The great advance in historical linguistics after the early publications of Grimm, Bopp and others was in knowledge of phonetics. Usually we assume that this increased knowledge clarified historical problems, as in accounting for the "first set of exceptions in the consonant shift" -- the retention of the voiceless stops after fricatives, for example in Gothic ist = Lat. est. But from von Raumer we learn that the influence also went in the other direction, that the problems which arose in historical linguistics led to an increasing need for competence in phonetics. In keeping with this need to move away from the "shuffling of letters," von Raumer set out to arrive at an accurate statement of articulatory phonetics.
The essay presented here recapitulates many of the conclusions presented by von Raumer in his monograph on Aspiration and the Consonant Shift, published when he was 22. In this he attempted to clarify the relationship of the Greek stops to the Germanic; he concluded that ph th kh were aspirates, like the related bh dh gh in Sanskrit. By this clarification he defined with greater precision the variety of sounds that Grimm combined in his aspiratae. He also made the suggestion for which Grassmann later was given credit -- that Sanskrit never shows aspiration in two successive syllables: Aspiration and the Consonant Shift p. 74, § 64: "For since Sanskrit never aspirates two successive syllables, one can assume as the original form of bud' with equal justification b'ud'. Then biudan would simply be the usual transition of b' to b." But this insight of von Raumer's was not noted by his contemporaries. Apparently review of all the relevant examples, as by Grassmann, was necessary to attract the notice of linguists.
In devoting attention to phonetics, von Raumer dealt with the spoken dialects, citing the pronunciation of Low German, Bavarian, Swabian variants. This attention to the dialects was followed by specific concentration on them, as his concern with phonetics led to the definitive treatments of Sievers and Jespersen at the end of the century.
Rudolf von Raumer (1815-1876) had a quiet career, completing his studies at the University of Erlangen and holding a position there until his death. With a chair at a university which was not a center for linguistic study he was somewhat of a loner, who did not participate in the struggles between the traditional grammarians and the more rigorous linguists. Possibly his aloof position also led to the neglect of his writings. Both in content and style, however, they seem more modern than most linguistic works of his time. In spite of his problems in developing a linguistic vocabulary and his occasional faulty reasoning (as in his interpretation of Gothic þ as an aspirate) we consider him one of the important contributors to the developing methodology of historical linguistics.
Through the discoveries of historical linguistic investigaon, the significance of phonetics has been placed in a new light. The more the importance of phonetics becomes recognized, the more apparent becomes the need to understand as clearly and precisely as possible its subject matter, namely the sounds themselves. This understanding, insofar it is in the sphere of direct observation, lies in the area of the physical sciences. Therefore it is highly desirable that important scientists should devote themselves to the study of this subject. Among the many valuable studies, which have been recently undertaken in this area, I intend to dwell only upon those which Johannes Müller and Ernst Brücke have achieved for a determination and arrangement of the sounds of language. Investigations of this type need above all a common solid foundation. And exactly that has been furnished by Brücke's publication1 in a manner as clear as it is accessible. Only after agreement has been reached on such a basis can one discuss the more complex and deep-seated questions of scientific as well as historical investigation of sounds. If in regard to some of these questions I maintain previously proposed findings opposed to Brücke's views, I ask that this not be considered as personal obstinacy. I have subjected my assertions to a renewed careful examination. But the result of this examination has only convinced me anew, that my views concerning aspiration and the sound shift, stated in 1837, are essentially correct. The relationship, however, in which this particular sound development stands to the various types of sound change in general, will be clarified partly by the following treatise, partly by comparison of it with my other linguistic works.
1. The natural-historical determination of sounds must first devote itself solely to the sounds of the present as object of immediate observation. The chief aim of such observation is the manner of utterance of the sounds. Differences, which the ear perceives or believes it perceives, are not to be dismissed. But in the realm of precise natural-historical observation they are only then to be dealt with, when one can establish with certainty their diversity of production.2
2. We distinguish primarily the tones of the human voice and the sounds of human speech. The tones are produced through the vibration of the vocal cords in the glottis; the sounds are produced through the deflection of the exhaled airstream against the organs lying between the epiglottis and the lips. Of the tones produced in the glottis, one distinguishes between loud and soft speech (vox clandestina). Loud speech is produced when we accompany utterance of sound, as far as it is possible, with tones from the vocal cords. Soft speech is produced when we speak without simultaneous sounding of the vocal cords. We accompany soft speech too with a noise, differing from the production of sounds, which we can clearly perceive especially in whispering the vowels.3
3. The sounds are divided into classes on the basis of three different criteria, namely 1) according to the position of the organs, 2) according to the type of air influx, 3) according to the organs, by which they are produced.
4. According to the position of the speech organs, the sounds are divided into 1) those which require for their utterance a complete closure of the organs (stops, literae explosivae), and 2) those which are produced without a complete closure of the organs (continuants, literae continuae4). The latter sounds are divided again into those, which are produced by air passing through such a narrow passage, that the noise of air deflection becomes clearly audible (consonantal continuants, consonantes continuae) and those for which, because of the wideness of the opening this is not the case (vowels).
5. The second criterion of classification is the type of air influx. Usually one divides the consonants into hard and soft, so that, for example, the German p is called hard, the b soft, and likewise ß (in gießen) hard, and the s (in sagen, Wesen) soft. The objection is raised against this division, however, that it is wavering and without clearly fixed limits, for what one person calls hard, another will consider soft. Another criterion has therefore been advanced, which provides a definite limit, namely whether the tone of the voice can be combined with the utterance of a sound or not. Sounds are accordingly divided into voiced and voiceless. The difference is most noticeable in the pronunciation of some continuants, for example, the s and the ß. While one sustains the (so-called soft) s, one can simultaneously produce a singing tone; as soon as one passes to the (hard or sharp) ß, however, the singing tone immediately ceases. This difference, which was known already by the old Indian grammarians, is an excellent criterion, because it replaces an uncertain and indefinable difference for the ear with a certain and verifiable distinction in utterance. But since the tone of the voice only accompanies the sound and is not essential to it, one will do better by seeking the reason why it becomes impossible for certain sounds to combine with the tone of the voice, and then to take this cause as the criterion for distinction. This cause, however, is none other than this: certain sounds are produced by blowing (flare), while others are formed through breathing (halare). But blowing and singing exclude one another; one can convince himself of this at once, if one tries to sustain a singing tone when one passes from breathing to blowing.5 The difference then that one designates by voiceless and voiced6 I would prefer to express by blown (literae flatae) and breathed (literae halatae).7
6. Concerning the division of the sounds according to the organs or the places of articulation, I refer to Brücke's exhaustive presentation; for illustration, however, I wish to give at this point merely a survey of the common stops and spirants (consonantes continuae spirantes) of the general New High German language. I limit myself to these two classes, because primarily they are to be considered in the course of this paper. I append the column for vowels merely in order to indicate the position of the consonantal continuants between the stops and the vowels.
|are not always assignable to individual places of articulation with the same certainty as the consonants.|
|I. Throat sounds
|k||g||ch (in Sache,
|II. Palatal sounds
|ch (in Sichel,
|j (in jeder)|
|III. Cerebral sounds
|sch (in schön)|
|IV. Dental sounds
|t||d||ß (in gießen)||s (in sagen, Wesen)|
|V. Labial sounds
|p||b||f||w² (= French V)
w¹ (the u in Quelle)
This table of the common stops and spirants of the general New High German language conforms to the one given in my paper on aspiration and the sound shift. Only in one single instance do I disagree thereby with Brücke, in reference to the cerebral sibilant (sch in schoen). I have repeated the experiments, which in conjunction with the old Indian grammar, disposed me to assign this sound to the cerebrals, and they have only convinced me again that its place of articulation lies between that of ch (in Sichel, Brücke's X¹) and that of ß (and s). One can convince himself of this fact, if one produces our spirants, one after another with vox clandestina, whether the sequence be from back to front or from front to back. One may first narrow the air passage in one such sequence by beginning with the ch in Sache, passing then to that in Sichel, thereupon to the sch and finally the ß. Immediately upon reaching the ß, one should reurn to the sch. It will be very easy to notice that the place of ariculation of the sch lies somewhat farther back than that of the ß. This is the most usual pronunciation of the NHG sch. The determination of this sound becomes somewhat complicated, however, in that there is a third sibilant besides ß and sch, which is pronounced somewhat farther back than the usual German sch. Sanskrit has this sound in the palatal sibilant श, according to Bopp's designation ʼs. It is produced by approaching the palate with the tongue in the same area, where we pronounce the ch in Sichel. But while we hold the part of the tongue, which lies in front of the place of articulation, as far as possible from the palate in pronouncing ch (in Sichel), in pronouncing the palatal sibilant we must approach the palate with the tongue. Through a gradual transition from the positioning of the organs of articulation of the palatal sibilant (श, s̓) to that of the pure cerebral sibilant (ष = s̒, sh, sch), we obtain an uninterrupted series of traditional sibilants lying between the palatal s̓ and the cerebral s̒. A portion of our fellow-countrymen use these sibilants in place of the pure cerebral s̒ (= sh, sch).
1. In the course of time the words of language have changed their sounds. So much is certain and, moreover, this is one of the most important facts for the history of languages. We ascertain that the sounds of words have changed when we compare the older state of languages with the more recent. The process of the change itself however has not yet been investigated enough. If we penetrate deeper into the darkness which in many ways veils these questions, we find a huge multitude of highly different processes at work. And that is even more troublesome, we find that to isolate these processes becomes even more difficult, because often quite heterogeneous occurrences lead to almost the same result.
2. When the change of languages and especially of language sounds is spoken of, there are almost immediately references to the "spirit of the language" and its wonders. I have no intent whatsoever of deprecating the profundity with which the more recent research distinguishes itself. But I think it is about time that we turn our attentions to reality and its phenomena with clear and impartial minds. When we do, we find that the "spirit of the language" in itself and apart from people does nothing, but rather that all changes in language actually are produced by the people. To just what extent their production is really a product of man remains a matter of conjecture. It is enough that the changes themselves are objects of observation as soon as they become apparent.
3. If first of all we direct our observation to that which happens before our very eyes, or better, before our ears, we will discern the following facts:
1) Every single person changes his speech in the course of his life. As a child, before the complete development of his speaking ability, he speaks many words with sounds, which he later abandons. If he attains an old age and loses his teeth, not only does the sharpness of his articulation disappear, but also in more than one instance real modifications of the previously pronounced sounds become apparent.
2) From this alone it follows that not even a single family, which consists of old people, adults and children, speaks one and the same language.
3) But even the adults among themselves never have exactly the same language, not even phonetically. This follows necessarily from the principle of individuation. Every human being has his own peculiarly formed organs of speech as well as his own particular facial features. Now the production of sounds is conditioned by the form of the speech organs, which confine the sound-producing air stream. Therefore, although our ear does not perceive the resultant difference, it is nevertheless present. But in many cases our ear is very readily able to detect the difference.
4) A further and not infrequently occurring difference results from the fact, that one person articulates a sound at a somewhat different place than another, and therefore, strictly speaking, actually produces an entirely different sound.
4. If we consider the possibilities which could result from the above discussed differences among relatively great numbers of people, we find them to be of most varied kinds. If the change of a sound heard in an individual's speech is caused by the inability of the vocal organs to produce the heard sound, this individual is forced to face the particular change wherever the sound in question occurs. Let us consider then an entire family, or an even larger social group consisting only of individuals which suffer from the foresaid disability. The sound pronounced earlier will necessarily disappear in this entire group, and the other sound will take its place.
On the other hand, however, let us consider a family in which one member, for example the father, has that peculiarity of speech, but the mother does not. The case may then occur that the children imitate either the father entirely or the mother. But it can also be that, being capable of imitating both, the children imitate their father in some words, the mother in others, and in some words perhaps they waver between one parent and the other.
5. If the change of the heard sound is not based on the inability of the speaker to produce the sound, but rather only on the fact, that the changed pronunciation is easier for the speech organs than the traditional one, then usually the results will also be different than in the previously discussed instances. It is possible then that certain members of the group will retain the old pronunciation. But since the change is not due to an individual peculiarity of the speaker, but rather to the mechanism of the human speech organs in general, among the other members of society it will also be effected, not merely through imitation, but also through the structure of their own speech organs. This is the case in most instances in which one sound is altered by the environment of another.
6. A large part of the changes, which the sounds of words undergo in the course of time, can be accounted for in the ways discussed above. Especially if one remembers in addition, that the mere inexactness of hearing and speaking causes sound changes, which are very similar to the four discussed already. We find however another type of alterations belonging to a class of sound changes different from those previously discussed. And these changes are namely those in which, firstly, there is no question of mere inexactness of transmission, because they are immanent within the entire vocabulary or at least a very large part of it. Secondly, in these changes there exists no inability to produce the earlier sound, because the same sound, which is abandoned in one place, reappears at another. And thirdly, in these changes it has not been possible to prove an influence of neighboring sounds as the cause of the change. In this category belongs the most remarkable sound change in the Germanic languages: the sound shift of the mutes.8
7. The ways in which one sound changes into another can be twofold. Either a certain sound changes swiftly into another particular sound, or it passes gradually through a continuous series of intermediate sounds. In the case of sound changes through neighboring sounds, especially among vowels, there is often this gradual transition. For the sound changes discussed under number 6, this gradual transition is especially applicable.
8. It is naturally not my intention above to exhaust the diverse types of sound change. Otherwise, sound change through analogy, for example, would also have to be treated. But I would prefer to reserve this and other related questions for another occasion.
1. We are not speaking here of the confirmation of the fact, that the sounds of words have changed. Neither is it a question of the form of words in one language or another. Rather, we are concerned about the process itself, by which the one form of words and sounds has replaced the other.
2. Neither is it a matter here of how word forms, which were already present in the spoken language, came to be taken into the written language. We are asking, rather, how one form of a word replaced another in the spoken language itself.
3. Although the question then about this process itself is different from the question concerning sounds which in one language are to be found in place of others in another language; the research on the latter question forms the basis for the investigation of the former. Indeed the excellent activity in the realm of comparative phonology and its admirable results lead us to believe that we shall also be successful in getting more clues about the above-mentioned processes. It is our very worthy linguists, whom we have to thank for the many results, and whose work has laid the foundation for all further investigations. Above all, it has been the comparative studies of the Indo-Germanic languages which have paved the way for us. I mention only the works of Rask, Bopp and Pott concerning the connection of the Asiatic and the European branches of the Indo-Germanic languages, the epochal Germanic grammar of Jakob Grimm, and the research in the Romance languages by Diez, in the Slavic languages by Miklosich and Schleicher, in the Celtic languages by Zeuß.
4. All of these works not only furnish the material for the investigation of sound change, but they also pave the way to this objective through the contributions, which they make toward the solution to what Diez, in the second edition of his Grammar of the Romance languages, has recently accomplished in this field. In fact the works of Diez quite clearly direct our attention toward both points; research must now be primarily focussed on these. The first of these is the investigation of the living dialects in the most specific and the most general sense of the word, and secondly, the physiological investigation of not only the living languages, but also of the languages no longer spoken.
5. Besides the very worthwhile treatment of dialects of entire ethnic groups, the investigation of living dialects will have to concern itself above all with the most accurate examination and representation possible of the particular diction of individuals. These studies will give us the possibility of drawing conclusions from the thousand-fold synchronic diversity on the diachronic succession.9
6. The second requisition: the investigation of sounds in nonliving languages is always attended by great difficulties. And yet this inquiry is the indispensable pre-condition if we want to advance from the mere demonstration of alphabetical modification to the investigation of sound change. The means at our command for converting the written letters of old languages into living sounds are quite diverse. To some extent they are provided in the structure of the languages themselves. Physiological and euphonic sound changes within the language concerned also offer numerous clues. Moreover, there is the value of the sounds in meter, which gives us so much information, especially in the classical languages and Sanskrit, and their positions in rhyme, which is so important for many languages of the Middle Ages. We do not want to review here all of the particular aids for determining the sounds of dead languages: the introduction of individual words in other languages, their transcription in another alphabet, and so forth. Rather, we shall limit ourselves to emphasizing only two means of determination relevant to the old chief languages of the Indo-Germanic family: the statements of the old native grammarians and the linguistic-historical change of the sounds themselves.
7. The importance of the old grammarians for determining the sounds is generally recognized. The general complaint, however, is that their assertions are partially ambiguous and partially difficult to understand. It is evident that this complaint is not entirely without foundation, in that the most discerning and candid scholars have arrived at quite different conclusions in many of the most important points. It should be pointed out, however, that the Indian grammarians are incomparably more accurate, more comprehensible and more explicit in their definition of sounds than the Greeks, who are in their own way also quite discriminating.
8. When I designate the historical-linguistic change of sounds as one of the means for determining sounds no longer spoken, I must first protest against a misunderstanding. In reference to Max Müller's estimable article on the languages in the area of the oriental war, Brücke says:10 "It must be noted that Max Müller considers the e and o to be diphthongs which differ from the true diphthongs, like the English J and ou in out only in degree. It is hardly comprehensible how a man of Max Müller's intelligence could defend such an error, however widely accepted, after he had read the investigations of Willis. The cause of this particular error is, as it appears to me, another error of even greater range, which he unfortunately shares with many other linguists. They believe that the nature of a speech sound can only be determined by historical and comparative philological research, for only this can be meant by the author when he refers to theoretical analysis. This determines how sounds replace one another at different times and among different peoples. But even if this occurred according to more immutable laws than it actually does, even then the analysis of individual sounds with regard to the conditions, under which they arise would be left to direct observation and scientific experiments." I subscribe fully to this statement of the discerning physiologist, and wish moreover that he would some day subject to a scientific examination the definition of sounds found in our otherwise quite laudable grammatical works. He would encounter there a great number of things which are almost more incredible than the above cited views of Max Müller11 concerning the alleged diphthongal nature of the Sanskrit ē and ō. I have therefore no intention of wanting to determine the nature of sounds according to linguistic-historical processes that would be contradictory to natural-historical observation. What I assert is rather: When we are uncertain about which sound is expressed by the symbol of a non-living language, among other arguments we can consider the question: What development has the sound of this symbol undergone in the course of linguistic history? From the answer to this question we can draw conclusions on the nature of the old sound. It is quite apparent, that by this method we would never enter into conflict with scientific conditions of sound production. For our entire investigation is to serve only the purpose of finding the historically true sound belonging to the symbol among the many scientifically possible ones. I shall demonstrate this with an example. Old High German had two i's, a short and a long. Etymologically, the short i corresponds to a Gothic (short) i; the long one to a Gothic ei. For example, Old High German stilu (furor) is in Gothic stila (with a short i); on the other hand, Old High German stîgu (scando, with long i) is in Gothic steiga. Now if someone should want to conclude that the OHG î is a diphthong, because it came from Gothic ei, then he would be guilty of the error of Max Müller, which Brücke rightfully criticizes above. If he says however: "The OHG i is long, where it corresponds etymologically with a Gothic ei; but where it stands in the position of a Gothic i (always short) it is short," then he will be right, insofar as direct proofs contrary to this assumption can not be adduced from elsewhere.
1. One of the most remarkable sound changes in the entire area of Indo-Germanic languages is the transformation which the mutes have undergone in the Germanic branch of this great language family. This transformation is not only among the most remarkable because it is one of the most important for etymological research, but rather because it runs through an entire family of sounds with amazing regularity and, moreover, has occurred in the course of centuries not once, but twice, according to the very same principles. This transformation, to which Jacob Grimm gave the name "sound shift," consists therein, as is well-known, that the Germanic languages of the Gothic stage have a tenuis in place of a Greek media, and in place of a Greek tenuis, an aspirata, and finally in place of a Greek aspirata, they have a media. But the same transformation, which Gothic experienced in relation to Greek, Old High German undergoes a second time in relation to Gothic. In spite of all restrictions and exceptions, which the course of this development undergoes, we have accordingly in this transformation a process which is undeniably based on the nature of these sounds.
2. In order to comprehend the progress of this development, however, it is absolutely necessary to determine correctly the nature of the sounds concerned in it. The tenues or blown (= voiceless) stops cause us no difficulty. The languages still living today have them as well as the dead languages, and the essential agreement of our k, t, p with the Old Greek k, t, p is not disputed. The mediae would give us somewhat more difficulty, if at the outset we have to attempt to determine relatively exactly the meaning of this concept, which the Old Greek grammarians identify with the expression mésa. For the moment, however, we can put aside this investigation, for the specific peculiarities of Greek pronunciation are not our concern in dealing with the law of the sound shift, but rather the sounds, which in the original Greek stage etymologically corresponded to the Gothic tenues, were the breathed (= voiced, = soft) stops, accordingly in the main our g, d, b.
3. The difficulty lies in the determination of the sound of aspirates. Twenty-one years ago I made the attempt to grasp the law of the sound shift more accurately by proceeding beyond the mere etymological comparison of letters and trying to penetrate into the historical-physiological process of the sound change itself.12 Among the results of this investigation was a more exact determination of the Greek and Sanskrit aspirates, a precise distinction of them from mere spirants (friction noises) and the proof, that precisely the aspirates played a major role in the process of the sound shift, a role, which the spirants, being quite different from the aspirates, were incapable of assuming. The main difference between aspirates and spirants was found to be, that the aspirate was a stop (explosiva) with after-sounding, while the spirant is a continual sound (continua), produced not through the closure, but rather through the mere constricting of the speech organs.
4. Now after many years of further research I would, of course, modify in many respects the views which I expressed in my article of 1837. Yet I still hold to the entire course of the investigation as well as to its essential conclusions, believing I can refute everything which has been said against my findings. The conclusions published by me would receive the strongest blow if the views which a perceptive physiologist recently postulated concerning the nature of the old aspirates had any basis. For Brücke is of the opinion in his article mentioned frequently above, that the old aspirates, the Indian as well as the Greek, were merely fricatives (spirantes). He attempts to support his opinion with the most diverse arguments, and I feel myself obliged, therefore, to analyze more closely his argumentation.
5. We shall first discuss the Sanskrit aspirates. Here Brücke begins his exposition with an argument, which he draws from the orthographical designation of the Sanskrit aspirates. "In the Dêvanâgarî," he says, "their signs have nothing in common with those of the respective stops; only the sign for t² (t of the cerebral group) has an unmistakable resemblance with that of its aspirate. This must be pointed out, because the almost complete lack of correlation of the signs is not entirely without importance for the evaluation of the nature of the sound."13 To the same degree as the latter is of importance precisely for the Dêvanâgarî, Brücke's argument will obviously refute his own views, as soon as the signs for the unaspirated stops reveal themselves to be in evident correlation with the signs for the corresponding aspirates. With a great number of signs, however, this correlation is not subject to the least doubt, and moreover is restricted by no means merely to the t of the cerebral series. One glance at the Dêvanâgarî signs will convince us of this fact. Let us compare in the guttural series क (ka) and ख (kha), in the palatal series ज (ja) and झ (jha), and in the labial series प (pa) and फ (pha). How one wants to explain the origin of this similarity depends naturally on the views one has in general on the origin and development of the Dêvanâgarî.
According to the present grammatical tradition of the Indians, "each aspirate is pronounced like its corresponding non-aspirate, but with an accompanying, clearly perceptible h. Consequently one may not pronounce ख (k̒) like a German ch, फ (p̒) not like f, or थ (t̒) like an English th; but according to Colebrooke ख (k̒) is read like kh in inkhorn, फ (p̒) like ph in haphazard, and थ (t̒) like th in nuthook. The relationship is the same with the other aspirates."14 Even Brücke cannot deny this. He is of the opinion, however, that the present pronunciation of the Sanskrit aspirates is not the original, that they were rather mere fricatives (spirants): ख (k̒) sounded like our ch in Spruch, फ (p̒) like f and so forth.15
To determine whether this is in fact true, we shall have to consult the older Indian grammarians. In the annotation to Pânini's Grammar we find a survey of the Sanskrit letters with indication of the manner of production in regard to the several speech organs as well as the position of these organs.16 Here all of the aspirates k̒, g̒, etc., as well as the corresponding non-aspirated stops are counted among the letters, whose utterance requires sprista, that is, the contact of the organs.17 I would not know, how one could any more clearly characterize the nature of the stops. But perhaps Brücke wants to maintain that the annotations to Pânini are not old enough for him, rather that they stood under the influence of that later change in the pronunciation of the Indian aspirates which was assumed by him. Let us look then for the earliest evidence of Indian grammar. In the Prâtisâkhya of the Rigvêda a representation of the sounds of Sanskrit has been preserved, which leads us quite far back in Indian antiquity. The very alphabet which precedes the first patala shows us how closely the aspirates were associated with the corresponding non-aspirates even in those very earliest times. For the letters are represented in such a way, that every aspirate forms a single word with the corresponding non-aspirate, which, through the dual ending (au), indicates the copulative composition of both letter-names. Thus ka and k̒a are joined to form the word kak̒au, ga and g̒a to form the word gag̒au, and so on.18 Decisive, however, is the naming and definition of the aspirates in this old grammatical work. For the ten Sanskrit aspirates (k̒, g̒, c̒́, g̒́, ṭ̒, ḍ̒, t̒, d̒, p̒, b̒) are brought together with the ten corresponding unaspirated stops and the five nasal consonants under the expression spars̓âs. This expression comes from the same word spris̓ (tangere), to which the word spris̒ta (contact) in the annotations to Pânini belongs. All of these sounds, including the aspirates, are accordingly designated as contact sounds by the old grammarian and are quite expressly distinguished from the semi-vowels (j, r, l, v) and the breathed sounds (ûs̒mâ), to which h and the sibilants are ascribed. As a clinching argument, one old annotation interprets the passage to the effect, that the spars̓âs are the letters in the utterance of which the speech organs touch one another.19 For all fifteen sounds, which with the aspirates form one and the same class, there is no doubt of this contact; for also in the nasal consonants the actual speech organs are entirely closed and only the nasal passage remains open. It is therefore quite clear what the old grammarian means with the designation spars̓âs contact sounds. Furthermore there can accordingly be no doubt, that also the aspirates were produced in his time through actual contact of the speech organs, that is, as stops.
I believe herewith to have given the proof, that the Sanskrit aspirates were stops, and it only remains to put in its right place an argument especially emphasized by Brücke. Max Müller says in the above-mentioned publication:20 "According to the Sanskrit grammarians we produce the aspirate as a modified tenuis, and not as a double consonant, in that we begin to pronounce the tenuis, but instead of breaking it off sharply, we allow it to be produced with what they call the corresponding wind (flatus, incorrectly rendered as sibilans)." From this Brücke wants to conclude:21 "Let us first turn our attention to this passage,"22 he says, "so far as it is concerned with the tenuis aspirates, that is, the voiceless aspirates. So far it does not give cause for the slightest doubt, since Max Müller mentions on p. 27, that the fricatives are called winds by the Sanskrit grammarians. In this passage the derivation of the voiceless fricatives from the voiceless stops is described. No one could invent a description of such simplicity and truth, if these fricative sounds did not exist in the language. The present pronunciation of the voiceless aspirates is consequently not the original." Rather according to him, the Sanskrit aspirates must have been voiceless fricatives, namely ख (k̒a), our ch in Spruch, छ (c̒́a) our ch in sprich, etc.
This entire argument, however, seems to me to suffer from an inner contradiction. Even according to Brücke, the Indian grammarians have a clear and correct concept of the fricatives. They classify the sibilants, the h and a few others as belonging to these fricatives (ûs̒mâ, a term of the Indian grammarians, rendered wind by Max Müller). But precisely the aspirates, which are our concern here, they do not include, but rather classify them among the spars̓âs, that is, sounds which require complete contact of the speech organs. What could be clearer, than that the Sanskrit aspirates were, in fact, not fricatives (spirants, ûs̒mânas), but rather stops?
As concerns the description of their production, which Max Müller passes on to us from the Indian grammarians, surprisingly enough it agrees completely with the statement I gave of the Old Greek aspirates twenty-one years ago. According to the description of the Indian grammarian, we have a Tenuis with a following, yet incompletely developed spirant, which, as regards the organs concerned, corresponds to the Tenuis. Twenty-one years ago I proposed to represent this undeveloped spirant with a proposed ħ and delineated the sound in question as follows:
From these result the hard aspirates pħv, tħs, kħhh. If the following spirant develops completely, then we obtain the double consonant pf, tß (new High German z) and kch, which still occurs in many dialects of German-speaking Switzerland. The difference between the aspirate and the corresponding double consonant is this: with the double consonants (pf, tß, kch), after production of the Tenuis the speech organs are brought into the steady position, which is necessary for the utterance of the clearly developed spirant. They are held in this position for a time, so that the spirant produced by it is distinguished from the preceding tenuis as a separately articulated sound. Because the speech organs have this steady position, one can also hold them there as long as one likes and, for example, make the uninterrupted sound: pfffff, tßßßßß and so on. The situation is quite different, however, with the undeveloped after-sounding of the aspirates. This is produced only through the slow opening of the organs after the closure of the Tenuis. The organs do not remain for one moment in the same position. Therefore no clearly determined, separable sound can be produced; and just so, this sound, which is involved in a continuous change from the time of its origin until it fades away, cannot be maintained steadily. For it begins with the point of opening and ends with such an expansion of the speech organs, that the stream of flowing air no longer makes audible friction.
In order to make myself as clear as possible, I have restricted myself intentionally to the simplest circumstances.
6. I have already treated the nature of the Old Greek aspirates quite extensively in my work concerning aspiration and the sound shift.23 From the agreement of the Old Greek grammarians with the development of the sounds within the Greek language itself as well as in the relationship of Greek to the other Indo-Germanic languages, I demonstrated the Greek aspirates to be the sounds already given in the previous paragraph: pħv, tħs, kħhh. I would not know very much to add to what I said then and I confess that Brücke's publication, which in other respects is highly instructive, has not shaken me the least in my convictions on this point. Brücke starts out from the fact that the Old Greek grammarians intended the same distinction with their division of the letters into phōnḗenta and áphōna, as he makes in connection with the Indian grammarians between voiced and voiceless sounds. But just one condition should have prevented him from making this assumption. The Old Greek grammar does not merely have phōnḗenta and áphōna but also hēmíphōna. Now what shall we do with these? By Brücke's view they must have been intermediary between voiced and voiceless sounds. Admittedly there is no such thing. We maintain the assumption therefore that reproduces phōnḗenta as vocales, hēmíphōna as semivocales, and áphōna as mutae. By vocales, the vowels are understood; by semivocales those sounds, which are formed through the narrowing of the speech organs. The division of the latter into khilá (tenues), daséa (aspiratae), and mésa (mediae) we explain as follows: grámma khilón designates the letter, whose sound is cut off sharply without after-sounding; dasú designates the stop with a strong air gust after the opening of the closure; finally méson designates a sound, which to be sure does not have the strong after-sounding of grárnma dasú, nor also the sharp cutting off of all after-sounding like the grámma khilón. From this assumption one can best explain the development, which not only the daséa, but also the mésa have undergone in New Greek.
7. If we take these results as a basis, which the investigation of the Sanskrit and Old Greek aspirates has furnished us, we find that our present High German language actually does not have any aspirates. The essence of the aspirates consisted therein, that it was a stop with an undeveloped after-sounding. In this class our f, ß, and ch (in sprich as well as in Spruch) do not belong. For they are continuant sounds (continuae), produced through narrowing, not closure of the speech organs. Neither are our pf and our z (= tß) aspirates. To be sure, they begin with a stop, but do not follow this with such an only half-developed after-sounding as we described above (§ 5), but rather a clear, fully-developed spirant. Our pf and z are therefore double sounds, which the Indian and Greek aspirates were not.
8. The Germanic languages of the Gothic stage no longer have a guttural aspirate; of the labial, they have preserved only a small part, and these medially. On the other hand, they have the dental aspirate initially as well as medially. The h and f, which these languages have in the positions where we might normally expect aspirates, are not aspirates but spirants. A remnant of the labial aspirates, Old Saxon possesses in its medial ƀ. All of the older Germanic languages of the Gothic stage have the dental aspirate th (þ). It has, however, been partly lost in the modern languages such as Swedish, Danish, and Low German; in others its change and eventual loss has been going on for centuries, as in English, which still shows a slight trace of the genuine old aspirate only in those instances, where the pronunciation of the th begins with the stop.
9. If we relate what we learn about the nature of the aspirate from the Indian and Greek grammarians with the results of etymological research, we recognize by a clear example how a real history of sound changes only results from the combination of scientific determination of sounds and etymological comparison of words. In place of the Greek and Sanskrit aspirates we find etymologically the soft (=voiced =breathed) stop in the Germanic languages of the Gothic stage. For example:
|Greek kh||=||Gothic g||(e.g. khéō,||Gothic giuta)|
|Greek th||=||Gothic d||(e.g. thúra||Gothic daúr)|
|Greek ph||=||Gothic b||(e.g. phérō||Gothic baíra)|
The same change is repeated for the second time in the relationship of High German to the Germanic languages of the Gothic stage. But only the dental th of the Gothic stage changes to High German d; the h does not go to g, nor the f to b. While the numerous Gothic th's consistently become d in High German (Gothic thanjan, High German denen, Gothic thata, High German daʒ etc.), the h and f remain unchanged (Gothic haubith, High German haupt; Gothic harjis, High German heer; Gothic fôtus, High German fuß; Gothic faran, High German fahren). What can be clearer than this process? The true aspirate contains the stop in itself, which remains after the cessation of its after-sounding. The Greek-Indian aspirates therefore become soft stops in Gothic, and similarly, centuries later Gothic th becomes High German d. On the other hand, the spirants h and f do not contain such a stop and consequently do not become High German g and b, but remain as they are.
10. We still have to show how precisely a media came from the aspirate. There is, of course, no question of a general necessity. The hard aspirate can also leave the hard stop by giving up its after-sounding. Such has been the case of the Old Norse th in Swedish. According to a very widespread but erroneous theory, one is inclined perhaps to say that it is impossible for aspirates, as we conceive of them, to begin with a soft stop. One will say: "The hard stop requires a greater amount of air than the soft. Now if the quantity of air pouring forth is increased further with the aspirate, how can it possibly begin with a soft stop?" But this argumentation is based on an erroneous physiological assumption. The hardness of the stop is dependent, to be sure, upon the amount of air emitted, but not upon this alone; it is at the same time also dependent upon the further condition, that the expelled current of air finds a firm closure of the speech organs, which can be opened only by strong pressure. If on the other hand the closure yields somewhat in firmness while the mass of onrushing air pressure either remains the same or even increases, two phenomena happen, which condition each other. First of all, a part of the breath rushes out only after the stop is opened and so creates the circumstances necessary for producing the true aspirate. Secondly, the stop will lose some of its hardness precisely through this premature yielding of the closure. For only the breath, which rushes on before the opening of the closure, conditions the hardness of the sound. So we see almost before our very eyes, the hard, non-aspirated stop (tenuis) gradually becoming through further intensification, an aspirate with a softer stop. For it is, in fact, the further strengthening of the breath beyond that of the tenuis, which causes the earlier opening of the closure. In this way the phonetic-historical series t - th - dh - d can result, as well as, more directly, the group t - dh - d. Old Saxon provides us with the documentary proofs for the whole process.