presence or absence of stress, which do not seem to fall under the principle applied here. Amongst these the change of e and o or e and o, both in roots and derivative syllables, is the most frequent. Thus we have in Greek Xeyu etXo%a, Xoyos ; XetTrw foiira, Xoiiros ; eXei - oojUtu eir)ou0a ; 5epKOfj.ai SedopKa ; rpewu rerpotpa ; prjyvvfu. (ppuya ; or t>yo-s ye ; 7^0? yevtos (for *yvecr-os) ; <ptpo-fj.fi> <j)^pe-re ; irarrip EiVdrwp, (ppdrtap ; Troi/Jt,r)v S.Kfj.(av, &c. It is abso lutely incredible that difference of stress could have changed either c into o, or o into c ; for the greater or less effort in pronouncing a vowel can have nothing to do with the quality of the vowel uttered, as vowel-quality is only regulated by the position of the tongue and lips. If, then, any distinguishing principle in the utterance of human speech governs these changes and that assumption is inevitable it must have been difference of pitch. This explana tion was suggested independently by Fick and M oiler 2 some years ago, but has not found its due share of attention, although it re commends itself both upon physiological and philological grounds. There is a natural physiological connexion between the palatal c and high pitch and between the guttural o and low pitch ; for in uttering a high tone we generally raise the larynx above its normal level, and consequently push the tongue forward with it towards a more palatal position ; for a low tone the larynx is lowered, and the tongue follows this movement by sliding backwards, that is, towards the position of the guttural vowels (as can easily be observed in singing the vowel a on different notes). On the other hand, we know that in Sanskrit the stress syllables were uttered in a high tone (uddtta), and regularly followed by a low -pitch syllable (svarita). This combination of high tone + low tone again corresponds with the sequel of c + o observable in a great many types of Aryan words or forms, such as eyw, yevos, deSopxa. ; compare also Evwdrup against Trar^p, dt-/xwr against troi/j.rii , &c. So far this theory seems very probable ; yet several difficulties still remain. In the first place, the additional hypothesis must be made, that not all "accentuated, " that is stressed, syllables had the high tone ; if o is the characteristic vowel of low-pitch syllables, words like XOYOS, <popos must have had low pitch on their first syllable, while the e of Xeyw, <pepu was uttered with the high tone. Strange as such an accentuation might sound to English or German ears, it involves no practical difficulty ; for there are at least some living Aryan idioms which possess similar distinctions : in Servian, for instance, the nominative vodd is pronounced with a high rising tone on the first and a falling tone on the second, the stress being nearly equally divided between the two syllables ; the accusative vbdu, again, has a well-marked stress on the first syllable, but is pronounced in a low falling tone. 3 In the second place, this theory requires a supplementary inquiry into the relations of pitch and stress in Aryan, for it seems evident that stress and high pitch did not always go together. That the reduplicated perfects like the Sanskrit daddrqa, Greek Sddopxe, for instance, originally had the stress on the root-syllable is certain from the evidence of Germanic, yet that same root-syllable has the low-pitch vowel o, while the unstressed reduplicative syllable shows the high -pitch vowel c. The original pronunciation of Aryan dcdorke, therefore, must have been something like ( * if, while afterwards the stress was w>r attracted by the high-tone syllable in Greek and the high tone by the old stress-syllable in Sanskrit. In this direction the investiga tions of Fick and Holier cannot be considered more than an open ing of the field for further research ; and the same must be said of what has been done hitherto with regard to an explanation of other vowel-changes of a similar character. 1 Mnrpholoyische Untersvchitngen, vol. iv., which treats of the Aryan I and u. 2 Fick, in Gottinger gelehrte Anzeigen, 1880, i. p. 417 stj., and Mb ller, in Paul and Braune, Beitrage, vii. p. 482 .^7. 3 See L. Masinj*, " Die Hauptfonnen des serbisch-chorwatischen Accents," in Mem. Acad. Imp. des Sciences, vol. xxiii., St Petersburg, 1876. 789 B. Grammatical Structure. A few short remarks must suffice here, as a full characteristic of Aryan morphology cannot be given without entering into a mass of more or less minute details. Since the days of Bopp comparative philologists have on the Deriva- whole accepted the theory of the old Sanskrit grammarians, that tion. all Indo-Gennanic words and forms must be traced back to simple, no longer divisible, monosyllabic elements, which have been called roots. We cannot undertake here to discuss the question how far this theory, which has never been uncontested and is beginning to be doubted more and more, is historically correct. However, so much may be conceded that, after removing all the elements which seem to serve in the formation of single words or forms, or the formation of groups of such only in contrast with the whole mass of a system of cognate words or forms, there generally remains a monosyllable, which for practical purposes we may take as a philo logical starting-point, without asking whether these preparations of the philological laboratory ever had an actual existence of their own or whether they are mere abstractions. The general means by which words and forms are derived from these "roots," or from other ready-made words and forms, are partly external, partly internal. On the whole, Indo-Germanic derivation and inflexion, looked at from this point of view, are based on a system of suffixes, that is, individualizing formative elements added at the end of less compound and less individualized formations. Infixes instead of suffixes occur only by exception, the chief instance being the insertion of a nasal, especially in certain verbal formations (as in Latin ju-n-go against jugum, Greek a-fj.- jSdvu against ZXafiov, San skrit yu-nd-jmi, yu-n-jmds against yugdm). The third external clement we meet is reduplication. Prefixes in the proper sense do not seem to occur ; even the verbal augment, which is the only case of an apparently real prefix, most likely was once an inde pendent word, so that augmentation must be reckoned among the numerous cases^of composition. As means of internal change we may mention the shifting of stress and pitch over the different syllables of words and forms, and the vowel-changes which, as we have seen, originally followed these variations of accent, yet may soon have become independent formative principles. As to inflexion, Indo-Germanic is known to hold the foremost Inflexion rank among all inflective languages. The distinction of nouns, of noun, pronouns,, and verbs is fully developed. In the nouns the intro duction in the substantives of grammatical gender is especially noteworthy. Substantives and adjectives were inflected in the same way, though some of the individual languages have deviated from this rule ; the pronouns, at least, in many cases had their own inflexions ; otherwise they agree with the nouns in the distinc tion of numbers and cases. There were three numbers singular, dual, and plural. The number of original cases cannot be settled with certainty. The highest number we find distinguished in any language is seven nominative, accusative, genitive, dative, instru mental, and locative (besides the vocative or interjectional case). 1 jut, judging from the fact that the same cases often have different endings in different declensions, one might be inclined to think that once a still greater variety of case-distinctions had existed. The single declensions are distinguished according to the various stem -suffixes immediately preceding the case - endings. The two chief subdivisions accordingly are the declensions of vocalic and consonantal stems. It may be noticed in passing that the so-called i and u stems follow the type of the consonantal declension ; this, however, appears but natural if we consider that the final i and u of these stems most probably are reductions of older diphthongs ending in a semi-vocalic or consonantal element. For declensional distinctions only one of the general external formative principles is used, namely, that of combining ready-made steins with suffixal endings, at the same time expressing case and number. The verb, too, has in like manner its inflexional endings to ex- Inflexion press the distinctions of number and person ; but it also makes use of verb, of all the other formative principles, both internal and external. The shifting of accent and the vowel-changes connected therewith are nowhere more distinctly traceable than in the verb. Besides, we find the use of special suffixes for the distinction of tenses and moods, sometimes the infixion of a nasal in the formation of tense- stems, then again on a larger scale the use of reduplication, and lastly, the use of the augment as a common sign for the different tenses of the past. None of the individual languages seems to have preserved the original stock of Aryan verbal forms to its full extent. The oldest Sanskrit seems to come nearest to Aryan. Greek has also been very conservative in one way ; it has lost hardly anything that was original, but has, like Latin, created a host of apparently new forms, some of which still continue to baffle all attempts at an explanation. Germanic may serve as a type of the opposite character ; it has lost all but the old present and the old redupli cated perfect, but supplied the loss by the extensive employment of auxiliaries. The differences thus exhibited by the different languages make it a difficult task to determine which formations belong to the primitive Aryan stock and which were added at later periods. General consent, however, seems to take the follow ing points for settled. Of the three voices distinguished in Greek,
