do not exhibit development, but decline, and since he admits occasional (although much rarer) exceptions to the prevailing phonetic laws.

But there is a very considerable difference in degree between the phonetic systems of the two scholars. What a stately appearance SCHLEICHER's Lautlehre presents, occupying, as it does, half of the whole Compendium, compared with BOPP's scanty and unevenly written chapter, which bears the title "System of writing and phonetics" [Schrift- und Lautsystem! It was SCHLEICHER's task to sift down and turn to account the great mass of detailed investigations which had been undertaken since BOPP's time by POTT, BENFEY, KUHN, CURTIUS and others. In his treatment of the subject we can observe the progress intimated above. The differences of the separate languages are taken into account, all related instances are carefully placed side by side, and the probability of each single instance measured from the result obtained. Thus SCHLEICHER established a long series of carefully weighed, well-grounded phonetic laws, which were destined to serve. as a regulating principle for every etymologist, and he has undeniably won great credit by this task of sifting and arranging.

Of course all such laws have only a provisional value. For since obvious etymologies form the material from which the phonetic laws are derived, and this material can perpetually increase and change, it is always possible that new phonetic laws should be recognized, or old ones transformed. This idea, whose correctness has been amply confirmed by experience (for how much that is new has been discovered by FICK alone!) was not sufficiently appreciated by SCHLEICHER. This was probably owing to the fact that he himself, with his methodical mind, had no conception of that combining fancy which is indispensable to the discovery of new etymologies, and therefore undervalued etymologizing in general.

We conclude from the foregoing that in all important points which have hitherto been mentioned, the difference between BOPP and SCHLEICHER cannot be called a difference of principle. But one point still remains, which at all events brings SCHLEICHER's originality into the clearest light, - I

refer to the reconstructed Indo-European parent speech [Ursprache). I find the earliest mention of this parent speech in the preface to the Formenlehre der kirchenslawischen Sprache, where we read:

"In comparing the linguistic forms of two kindred languages, I try first of all to trace back both the compared forms to their probable fundamental from, i. e., the shape which they ought to have, leaving out of account the later changes; or at any rate, to bring them upon the same phonetic plane. Now since the oldest languages of our family (even the Sanskrit) do not exhibit their oldest phonetic form, and since the different languages are known to us in very different degrees age, this difference in age must first be eliminated as far as possible, before there can be any comparison; the given quantities must be reduced to common terms before we can compare them, whether the expression thus obtained be the oldest form which can be deduced for both the compared languages, or the oldest form of one of them."

of

Hence, in comparing two languages, we can either reduce the form of one language to that of the other (e. g., Slavonic pekąsta to a Sanskrit *pacantyasya, - v. the work quoted above), or trace both forms back to a common primitive form. The first method, so far as I can see, has very seldom been actually applied by SCHLEICHER; the second, on the other hand, if for "the comparison of two languages" we substitute the words "comparison of all Indo-European languages", contains the following rule for the construction of fundamental IndoEuropean forms: from a form which appears in all languages, subtract all that is due to the special development of the individual languages, and what remains will be the primitive form. An example will make these directions clear. "Field" in Sanskrit is ájras, in Greek ȧypós, in Latin ager, in Gothic akrs. Now we know that in Gothic g has become k, and that an a was lost before the s; thus we obtain from the Gothic the primitive form agras. We know further that the Greek o is derived from a, so that we likewise obtain agras, and so with each language in turn. Hence agras may be regarded as the primitive form, and by a similar process we deduce the accusative agram, genitive agrasya, ablative agrāt,

nominative plural agrāsas etc., as well as a large number of pronouns, prepositions etc. All these forms together make up the Indo-European parent speech; or, expressed in historical style: the parent speech is the language which was spoken immediately before the first separation of the primitive IndoEuropean race.

SCHLEICHER did not always content himself with this simple and clear notion of the parent speech, for he often ascribes to it a quality which cannot be derived from the previous definition, the quality of complete integrity of its original structure. An example will best explain what is meant. The nominative of the word for "mother" is in Sanskrit mātā, in Greek unp, in Lithuanian motë, in Old Slavonic mati, in Old High German muoter. Nowhere does an s appear in the nominative. Accordingly, by comparing the separate forms we can only obtain the form mātār or mātā (the latter if we assume that the r, as for example in up, was in the individual languages introduced into the nominative from the oblique cases), but not the form mātars, as SCHLEICHER does. He assumed this form because matar is the stem, and s the suffix of the nominative, and he felt convinced that in the parent speech so-called "phonetic laws", mutual influences of sounds, and similar phenomena, were not yet in existence. But this supposition is quite arbitrary, for if the primitive speech was spoken by human beings, it must have shared the fate of all language, viz., change in phonetic and morphological constitution. There is, then, no reason why we should not ascribe to the parent speech forms like mātār or mātā. It is true that in a still older period the form may have been mātars, as SCHLEICHER assumes, but then it would be necessary to distinguish the different periods of the primitive language, so that we should not put older and younger forms upon the same plane, as SCHLEICHER seems to have done. The failure to make this distinction has undeniably introduced a certain ambiguity into SCHLEICHER'S notion of the parent speech. In the following discussion I venture to leave this difficulty out of account, and will understand the term "parent speech" only in the sense previously defined, i. e. in the sense originally intended by SCHLEICHER.

DELBRÜCK, Introduction to the Study of Language.

4

Is it, now, SCHLEICHER's opinion that a historical reality must be ascribed to the forms of the parent speech, taken in this sense? I think the reader of the Compendium will be inclined to answer this question in the affirmative, and will be somewhat surprised to find in the appendices (Chrestomathie, page 342) the following remark: "The assumption of these primitive forms does not necessarily imply the assertion that they ever possessed an actual existence." In order to explain this apparent contradiction I will choose the form of independent discussion, proceeding in the following order: I will first formulate the objections which have been brought against SCHLEICHER'S parent speech (in the sense described above), and then try to determine their true value.

The first difficulty is naturally due to the demand that in the case of a certain form each individual language shall be consulted. This demand, however, can be satisfied only in the rarest instances, for how few are the words and forms which we can trace through all the languages! But in actual practice this objection has little weight. For we must consider that it is possible to point out in all languages quite a number of inflectional suffixes, or at least traces of them; and since we know the phonetic laws which would come in play, we can say in the case of a good many word-stems what their form must have been in a given individual language.

A second objection is of a more serious nature. Is it really possible to define the point where the development of each separate language began? Can we determine with certainty whether a certain modification of sound or form belonged to the primitive speech or originated in the individual language? SCHLEICHER had definite views on this point. For instance, he thought it possible to assert that the parent speech possessed the following sounds :

How did he reach this conclusion? In separate fields the way had been prepared for him, as in the field of the a-vowels. It is well known that the Indo-Iranian group of the IndoEuropean possesses no è and o, but exhibits an à where the other languages have these vowels. Bopp was only at first of the opinion that ĕ and ŏ originally belonged also to the Sanskrit, and were subsequently lost; he then adopted the view of GRIMM (Grammar, I, 2nd edition, page 594), who in connection with his Gothic researches denied that è and ŏ were original, so that for the Indo-European there remained three simple primitive vowels, a, i, u. This assumption also gained favor in consequence of the high esteem which the number three is wont to enjoy; POTT, for instance, begins the section on vowels in his Etymologische Forschungen with the remark: "It seems to follow from historical and physico-philosophical grounds that language possesses but three simple fundamental vowel sounds, viz., ă, i, u." Thus the hypothesis of GRIMM seemed to find confirmation in all directions, and was accepted by SCHLEICHER. He assumed that the primitive language agreed with the Sanskrit in the simplicity of its vocalism, while the more diversified Greek exhibited a condition of

greater development or deterioration. For the consonants, however, an opposite conclusion was reached. The cerebrals of Sanskrit had been early regarded with suspicion, the assumption being made that the Hindus had obtained these extraordinary sounds from barbarian aborigines; the palatals also were found in many cases to be younger than the gutturals, as for instance in reduplication (cakára from kar). In this point, therefore, the Greek appeared to have preserved the original condition of things, while the Sanskrit had deteriorated, and the main conclusion was that the rich and diversified phonetic material, which the individual languages either exhibit or must have once exhibited, arose by means of various processes of division and multiplication out of a limited and simple phonetic material in the primitive speech. From the analogy of this result SCHLEICHER drew the further conclusion that the phonetic condition at a still earlier period must have been yet more simple:

"At an earlier period in the life of the Indo-European