“A Baudouin de Courtenay Anthology” on Digital Publishing at Indiana University Press

I have published a larger work, Proba teorji alternacyj fonetycznych (An Attempt at a Theory of Phonetic Alternations) in Rozprawy Wydzialu filologicznego (Transactions of the Philological Section) Cracow Academy of Sciences (now Polish Academy of Sciences in Cracow), vol. XX, 1894, pp. 219-364 (also published separately). According to usual procedure, I was also supposed to supply a short summary in German or French for the Anzeiger (or Bulletin) of the Academy. However, since I felt it necessary to make available to non-Polishspeaking scholars the particulars of the exposition and the course of my argument, and especially the formulas which I have advanced, I had greatly to exceed the maximum length of such a résumé and instead of a short summary, give a German version of the same material. Nevertheless, my entire manuscript was accepted without reservations for preparation by the editors of the Anzeiger, set up completely in galley proofs, and painstakingly corrected by me with the kind assistance of Mr. St. Rosznecki, Cand. Mag.; then nearly half the type was set under the heading of the Anzeiger and my German spelling was changed to conform to the style of the Anzeiger. Suddenly, however, toward the end of October, I was told it was impossible to accept such a long work for the Anzeiger (of which a single volume could under no conditions exceed four folios and still cost only 40 Kreuzers). And, in fact, the Anzeiger is not the most suitable place for a work which has cost so much effort. There-fore I decided to publish it as a separate book, even though this entailed the tiresome, not very profitable, and time-consuming task of changing the résumé style into a direct scholarly exposition and the expression the author into I. It also entailed the division into individual chapters and so forth. And I had to do this with the material already in type.

The history of the origin of this modest contribution likewise explains why I have taken my examples primarily from Polish and have cited them generally without translation. If I had composed the German version independently of the Polish original and had not been forced to merely transform an Anzeiger résumé into a separate study, I would have aimed at a greater selection of exampies from various languages. Nevertheless, I believe that the Polish examples are quite comprehensive and that anyone will be able to test the general propositions of my work on any other Ianguage he might like.

The subtitle, A Chapter from Psychophonetics, sounds somewhat pretentious; however, by this I merely wanted to indicate that I consider myself an adherent of the linguistic school which emphasizes the psychological factor in all linguistic phenomena.

I would advise the beginner as well as the reader who is not accustomed to such explanations not to read this book through from beginning to end without interruption, but rather to study it in the following order: after the Explanation and Definition of Some Technical Terms and the Explanation of Symbols and Abbreviations (p. 132), read through Chapter I (p. 153) attentively, then pass over Chapter II and go directly to Chapter III (p. 170) and after that go through the four following chapters, IV-VII (p. 175 ff.) in sequence; after Chapter VII one can return to the Introduction, followed by Chapters II (p. 161) and VIII (p. 197).

INTRODUCTION

_________________________________

At present I am giving only the first part of my work. The second part should encompass: (1) an analysis of the conditions under which alternations arise; (2) a classification of alternations, at the time of their emergence and (3) during their further existence, when they are maintained through the force of tradition and linguistic intercourse; (4) an account of the different means of utilizing alternants for psycholinguistic purposes; (5) a specification of the limits within which the alternants may move in various directions; (6) an analysis of the different layers of alternation; and (7) an analysis of the correspondences of alternations, i.e., of alternations which correspond in different languages. Moreover, I intend to give special consideration to: (1) the alternations of Sanskrit, (2) the alternations of the Indo-European (Indo-Germanic) Ianguages, which are of common Indo-European origin, and (3) the alternations common to all Slavic languages.

The etymological relationship of speech sounds has been observed for a long time, roughly since the time when man began to be seriously concerned with grammatical and especially with phonetic questions.

The greatest heights here were attained by the Indie grammarians, who developed a very refined theory of the laws of sandhi, on the one hand and the laws of guna and vrddhi on the other hand. But the Indie grammarians lacked a feeling for history and were unable to grasp the significance of gradual development, historical sequence, or chronology in general. As a result, their findings lay, so to speak, on a single temporal plane; everything happened simultaneously, as though there were neither a past nor present nor future. Thence also the purely mechanical character of their gramamtical rules; they give excellent prescriptions for the formation of all kinds of grammatical forms, but we would look in vain for a scientific explanation of the ways and means by which these forms originated.

The concepts of guna and vrddhi have been taken over by European grammarians under the names Ablaut, phonetic gradation and the like, and the theory of phonetic changes has now reached a high level of sophistication. Although European linguists have been dependent on the views of the Indie grammarians, they have surpassed the latter in two ways: first, they have introduced the concept of chronology into their investigations and have made use of it with greater or lesser success; second, they have formed their conclusions on a much broader comparative base, using material of a greater variety of languages, including not only related but also unrelated languages.

According to the most recent studies of phonology, the relationship between two different but etymologically related sounds is completely the reverse of what was supposed by earlier linguists; what had passed earlier for the basic form turned out to be a derivative (secondary) form, and vice versa. A characteristic example of this change of viewpoint is the replacement of the theory of vowel gradation by the theory which posits the weakening of a stronger phonetic structure and the loss of a part of that structure (F. de Saussure, Brugmann, Osthoff, Hübschmann, and many others).

But even the most recent linguistic works deal with the alternations themselves only indirectly, focusing attention on the determination of phonetic changes and the establishment of the historical priority or origin of the speech sound in question. Furthermore, these works fail to give a satisfactory account of the very concept of alternation or coexistence. One of the works which most closely approaches the concept of alternation developed in my preceding study, and which affirms first of all the existence of alternations, is August Leskiens Der Ablaut der Wurzelsilben im Litauischen (Leipzig, 1884).

The expression alternation in the sense in which I use it does appear, however, here and there in contemporary linguistic works. Thus, for example, de Saussure says (Mémoire sur le système primitif des voyelles dans les langues indoeuropéennes, Leip. zig, 1879, p. 12): The Italic languages have too greatly leveled the inflection of the verb for us to expect to find in them the alternation of weak and strong forms.

I arrived at the concept of alternation developed in my preceding study more than eighteen years ago, when I was giving lectures on certain aspects of comparative grammar and general linguistics at Kazan University and at the Kazan Theological Seminary.

The customary approach to phonetic differences in the works available to me at that time was to determine the change of one speech sound into another, and particularly to establish sound laws and so forth. This approach could not satisfy me for, first, it gave an inadequate account of the chronology or order of the historical layers and, second, it gave an inexact formulation of the actual facts. One such overriding fact is the coexistence of phonetically different but etymologically related speech sounds; only after establishing this fact can one proceed to the investigation of its causes.

My views on the subject at that time were expressed, or rather noted in passing, in my Detailed Program of Lectures for the Academic Year 1876-77, pp. 92-95, 107; in my Detailed Program of Lectures for the Academic Year 1877-78 p. 114ff.; in Iz Lekcij po latinskoj fonetike, Voronež, 1893, and even much earlier in Wechsel des s( š, ś) mit ch in der polnischen Sprache (. Beiträge zur vergleichenden Sprachforchung von Kuhn, VI, pp. 221-22) (where in 1868 this alternation s ||ch was called a constant gradation used to differentiate meaning ). My ideas on this matter have gradually become more refined and precise. In my Russian lectures I used the word čeredovanie <“alternation> to indicate the со-occurrence in one and the same language of phonetically different but etymologically related sounds.

One of my Kazan students, M. Kruszewski, in his dissertation Nabljudenija nad nekotorymi foneticeskimi javlenijami, svjazannymi s akcentuaciej (Observations of Some Phonetic Phenomena Connected with Accentuation), Kazan, 1879, collected the material available in the Rig-Veda to study the alternations caused by the influence of the accent. He acquired a clear and independent view of similar phonetic phenomena and presented it in the introduction to his masters thesis К voprosu o gune. Issledovanie v oblasti star oslav jansko go vokalizma <On the Question of Guna. A Study in the Area of Old Church Slavonic Vocalism>, Separate offprint from Russkij Filologiceskij Vestnik, Warsaw, 1881 and also in the German version of this introduction Über die Lautabwechslung, Kazan, 1881.¹

Kruszewski developed the theory of alternations more philosophically, more comprehensively, and more precisely than I myself have done, thanks mainly to his strict application of the analytic method. It cannot be denied, however, that Kruszewski merely gave another, finer form to what he had learned from someone else. (This was admitted by Kruszewski himself; cf. Über die Lautabwechslung, p. 1.) One might take exception to certain shortcomings in Kruszewskis presentation of the theory of alternations. Despite his rigor and analytic method, Kruszewski left many things unnoticed; he did not define the boundaries between the individual classes of alternations precisely enough, and he set up alleged characteristic features of individual classes that could by no means be considered characteristic; on the other hand, he ignored other features which should have been mentioned as the most characteristic for a given class. Moreover, Kruszewski made occasional mistakes in logic. All of this, however, should not surprise us in view of the novelty and difficulty of the subject, and especially if we consider that Kruszewski based his general conclusions on facts drawn from the limited area of phonetic changes caused in Russian by palatalization or softening of consonants.

Kruszewskis terminology is not satisfactory and cannot be retained today. After all, both works of Kruszewski, his masters thesis and his brochure Über die Lautabwechslung , belong to a time when the Kazan linguists became infatuated with nomenclature, developing a mania for inventing new and unusual techical terms; Kruszewski was wise enough to use some restraint in this respect in his works. This disease reached monstrous proportions in my own Nekotorye otdely sravniteVnoj grammatiki slavjanskix jazykov, the reading of which could only be hampered by such technical terms as coherents, homogenes, heterogenes, monogenes, polygenes _{_גי}י amorphism and secondary heterogeneity of morphemes, amorphous correlatives, divergence and anthropophonic coherence, mobile correlation and morphological coincidence, coincident correlatives՝, coexistent correlatives, etc.

However, despite this frightful number of newly coined technical terms, there is a sound kernel of useful observations in this work. Among one of its most original ideas (which was, however, nothing new for me personally, since I had been developing it for several years in my lectures) was the requirement to distinguish native alternations from foreign (borrowed) alternations. Of a certain methodological value are also: first, the distinction of the concepts sound and phoneme²; second, the unification of the concepts of root, affix, prefix, ending, and the like under the common term, morpheme; third, the distinction between a purely anthropophonic, physiological, and acoustic analysis of human speech and an analysis based on a morphological and semasiological view-point; fourth, the distinction between basic (primary) and derived (secondary) members of an alternation.

Another work of mine appeared almost simultaneously with this selection, Otryvki iz lekcij po fonetike i morfologii russkogo jazyka, citannyx v 1880-1881 akad. godu . . . This work was mainly devoted to an analysis of the divergents in Russian.

The renowned Orientalist Dr. V. Radloff (at present a member of the Imperial Academy of Sciences in St. Petersburg) in his article devoted to the application of Kruszewskis principles to the Turkic languages Die Lautalternation und ihre Bedeutung für Die Sprachentwicklung, belegt durch Beispiele aus Den Türksprachen, Abhandlungen des fünften internationalen Orientalisten Congresses gehalten zu Berlin im September 1881, Berlin, 1882, has replaced the term phonetic interchange by phonetic alternation. The alternations of the first type he calls, as I do, phonetic divergence, divergents; alternations of the second type which in my present terminology are called traditional alternations he calls phonetic compensations, compensatory sounds (a not particularly apt expression), and finally he has restricted the term correlation, correlative to alternations of the third type, as I do, too. In R. Brandts recently published book Lekcii po istoriceskoj grammatike russkogo jazyka, čitannye Romanom Brandtom, Vypusk I Fonetika <Lectures on Russian Historical Grammar by Roman Brandt, vol. 1, Phonetics>, Moscow, 1892, we also find a chapter entitled Ceredovanie zvukov <Phonetic Alternation>. But Brandt evidently understands under this term something different from what I have always understood; for he is concerned not with the phenomenon of simultaneity, or coexistence, but with the explanation of phonetic changes which he divides into changes (perexody) and substitution (podstanovki) depending on whether they occur now or whether they occurred in the past. Professor Brandt is absolutely right when he reproaches me for the mass of newly coined technical terms in my Nekotorye otdely sravniteVno] fonetiki. But the same reproach also applies to him, for he has probably invented more than a hundred completely new technical terms, which make the reading of his works infinitely more difficult, and which, in comparison with my own terms, have the sole advantage of having been coined from native Russian rather than from Latin. This is, however, a very dubious advantage, for a Latin technical term can be easily understood by scholars in Europe and America, irrespective of nationality, while an exclusively national term, coined under the influence of puristic tendencies, only increases the difficulty of mutual understanding. Our times, infected as they are by international animosities, cannot well afford them.

My attempt at presenting a theory of alternants will perhaps receive no recognition. It cannot, however, be denied that the concept of alternation and alternants is relevant to an enormous mass of linguistic facts, for there is probably no sound in any Ianguage which is completely isolated and does not alternate with another sound, just as there is no word to which the study of phonetic alternations cannot be applied.

Explanation and Definition of Certain Terms

The phoneme = a unitary concept belonging to the sphere of phonetics which exists in the mind thanks to a psychological fusion of the impressions resulting from the pronunciation of one and the same sound; it is the psychological equivalent of a speech sound. The unitary concept of the phoneme is connected (associated) with a certain sum of individual anthropophonic representations which are, on the one hand, articulatory, that is, performed or capable of being performed by physiological actions, and on the other hand, acoustic, that is the effects of these physiological actions, which are heard or capable of being heard.

Phonetics, as a whole, concerns all phonetic phenomena, both anthropophonic (whether these be of a physiological or auditory nature) and psychophonetic in which the former, sensory phenomena are reflected. Phonetics consists thus of two parts, of an anthropophonic and psychophonetic one.

The morpheme = that part of a word which is endowed with psychological autonomy and is for the very same reason not further divisible. It consequently subsumes such concepts as the root (radix), all possible affixes, (suffixes, prefixes), endings which are exponents of syntactic relationships, and the like.

Explanation of Symbols and Abbreviations

= . . . identity in the mathematical sense.

│. . . alongside, one occurring alongside the other.

││. . . symbol of alternation or co-occurrence, of a unilingual correspondence, of etymological relationship in one language.

╫ . . . symbol of correspondence, of a multilingual correspondence, or of etymological relationship in different languages.

→ ... symbol of transition, change; whatever stands on the left side of the symbol had changed into whatever stands on the right side.

← . . . the reverse symbol, symbol of the origin of whatever stands on the left side of the symbol from whatever stands on the right.

. . . symbol of the lack of any connection from either point of view.

# . . . symbol of parallelism.

* ... assumed, reconstructed form.

« + » ... denied, impossible, or invented form.

CHAPTER I

_________________________________

DEFINITION OF ALTERNATION AND ALTERNANTS.
ESTABLISHING THE NOTION OF ALTERNATION
ETYMOLOGICALLY AND PHONETICALLY.
THE ORIGINAL CAUSE OF EVERY ALTERNATION.

In every language and in the speech of every individual we observe a partial phonetic difference between etymologically identical morphemes. In other words, in every language there are etymologically related morphemes which differ phonetically in some of their parts. For example, in the etymologically related morphemes mog- and mo ž- of the Polish words mog- ę| mo ž-esz, the first two phonemes m and o are identical, but the final phonemes g and z are different. Such phonetically different phonemes, which are parts of etymologically related morphemes and which occupy the same position in the phonetic structure of the morphemes (in the cited example, the third position), we shall call alternants, and their relationship to each other, an alternation.

Similarly, there are two obvious alternations in the root morphemes of the Polish words mróz | mroz-u (gen. sg.): u( ό) ║ о and -s(-z) ║-z-. A distinct alternation also occurs at the juncture between the root morpheme and the formal morphemes in the Polish words plac-i-c│plac- ę :ći ║с. In the base morphemes of the German words lad-en │ Las-t, Ver-lus-t │ ver-lor-en, Fros-t | frier-en, geb-en | gab . . ., we find the following clear alternations: 1) d ║s, ā ║ ă; 2) s║r, ŭ ║ō; 3) s ║ r, ǒ ║ í (ie); 4) g ║ g _{_í,}e ║a, -Ь║֊Р(Ъ).

Strictly speaking, in all such cases we could consider the alternating units to be not phonemes, but morphemes, since only the latter form semantically indivisible linguistic units. Thus, from the point of view of the psychological reality of language, there is an alternation between entire morphemes and their combinations; e.g., Polish mog-║mož-, mrus ║ mroz-, plać-i ║ plac-; German lād- ║ lăs—, lŭs- ║ lōr-, frǒs- 11 frir-, g _{_ie}eb- ║ gap. The phonetic difference between related morphemes we shall call a phonetic alternation. (This alternation is connected with a semantic alternation between the morphemes and entire words.) The phonetic alternation of entire morphemes, however, can be reduced to alternations of their phonemes, or the phonetic components of the morphemes.

In terms, then, of phonemes, we shall call phonetic alternants or alternating phonemes such sounds or phonemes which differ from each other phonetically, but are related historically or etymologically. Phonetic alternants, or alternating phonemes are, in other words, sounds or phonemes which, though pronounced differently, can be traced back to a common historical source, i.e., originated from the same phoneme.

Establishing the Notion of Phonetic Alternation and Phonetic Alternants by Etymology.

The semantically basic morphemes of the Slavic words, Pol. prosi-ę| Cz. pras-e | Rus. рэгаś-όпэк (porosënok ) | Ukr. poroś-á (porosja ) ן SCr. prās-àc | Slov. (Krajna) pras-c are etymologically related, since they can be traced back to a common historical source, *pors–. But this common historical source *pors- can, in turn, be compared with etymologically related morphemes of other Indo- European languages: Slavic *pors- | Lith. pařš֊as | Lat. porc-us | Germ. *fahr–. Likewise, Slavic *vez- {Pol. wiezi-e | Cz. vez-e | Rus. vez֊ót (vez-ët) | Ukr. vez-é (vez-e) | SCr. vèz-e . . .} 4= Lith. vež- ╪ Germ. *vig- ╪ Lat. veh- ╪ Gk. fex– ╪ Skt. vah- and others.

Any such comparison of words and morphemes of different languages rests on the assumption that the morphemes in question are etymologically related. The etymological relationship is established on the basis of semasiological similarity, on the one hand, and partial phonetic similarity of the morphemes, on the other hand.

The phonetic similarity must be neither accidental nor arbitrary, but must recur in a series of morphemes comprising at least partially the same phonemes. Thus, for example, we posit the etymological relationship of the basic morphemes of the above-quoted Slavic words prosi-ę │ pras-e │poroś*–a . . . first, on the grounds of their semasiological similarity (they all mean pig, boar, swine), and second, on the grounds of the occurrence of identical phonetic correspondences in a whole series of words (p; s with different reflexes; ro ╪ra ╪ oro. . .). The etymological relationship of morphemes belonging to various Indo-European languages (e.g., *pors–╪ parš– ╪ pork– ╪farh– . . . ; *vez – ╪vez– ╪*vig– ╪ veh– ╪ fex– ╪vah-) is likewise founded on the semantic similarity of the morphemes and the occurrence of regular phonetic correspondences in a series of words (ρ ╪ ρ ╪ ρ ╪ ƒ…., or ╪ ar ╪ or ╪ ar…s ╪ š ╪ k ╪ h…; υ ╪ υ ╪ υ ╪ υ(f)…,e ╪ e ╪ i ╪ e ╪ e(ε) ╪ a…, z ╪ ž ╪ g ╪ h ╪ x ╪ h…).

It is on the basis of similar comparisons that we say: Polish ro corresponds to Czech and South Slavic ra and to Russian oro; these diverse Slavic sequences can be derived from a proto-Slavic sequence or; all these Slavic sequences, as well as the assumed proto-Slavic sequence or, have correspondences in other Indo- European languages, i.e., ar in Baltic (Lithuanian and Latvian) and Germanic, or in Latin; Slavic s corresponds to Lithuanian š, German h, Latin k, etc.

It would be a mistake however, to regard the historical-phonetic inference as being unimpeachable. What is etymologically related in the different languages are not independent phonemes, but the morphemes in which they occur, i.e., those simplest, indivisible semasiological parts of a word which have an autonomous psychological existence. Consequently, when we speak of the affinity of Slavic z with Lith. ž, Germ, g, Lat. h , Gk. _x, Skt. h , we do not mean the aboslute relationship of these phonemes, regardless of the morphemes in which they occur, but the relative relationship of these phonemes in particular morphemes (vez–, liz–ać, zim–a). The phoneme z is related to certain phonemes of other Indo-European languages in words such as zqb, zn-ас, ziarno. . . . Similarly, Slavic s corresponds to Lith. s ╪ Germ, h =j= Lat. and Gk. k =|= Skt. ç in the word *pors-ę (prosi-ę), etc., but not in such words as *sedmi, sebe, syn-, bos-, etc.

Thus, all etymological comparisons of words belonging to different languages are based on the recognition of an etymological relationship between the morphemes comprised in these words. But the etymological relationship of the morphemes can be treated as an etymological relationship of the particular phonemes and their combinations.

Etymological relationship based on the comparison of various languages, we shall call correspondence, or interlingual correspondence. But in addition to this type of relationship, there is also the etymological relationship of morphemes belonging to one and the same language. Since we recognize the etymological relationship of different, though similar morphemes, belonging to various Ianguages, e.g. {pros- │ pros- │ poros- │ *pors-} │ parš- │ pork- | farh- we should with even more reason admit the etymological relationship of morphemes belonging to the same language; e.g., mog- │ mož—, rod- │ rut, mroz- | mrus, plot– │pieś-, etc. The second, or intralingual type of etymological relationship, we shall call alternation, and the alternation concerning specifically the phonetic elements of the morphemes, phonetic alternation.

In the case of alternation, or intralingual etymological relationship, we are dealing with speakers belonging to the same speech community, while in the case of correspondence of interlingual etymological relationship, the speakers, or more specifically the carriers of elements forming a relationship of correspondence, belong to multiple and ethnically diverse speech communities.

As in the case of correspondence, we can treat the alternátion of morphemes, in their identity and difference, as an alternation of their constituent phonemes (mog- ║mož- can be broken down into the phonemes m ║m, o ║o, g ║ž; and mroz- ║mrus- into mr ║ mr ,o ║u, z ║s. . .). The existence of corresponding and alternating morphemes is due to the fact that the respective morphemes were in either case historically the same.

What has been said with respect to correspondence—namely, that it concerns morphemes, and not individual phonemes, and that the latter form a relation of correspondence only insofar as they are a part of certain groups or types of morphemes (i.e., not every Slavic о corresponds to the Latin о, but only the о of some morphemes)—holds likewise with respect to alternation. Thus, for example, not every Polish z alternates with g, but only the z which occurs in morphemes that alternate with morphemes containing g. But while the correspondence of phonemes implies a correspondence of morphemes, there are also purely phonetic alternations (variations, ramifications) of phonemes that are not bound to one or another morpheme. These variations I have called general phonetic divergents. Thus, for example, the Polish i _₂ (y) || i₁ (i) in the phonetic sequences py, by, my, ty, dy . . ., cy, dzy, sy,ry . . . , czy, džy, szy, žy . . ., pi, bi, mi. . ., ći, dźi, śi, źi . . ., ki, gi, chi . . . , ji, do not depend on the semantic value of these sequences. The same holds true of Polish e ₂ || e₁ (pe, be ... \ pe, be ...), s || ś (st, sp .. . \ ść, śp . . .), ę\\en\\em\\eh\\en (ęs, ęz, ęš, ęž, ęch \ ęt, ęd, ęc, ędz \ ęp, ęb \ ęk, ęg \ęc, çdz . . .).

Establishing the Notion of Alternation Phonetically

In almost all works and studies dedicated to phonetic questions we find pronouncements concerning the transition of some sounds into others, the transformation (or change) of some sounds into others, and so forth. Thus, for example, Polish grammars speak of the change of k into cz in the words piecz-e, racz-ka . . . <dim.>, from piek-ę, ręk-a, and of a into a in the words maz, dab . . . from męz-a, dębu . . . <gen.>. . . .

Such a formulation of the relationships in question is misleading. Whoever uses it is confusing an arbitrary, subjective (linguistic) experiment with the objectively given historical processes.

In an arbitrary experiment, we can pass from one sound to another by substituting, according to need, the various movements of the speech organs. Thus one can easily pass from b to ra if one merely keeps open the nasal cavity by lowering the soft palate without changing the position of the other speech organs. In the same way, one can pass from a hard (nonpalatalized) p to a soft (palatalized) p, from о to u, from e to o, etc. In this manner, by changing only one property at a time, one can move from p to a through almost the entire scale of the speech sounds: p - p - b - b-m֊n-d-z-ž֊š-s-x (ch) -k-g-g- γ (h) [-j] -i e-o֊u֊y֊a.

But what do we actually do in such an experiment? We gradually alter the groups of phonetic representations and perform the corresponding physiological processes. But we need not confine ourselves to such a gradual change; after pronouncing p we can directly imagine the representation of a and implement it phonetically. And then we are indeed entitled to say: p has changed into я.

But what does actually happen? One group of phonetic representations is substituted by another, and their successive implementation gives us a certain right to say: p passes into a, x passes into y. In reality, though, the various associations of representations follow each other, like slides in a kaleidoscope, only in the head of the experimenter. The successive production of sounds corresponding to these associations is only coincidental and certainly not obligatory. Thus even at this stage there is no change from one pronunciation to another, but only replacement of one mental image by another.

In the objective historical development of a language such sound changes are purely fictitious, to say nothing of the types of transition such as k into cz, g into z or ç into q. There are neither pronetic changes nor phonetic laws, and there can never be such, if only for the simple reason that human language in general, and the sounds of a language in particular, are not endowed with continuous duration. A word, or a sentence, once uttered, disappears at the moment of its utterance. There is no physical connection between utterances. What links the separate speech acts—be they sounds, phonetic words, or utterances (that are heard and perceived by the ear)—are representations, or images of the memory , which during the utterance itself serve as a stimulus to set the speech organs into appropriate motion.

(The production of sounds) allows for two possibilities: the physiological conditions determining the activity of the speech organs may allow the full realization of the processes intended by the brain center, or they may inhibit them. In the first case, the phonetic intention coincides with its realization (for example, za, ra, ar, la . . . Pol. mech, jablek <gen. pl.> . . . ; rodu, mrozu, mçia <gen. sg.>; ivoda <nom. sg.> . . .); while the second case produces a discrepancy (e.g., Pol. zta— monosyllabic and with a voiced z, rtęć— with an ordinary voiced r, atr, tka, mchn, piekt, jabtko . . . , rud (rόd)—with -d, mruz (mróz) with -z, maz with z, wud-ka (wόdka) with -d-). In the second case, our phonetic habits and the universal conditions determining the production of phonetic sequences, compel us to modify the pronunciation of the intended sequences; e.g., sta (with s instead of z), γtçc, atr (with voiceless and weakened r), Ika, mchu, piekt, japtko . . . , rut (with a weakened t), mrus, maš, wut-ka. . . .

But even in forms such as cnót or matka, the phonetic environment prevents us from pronuncing the t as clearly and independently as in cnota and matek, . . . and compels us to substitute for it a t which is weakened and dependent on the following sound.

An intended pieke instead of piecze or rqkek, rqkka instead of rqczek, rqczka could, on the other hand, be pronounced without the slightest difficulty and without necessitating a change of k into cz.

The discrepancy between the phonetic intention and its realization is solved by substituting an intended impossible activity by a possible one.

The substitution can be of two kinds: (1) if the intention, which is founded on related words and forms, cannot be realized, the intended phoneme is replaced by one which is closest to it phonetically (examples above); or (2) the actual pronunciation is imitative of foreign sounds in the speech of others which we intend to reproduce. The latter type of substitution is known to occur (a) in the speech of children and, more generally, within the limits of the native language; and (b) in the rendering of foreign words whose pronunciation is modified in accordance with our own phonetic habits (for example, when French sur is pronounced in Polish sur [siur]).

Phonetic change or transition results in all these cases from the discrepancy between the intention and its realization.

Only the first type of substitution applies to the alternations and alternants.

The substitution of an intended pronunciation by a possible one constitutes the only type of phonetic change or transformation that may occur in the synchronic state of a given language. What is ordinarily called phonetic change or transformation of one sound into another is, from an objective point of view, only соexistence, or alternation.

Such coexistence, or alternation, is neither phonetic change in the present nor succession in the historical sequence. It is simply the phonetic difference between etymologically related morphemes. However, its cause is still considered something of a puzzle.

If it is claimed that the č(cz) in such contemporary words as pieczç, rqczka, etc., is derived from k, then one could with equal right insist that on the contrary, the k of piekę, ręka is derived from c. Notice the complicated alternation:

It would be a sign of poor thinking, and a historical error, to claim that the с in ciec, móc ... is based on kć (or gć).

In short, phonetic change, as it is ordinarily understood, is a fiction, a delusion. There can be only:

1) substitution of intended activities by possible ones, (stemming from) the lack of coincidence or discrepancy between the phonetic realizations and intentions, and

2) synchronic phonetic differences, i.e., alternations of a historical origin of morphemes and of their phonetic components, the phonemes.

The two types of phenomena are closely interconnected. Active, dynamic substitutions give rise to embryonic, incipient phonetic alternations; while the alternations which from a contemporary point of view seem to have no cause, can be traced back to substitutions which took place in the past.

Causes of Alternations

If the history of a given language is viewed as something continuous and uninterrupted, it appears that the cause or stimulus (impulse) of a given alternation is a purely phonetic or anthropophonic one. But in the case of mixed languages, the situation must be defined more precisely: the primary stimulus of an alternation is probably always of an anthropophonic nature, but it may have its roots (1) in the native language, as is most frequently the case, or (2) in a foreign language from which the given speech community has borrowed the entire alternation or one of its elements.

The original cause of the alternation may still be active in the present, synchronic state of the language or it may have been active in the past. In the latter case it can be discovered by way of hypotheses and historical inferences.

CHAPTER II

_________________________________

CLASSIFICATION OF ALTERNATIONS AND ALTERNANTS.

I. Classification of Alternations according to their Causes

The complexity of causes accounting for the emergence and preservation of alternations must ultimately be ascribed to communal life and the physical (anatomico-physiological) and psychological make-up of the members of a speech community.

1. Classification of alternations according to the possibility of determining the anthropophonic causes operating in the synchronic state of a language.

All alternations are either due to living anthropophonic tendencies and to fixed and constantly repeated anthropophonic habits, or they are not due to them. Hence we may divide them into two large classes:

1) neophonetic alternations, and

2) non-neophonetic, paleophonetic alternations.

Alternants of the first class we shall call divergents, and their relationship divergence՛, alternants of the second class we shall call nondivergents, and their relationship, nondivergence.

2. Classification of alternations according to the possibility of establishing their psychological causes.

All alternations are either associated with representations of certain psychological differences (nuances) of a semasiological (significative) or morphological (structural) nature, or they are not associated with them. On this distinction we base the division into:

1) psychophonetic alternants, or correlatives, and

2) non-psychophonetic alternants, or non-correlatives.

The relation of correlatives forms a correlation, while the relation of noncorrelatives forms a noncorrelation.

3. Classification of alternations according to traditional and_₉mmore generally, social causes.

All alternations are either the result of repetition and imitation (including transmission from one generation to another) or they are independent of this factor.

All genuine paleophonetic alternations are preserved through tradition and are, consequently, due to social conditions.

Divergences, or neophonetic alternations, arise and are maintained independently of tradition, although there are some divergenees whose anthropophonic connection with their causal factors is supported by tradition and linguistic intercourse.

Finally, the correlations or psychophonetic alternations are due, above all, to tradition and linguistic intercourse; but when they become associated with psychological distinctions, the influence of the social factor upon the speaker ceases.

4. Classification of alternations according to internal (autoglottic) or external factors.

All alternations arise either in the uninterrupted historical evolution of a given language, or they are borrowed from other, closely related languages. In other words, the present or past causes of some alternants either lie within the given speech community or result from linguistic contact of this community with other communities or literatures.

Divergences are always of internal origin, for their difference is the result of the varying pronunciation of the members belonging to the same speech community. On the other hand, traditional or even psychophonetic alternations (correlations) may have their origin in foreign sources (Polish h ║ z in blahy │ blazen of Czech origin; or the Russian er ║ra in smer det │ smrad of Church Slavonic origin).

From a synchronic point of view all alternations are internal and peculiar to the given language. They are foreign only from the viewpoint of their provenience.

Alternations stemming from a foreign source are: (a) either totally foreign or (b) partially foreign (e.g., Polish g ║h in gardzic | hardy, ganić | hańba; Russian olo | la in golová | glavá).

In terms of their origin, the alternations of the first type could be called foreign monolingual , and those of the second type, foreignnative bilingual.

5. Classification of alternations according to the difference between individual and social causes.

Divergences and correlations are essentially due to individual or, at best, collective-individual causes, whereas traditional paleophonetic alternations are due exclusively to social causes.

The individual or collective-individual causes of divergences are anthropophonic.

The individual or collective-individual causes of correlations are psychological.

Inasmuch as we attribute the traditional alternations of one language to social causes, we have even more reason to attribute foreign alternations to such causes; the social factor here embraces not one, but two speech communities.

6. Classification of alternations according to the simplicity or complexity of their causes.

a) All alternations have either one or two causes.

One cause is involved in

divergences which are not supported by tradition, and in

traditional alternations which are neither divergences nor correlations.

Two causes may be involved in

divergences which depend not only on anthropophonic conditions but also on linguistic intercourse, i.e., divergences which are transitional between divergences in the strict sense and traditional alternations; and in

traditional alternations which are at the same time correlations, or psychophonetic alternations.

The combination of divergent, or neophonetic, with correiational, or psychophonetic causes, is impossible.

Change of causality comes about in the historical succession:

when divergences become traditional alternations;

when alternations of foreign origin become traditional alternations;

when traditional alternations become correlations, or psychophonetic alternations;

when correlations become traditional alternations.

b) The causality of certain alternations is either simple or complex.

One cause accounts for

pure divergences or purely neophonetic alternations, which are not affected by tradition and linguistic intercourse in general, and

purely traditional alternations.

Two causes account for correlations which depend, on the one hand, on tradition and, on the other, on individually formed psychophonetic relationships.

II. Classification of Alternations according to the Conflict of Different Tendencies

1. The conflict of tradition with individual needs and tendencies.

Such a conflict pertains to traditional alternations which are not at the same time correlations. Because of their very nature, correlations exclude the conflict between individual tendencies and tradition.

Such a conflict arises when divergences are about to emerge; the final establishment of a divergence marks the unconsciously won victory of individual peripheral tendencies over tradition and linguistic intercourse in general.

However, there are divergences in which the individual tendencies not only do not contradict, but are supported by tradition (e.g., s ║ś in Polish Kostka | kość). Such a state is transitional between divergence in the strict sense and a traditional alternation which, in time, produces the above-mentioned conflict.

In the case of traditional alternations, tradition supports the complexity of phenomena and taxes the memory, while the individual tendencies strive to reduce this complexity and to relieve the memory.

Conversely, in the case of emerging divergents, tradition lends simplicity and unity, while the individual tendencies introduce previously nonexistent distinctions.

Finally, in the case of correlatives, there is agreement and harmоnу between tradition and individual tendencies.

2. A conflict between individual anthropophonic or peripheralphonetic tendencies with individual central-psychological tendencies must necessarily appear in divergents which differ because the phonetic intention of one of them could not be realized. In the case of other alternations there can be no such conflict.

III. Classification of Alternations according to their Genesis, or Distance from their Causes

Strictly speaking, we consider here not only clear and distinct alternations, but also incipient (embryonic) and residual alternations. Accordingly, we shall distinguish (a) embryonic, (b) live, and (c) petrified alternations.

1. Classification of alternations according to their distance from the source of anthropophonic causality.

Here we shall distinguish the following stages:

a) embryonic alternations, due to individual tendencies;

b) alternations that develop both individually and socially;

c) developed alternations;

d) established alternations;

e) transitional alternations that are gradually receding and are felt to be obsolescent;

f) petrified, residual alternations.

Stages (c) and (d) are preserved mainly through tradition and social intercourse in general; stages (e) and (f) persist as a result of the conflict between individual and social tendencies, in which the former gain the upper hand.

Stages (a) and (b) represent divergences, or neophonetic alternations; (c), (d) and (e) represent traditional or psychophonetic alternations; a variety of stage (d) forms correlations, or psychophonetic alternations; stage (f) represents an alternation only in retrospect, for by itself (from a synchronic point of view) it lacks the kind of phonetic difference which is required to view it as an alternation proper.

2. Classification of alternations according to their distance from the source of psychological causality.

Here we shall distinguish the following stages:

a) incipient alternations; paleophonetic alternations which are beginning to be utilized for psychological distinctions; alternations whose psychophonetic associations are rudimentary. Each member of a speech community must accomplish for himself this process of associating concepts;

b) established, i.e., clear and distinct alternations;

c) disappearing alternations;

d) obsolete alternations.

The last three stages, (b), (c), and (d), are due to both individual tendencies and social factors.

3. Classification of alternations according to the distance from a source originating in a related language.

Here we shall distinguish the following stages:

a) borrowing of foreign alternations by single individuals of a given speech community;

b) individual as well as collective adaptation of foreign alternations and their estrangement from the original source;

c) ultimate establishment of the foreign alternations (in the borrowing language) and their preservation through tradition and social intercourse in general.

Established alternations of this category as subsequently subject to the same changes or loss as the other types of alternations (III. 1 and III.2).

With regard to the last type of alternations (III. 1-3), one must mention the role of turning points in the history of a language which affect alternations no less than other phenomena of language. Such a change takes place most conspicuously when the phonetic representations and peripheral activities (i.e., activities involving the speech organs) and their corresponding acoustic results form new combinations which differ <. . .> from previous ones. <. . .>

IV. Classification of Phonetic Alternants and Alternations according to their Etymological Relationship

The phonemes which constitute alternating pairs must always be etymologically related, that is, they must originate from a single proto-phoneme; but this relationship can be of a dual nature:

a) either the phonemes alternate in etymologically related, and hence alternating, morphemes (e.g., g║źin Pol. mog- | moź-, etc.);

b) or the phonemes alternate as components of whole groups of morphemes having a similar phonetic structure (e.g., Pol. i(y)║ e in the verbs wy-cin-a, wy- źyn -a, na-gin-a │ roz-bier-a wy-cier-a, po-źer-a . . . ; e ║ о in piek-ę, ciek-ę, strzeg-ę, grzeb-ę | bior-ę, wiod-ę, nios- ę, plot-ę . . . ; Lat. i ║e in col-lig-o, con-tin-et, ab-ig-it, af-fic-it ... │ con-fer-o, at-ter-it, im-per-at . . . ; Goth. i ║ e(r)...).

As to the degree of etymological relationship between the alternating phonemes, the two types of alternations differ only quantitatively, not qualitatively. The morphological relationship between the morphemes, i.e., their relationship as members of certain morphological categories, is simply a result of generalization. This generalization is based not only on the comparison of a whole series of morphologically related morphemes, but also on their etymological relationship. Thus the comparison of the Polish forms

leads us to establish an alternation which can be regarded as an abstraction of all of these cases, namely to the alternation o 11 e in

The comparison of morphemes according to their morphological, formal relationship, and the discovery of phonemic alternations of such types as e 11 о in piek- . . . bior- . . . , has proved to be extremely fruitful in the recent historical studies of Indo-European phonology which were so auspiciously launched by Brugmann and de Saussure.

The distinction between two types of alternation according to the degree of relationship between the alternating phonemes is closely related to the question of the different directions of the morphological assimilation of morphemes, i.e., to the directions of leveling triggered by psychological factors. Two directions are most important:

a) assimilation according to etymological relationship, and

b) assimilation according to morphological or structural relationship.

V. Classification of Phonetic Alternations according to the Simplicity or Complexity of their Relationship

1. The distinction between equivalent and nonequivalent phonemes.

If one phoneme corresponds etymologically to another phoneme, or if two phonemes correspond to two other phonemes, their relationship is simple. But if one phoneme corresponds to two or more phonemes, then the alternation is one of nonequivalent phonemes. This type of alternation also includes the alternation of a specific phoneme with a zero-phoneme, i.e., the absence of a phoneme.

There are, furthermore, cases in which a particular phoneme alternates not with another phoneme, or with two or more phonemes, but with a part of another phoneme, or with a phoneme plus a part of a neighboring phoneme<...<

2. Classification of alternations according to the number of their constitutive members.

Ordinarily there are only two alternants in an alternation, as in the relationship of a unified causality. However, there are also some (though quite rare) cases when an alternation arising from a single cause, a single anthropophonic tendency, includes three or even four phonemes in the various degrees in which it is manifested. An example is the Russian divergence of vowels depending on the accent (ό ║ă ║ў (ə) in gód-, gód-a │găd-á │gўdăvój pół-gўd-a. . .).

3. Classification of alternations according to the simplicity or complexity of the morphemes containing the alternating phonemes.

A series of contiguous morphemes is opposed to a single morpheme. In such a case we either deal with simple morphemes, or we must compare the phonemes of two contiguous morphemes.

Examples of the first type have been cited above; as an example of the second type we can adduce the Polish alternation ći ║ c, dźi ║ dz in płac-i │ plac-ę, rodz-i | rodz-ę.

4. Classification of alternations according to the distinction between morphemes belonging to a single word and the etymological relationship of morphemes belonging to different words.

All the above-mentioned alternations occur in morphemes belonging to a single word. But if we were to look for a phoneme or a combination of phonemes which alternates, for example, with the final с of Polish dec, we would find that it consists of two morphemes which can never be combined: 1) cie k-, and 2) -ć in the infinitive:

The pertinent alternation is:

Thus, we shall distinguish two types of alternations and alternants:

a) simple ones, which involve sets of phonemes belonging to a single word (e.g., Polish {dźi ║ dz} ← {di ║ di)), and

b) complex ones, in which at least on one side, the phonemes belong to two different words <sic!>:

5. Comparison of simple alternations and simple alternating pairs with alternations of alternations, or of alternating relations.

Simple alternations require no further elucidation here. As an example of an alternation of alternations, we can adduce the Polish:

Alternations of alternations, or of alternating relations, are based on a purely formal or structural relationship of morphemes.

One could actually set up many more criteria for the classification of alternations. But I shall limit myself for the time being to those discussed in this chapter and shall proceed to a more detailed analysis of some alternating types.

CHAPTER III

_________________________________

ALTERNATIONS CLASSIFIED ACCORDING TO ANTHROPOPHONIC CAUSALITY.

THE ANALYSIS OF DIFFERENT TYPES AND THEIR CHARACTERISTIC FEATURES.

DIVERGENCES.

As is apparent from the forgoing remarks, every language has gone through the process of splitting originally homogeneous phonemes for purely anthropophonic reasons, regardless of their occurrence in one or another kind of morphemes or morphemic categories. For example, the original phoneme k in Polish has split into k, k, č (cz) and c. A special, psychologically conditioned case of phonetic splitting is the split of phonemes in etymologically related morphemes; for example, of k into k, č and с in the morphemes wilk–, wilč–,wilc- (wilk, wilczysko, wilcy).

In both cases we find a split of what constitutes, or at one time constituted, a psychological unity. But a phoneme considered apart from meaning-carrying morphemes forms a unity only as a phonetic representation, as an image imbedded in memory, while the phoneme as a component of a morpheme owes its psychological unity to the etymological connection of morphemes.

We must keep in mind another important fact: the conclusion that phonemes which are now distinct, without any apparent reason for their distinction, must have formed at one time a single phoneme is suggested only by etymology, i.e., by etymological comparisons within one language (alternations), or between two or more languages (correspondences). Where we cannot prove the existence of alternations or correspondences, we have no right to assume that one phoneme is derived from another.

One could speak of the splitting of a particular phoneme into several phonemes independently of the etymological connection of the morphemes containing them only if the causal factors are still at work, if they are still fully alive, if one can pin them down, so to speak, in flagranti.

Neophonetic factors operate independently of etymological relations. Phonemes or sounds of the same intention but different realization can thus be compared without regard to their etymological connections.

The Russian vowels a , e, and o, particularly a and e , exhibit different variants according to the following consonant (mat, mel, zakon ... │ mat, mеl’, kon. . .).

In literary German 5 varies its pronunciation according to its environment.

This kind of contemporary, live split of a psychologically homogeneous phoneme into two or more phonemes we shall call divergence. Divergence is of a purely anthropophonic phonetic nature; it is a divergence of the phoneme itself, regardless of its membership in related morphemes.

But when the differentiation is determined by the etymologically (psychohistorically) related morphemes, the result is a phoneticetymological divergence or a neophonetic alternation, i.e., a rudimentary alternation of morphemes and their constituent phonemes. Examples

Pol. s ║ ś [ kostka │kość, cząstka │ czçść, piosnka │ pieśń];

Pol. ń ║ ń (voiceless ń) ł ║ ł (vl. ł), r ║ r (vl r) [pieśni │ pieśń jabłek, jabiko, wiatru | wiatr . . .];

Pol. i₂ (y) ║ i₁((i) [głowy, cnoty ... │ postad, soli . . .];

Pol. -t- ║ -t (weakened t) [cnota | cnót];, . . .

Pol. ęn (en) ║ ęń (eń) [bçdç | bçdzie . . .];

Rus. o ║ o₁ a ║ a ₁,, e ║ е₁[vóza | vozit, bàba | bábe, ètot | èti. . .];

Rus. i₂(y) ║ i₁ ((i) [balý darý . . . │ koroli, cari . . .];

Rus. ό ║ á ║ ў [gód, góda | godá | godovój pólgoda . . .];

Germ, -b-║-p [Grabe, Stabe . . . │ Grab, Stab . . . ] ;

Germ. X_a (ch) ║ x₁ ((ch) [brach| bricht, Loch | Löcher. .. ].

In both the purely anthropophonic and alternating divergences there occurs either:

a) substitution, necessary adaptation to the conditions of pronunciation, or

b) in addition to adaptation, unconscious recollection of individual anthropophonic modifications of the psychologically homogeneous phoneme.

In the first case (a) there is a discrepancy between the intention and its anthropophonic realization: we desire to pronounce a given phoneme with all its properties, but we are able to pronounce only the modification of this phoneme, substituting possible characteristics for intended ones.

The proof of this lies, first, in the orthography, and second, in the frequent discussions concerning the actual quality of a sound perceived in a particular case. The perception of educated people is often influenced by orthography.

For example, the expression <German>z is pronounced like s is in some sense justified: the z represents the imagined psychological unit, but the s represents its implementation on the linguistic periphery.

An example of the second case (b) is the Polish s in kość, gość, pieśń, pieśni . . . , s in śpi (also pronounced spi) in which adaptation and substitution are supported by the unconscious recollection of individual characteristics.

The unconscious recollections of individual characteristics occurring in etymologically related morphemes represent a transitional stage between divergents and traditional alternants.

The anthropophonic splitting of psychologically homogeneous phonemes, which accompanies purely anthropophonic or alternational divergents, is caused by:

a) the development of different properties in one member of an alternating pair or, rather, through the substitution of some properties in place of others [ -t- ║-d- in rad | rada, s ║-z- in mróz | mrozu, ŗ ║r in wiatŗ | wiatru, m ║m in mehu | mech . . .], or

b) the weakening of the autonomy of one member of the alternating pair.

Here we must distinguish between conditions which favor the manifestation of the properties of a given phoneme and those which prevent it. Thus, for example, the position of t in ta , sta is favorable for the manifestation of the properties <of t>, while in at it is unfavorable. The same holds true of r in ra, rydz, a r as opposed to rda, rdza . .., of l in ligać as opposed to lgnąć, and of m, m in my, migać as opposed to mgła, ragnie. . . .

The cause of any particular divergence is either universal or ethnic, i.e., spatially and temporally conditioned. In other words, the combinatorial-anthropophonic changes, which provide an impulse for the splitting of a phoneme, come about either

1) as a result of continuous, one might say, eternally acting forces, or

2) as a result of transient, temporary forces which act only in a particular period of a given language.

In the latter cases these forces operate

a) synchronically, or

b) they belong to the past of a given language.

Within both the purely anthropophonic and alternational divergences we must distinguish:

1) phonetic habits (e.g., beginning any syllable with a consonant, which accounts for, among other things, the use of the so-called prothetic consonant to avoid hiatus), and

2) accommodation, for the sake of easier pronunciation.

This latter is, in turn, either

a) mandatory and without exceptions in normal, correct pronunciation (e.g., the divergence of t in ta │tr │ tl │ ts │tn │ at | ant. . . ; the alternation of d in dno | den . . .), or,

b) reduced to a weak anthropophonic tendency, a tendency curtailed by prohibitive analogy, i.e., by the striving toward phonetic leveling of psychologically different morphemes.

Features of Divergence

1) The first and basic feature of any divergence can be formulated as follows: the alternating phonetic properties are not individual and independent properties of the anthropophonic variants (modifications) of a given phoneme in a given phonetic position of a morpheme, but are conditioned combinatorially, i.e., they depend on the combination with other phonemes or, more generally, on the conditions of the anthropophonic environment.(. . .)

Two other features are closely allied with the first feature of divergence or with the alternations that are caused anthropophonically:

2) Direct definability (Bestimmbarkeit) and the presence of anthropophonic causes for the alternations, and

3) Generality and anthropophonic necessity of the alternation.

Which means that such an alternation

a) occurs without exception in all words and phonetic combinations containing the given phonemes in the given language, and

b) is causally connected with anthropophonic alternations in such a way that

x^’ may combine with y, and

x may combine with y, but

x . . .y or

x . . . y are impossible.

4) Divergence is independent of psychological (morphological or semasiological) factors. It occurs on the periphery of language, in the articulation of phonemic complexes which need not be specified syntactically or morphologically.

5) Since the properties of the phonemes x, x are not individual՝, psychological properties that are stored in the brain center, but conditioned variants which depend on the phonetic environment (y, y) rather than on psychological factors, the anthropophonic variations of the dependent phoneme may go unnoticed. These variations are established automatically by the peripheral phonetic conditions. The possibility of perceiving and marking these variations is, however, by no means excluded and may, in turn, establish a link between divergence and the traditional alternations.

Within the area of divergence or neophonetic alternations, we may, further, distinguish different degrees and varieties, such as:

a) embryonic alternations, which we can discover by means of minute, as it were, microscopic observation, or which we can simply postulate;

b) alternations with clearly definable consequences, i.e., alternations which cannot be perceived in a state of weakened consciousness, but which can be discovered in a state of sharpened or intensified consciousness [e.g., Pol. m ║ m in mech │ mchu, r ║ r in Piotra ║ Piotr, t ║ t in kota │ kot, Rus. a ║ a in brata │ brate ]. Here a psychologically homogeneous phoneme appears together with its anthropophonic bifurcation;

c) alternations which can be perceived not only consciously but also unconsciously, and which come about, on the one hand, from a split (bifurcation) of the psychological unity and are, on the other hand, supported by tradition [for example, Pol. -b- ║ -p in łba │łeb, p(p_{_i)}) ║ p in kupiec │ kupca . . . ; Rus. t ║ d in svatat’ │ svadba . . . ].

In deciding whether a given alternation belongs to type (b) or to type (c), we can examine the pertinent facts of language and of writing (the test of writing applies, of course, only to literary Ianguages and literate individuals). Thus Polish -b- ║ -p in łba │ łeb belongs to type (c), since there is also a form lepek <dim.>; so does Russian d ║ t in búdok | búdka in view of bútočnik, and t ║d in svatat │svádba in view of svádebnyj <adj.> The Polish spelling tchu <gen. sg.> tchnqc shows that d ║t in dech, oddech │ tchu, tchnąć belongs likewise to type (c). In the speech of Polish children who pronounce vrušek (wruszek) instead of vruzek (turótek <gen. pl.>), the ž ║š alternation wróź—yć │ wróź-ka also belongs to type (c) or, to be exact, is not yet established as a proper alternation.

The three types cited obviously admit in each language a series of transitional stages and oscillations in one direction or another.

CHAPTER IV

_________________________________

CORRELATIONS, OR PSYCHOPHONETIC ALTERNATIONS.

A correlation is an alternation of phonemes in which the phonetic difference is connected (associated) with some psychological difference between forms and words, that is, with some morphological or semasiological difference.

Strictly speaking, the alternation concerns not isolated phonemes (sounds), but entire morphemes, or even words.

At this stage of the development of alternational relations, morphology utilizes homogeneous phonemes (i.e., phonemes stemming from an originally single phoneme) as mobile correlatives in the same way as it utilizes mobile word-forming morphemes (affixes), i.e., prefixes, suffixes, desinences, etc. The mobile correiatives form a necessary link, an integrating part of some mobile homogeneous morphemes. And like the suffixes or prefixes, the correlatives may serve to distinguish morphological categories.

Thus, in Polish, as in other Slavic languages, one class of denominal verbs is still productively formed by means of adding the suffix i to the primary stem; the final consonant of the stem corresponds to the palatalized consonant which had developed historically by way of spontaneous degeneration at the time of the first Slavic palatalization (which applies at least to the velars k, g, ch): [brud- │ brudź-i–ć, łup │ łup-i-ć, tok- │ toč-y-ć, trwog │ trwoz-y-c..,.].

The locative case of Polish substantives is marked not only by the desinence -e, but also by the change of the final thematic consonant into a sound which developed by way of spontaneous degeneration from a palatalized consonant at the time of the second Slavic palatalization (which applies at least to the velars): [: narodź-e, wol-e, boř-e, strać-e, ręc-e, wódc-e, nodz-e, strudz-e . . .].

To cite other examples:

In the Polish conjugation, 1st p. sg. nios֊e (nios-ę), plot-e, gn-e, bior-e, piek-e, mog-e . . . 3rd p. sg. nieś-e, pleć-e, gń-e, bieř-e, pieč-e, mož-e . . . ;

also in the Polish conjugation 3rd p. sg. (2 sg., 1-2 pi.) {lubi mówi, w oli, rańi, twořy, točy, trwožy, sušy,} nośi woźi, świeći, chodźi │ 1st sg. (3rd pl.) lube, mów-e, wol-e, ran-e, twoř-e, toč-e, trwož—e, suš-e,} noš-e, wož-e, świeć֊e, chodz-e . . . ;

nośi-, woźi-, świeci֊, chodźi-│noš-enie, noš-ony, wož-enie, wož-ony, świec-enie, świec-ony, chodz-enie, chodz-ony . . . <verbal nouns and past passive participles>;

simple verbs vs. duratives and iteratives: pali-ć, czyńi-ć_ttrudźi- ć ן pal-a-ć, -czyń-a-ć, ֊trudz-a-ć . . . , but stroi-ć, toćy-ć, trwoźy-ć, chodźi-ć, mnoźy-ć | -straj-a-ć, tać-a-ć, trwaź-a-ć, chadz-a-ć, -mnaź-a-ć and gnieś-ć, pleś-ć, mieś-ć, leć-e-ć, siedź-e-ć | gniat-ać, plat-a-ć, miat-a-ć, lat-a-ć, siad-a-ć . . . ;

the same relationship holds in Russian verbs: bros-á-t’ | brás-yva-t, koló-t’ │ kál-yva-t, strói-t | -stró-iva-t, (strá-iva-t’), próči-t, -proč -iva-t, Ijubi-tי | -Ijúbl-iva-t, dolbit │ dálbl-iva-t, xodi֊t │xáz-iva-t, nosi-t | nášiva-t, zapodózri-t | zapodózriva-t’ (zapoddzr-iva-t’). . . .

The nom. pl. mase, of Polish nouns (also adjectives and past tense of verbs) in combination with the desinence -i (-y) has a final consonant which underwent palatalization (in the case of the velars, palatalization of the second period) by way of spontaneous degeneration [chłop-i, kać-i . . . , silń-i, mil-i . . . , wilc-y, ptac-y . . . , wielc-y, drudz-y . . ., szl-i, chodzil-i, dal-i . . .].

In Modern High German the plural of certain masculine and neuter substantives is formed not only by adding the desinence -e or -er, but also by changing the nonpalatal stem vowel into a palatal (by so-called Umlaut): Wolf, Dorf, Grab, Loch, Wurm | Wölfe, Dörfer, Gräber, Löcher, Würmer. . . .

In ancient Indie <Sanskrit> some derived names are formed simultaneously through the addition of the suffix -ya- and the correlative change of the simple vowel of the stem morpheme into the second degree of the vowel gradation (vrddhi-): kāunteya-, sāubhagya, vāirya, pārthava- . . . from kuuti-, sub haga, vira, prthivï־־־. . . .

As is well known, some suffixes or subordinate morphemes (suffixes, prefixes) can give a word a nuance of coarseness, abstraetion, etc. A word can acquire similar nuances through a correlative relation of morphemes, as, for example, in the correlation x (ch) ║ 5 in Polish włoch-y, kluch-y │ włos-y, klus-ki . . . ; Russian ra, la ║ oro, olo ; and šč, žd || č,ž as in graždanin | gorožánin, glavá | golová, prevraščát │voróčat . .. ; French k ║š as in cause │ chose . . . (in this last case one could hardly speak of an actual, live correiation that is still perceived by the speakers).

The so-called vowel gradation found in proto-Indo-European and in the older stages of various Indo-European languages is closely connected with a correlative alternation of phonemes in which one member of the alternating pair has a zero-phoneme, i.e., lacks any phoneme:

This Vowel gradation may continue to be alive and productive, or its productivity may be waning.

In this context we can also mention the so-called infixes, which are widely used in the Semitic languages.

The possibility of forming new alternating pairs may serve as a proof of the productivity of correlative alternations, particularly when such alternations could not have arisen by way of simple phonetic change. This is the case of the c ║ č (cz) correlation which is used in the formation of Polish diminutives from simple substantives. Having developed phonetically in words where both č (cz) and с were the result of various palatalizations of k in different periods of the language, the correlation is now also found in words in which the с goes back to a ti , and not to a k. The correiation с ║č (cz) was caused by a phonetic development in the words donica miednica, krynica, lice, słonce, kupiec| doniczka miedniczka, kryniczka, liczko, słoneczko, kupczyk . . . , just as the correlations k ║ ć, g ║ ž, x (ch) ║ ś (sz) in ręka │rączka, noga │ nóźka, mucha | muszka. . . . But the correlation c 11 с in correlative pairs of the type świeca │świeczka . . . was brought about by morphological assimilation (analogy).

The older nom. pl. of the possessive pronouns nasz-y, wasz-y, has in modern Polish been replaced by naś-i waś-i, because the consonant s is felt to be a characteristic feature of this case for stems ending in the consonants s,š (sz) or x (ch): naš | naś–i. . . .

The older diminutive forms of the Polish grosz, arkusz were groszyk arkuszyk (as they are still in some parts of the Polish linguistic territory, e.g., in Lithuania, the Ukraine, etc.); at present these forms have a palatal ś as in arkuśik, grośik. (A similar palatalization is used in childrens speech to express endearment and affection.)

The vowel alternations of Sanskrit which are subsumed under the general term guna–, were originally mostly correlative, i.e., morphologically mobile; in the period from which Indie literature dates, this mobility began to change into the psychological immobility of a purely traditional alternation. The so-called vrddhi- on the other hand, became a living, mobile, and productive correlation.

Correlations are always in a state of transition between one type of simple traditional alternations and another. A correlation arises only as a result of the utilization of alternational distinctions for psychological purposes, and this utilization may be repeated generation after generation until it ceases; and when it ceases, the psychophonetic alternation or correlation is converted into a simple traditional alternation. We shall illustrate this general statement with some examples.

The Polish alternation Ø ║ i (ń ║ in) in pn-ę, pń-e | pin-a; –čn-ę, čń-e │ -čуп-a; tn-ę, tń-e | -ćin-a . . . was at first a mobile correlation, as shown by its extension to the pair gn-ę gn-e [ gin-a . . . ; but at present it is shifting into a state of total psychological immobility, as are the other alternations closely related to it (čt- ę, čc-ečyt-a, sła ć| sył-a֊c,tk-a- ć| tyk-a- ć, br-ać| ber-a ć.

The category of rudimentary psychophonetic alternations also includes the Polish о || u(ό), as in chod-u, bor-u, stoł-u, grod-u | chód, bór, stół, gród . . . ; ę, ą|| ę, ą, as in ć ęž -kićąž -a | tęg-i ws-tąž-ka; u|| Ø as in such-y | sch-ną- ć. . . ; German i (e) || a in bind֊e ן band, ess-e | aβ . . . ,

We can assume that even the Polish alternation o || e in bior-e, nios֊e, wioz-e | bierz-e, niesi-e, wiezi-e . . . was once felt to express a relation of different forms or in other words was a true, albeit short-lived, correlation. Now, of course, it shows no trace of correlativity, for otherwise there could not be any tendency to level the stems and to change such forms as bior-e _ג nios-e, wioz-e . . . into biere , ni e s-e, w ie z-e.

There are also cases in the history of a language when a particular correlation seems to disappear, but in actuality only changes its outward appearance to be subsumed under a broader, more general correlation.

An interesting example of such a change is found in Russian, where the alternation k | | č, g || ž in the present tense of verbs of the literary language, e.g., p _ieek-ú, t_ieekú , b_ieer_ieegú, st_ieer_ieegú . . . p_ieeč -ót, t_iiieč-ót,b_ieer_ieež -ót, st_ieer_ieež-ót ... is replaced in dialectal spech by the alternation k || k _i g || g _i,, e.g., p _i iiek-ú, b_i eer_i e egú . . . |p _iek _i-ót b _ier _ieg _i-ót ... on the model of the alternations:

b|| Ь_i , t || t _i d || d _i s || s _{_i,}, z )( z_i r || r _{_i,},n||... (in the forms gr_ieeb-ú, pl_ieet-ú, v_ieed-ú, n_iees-ú, v_ieez-ú, b_ieer-ú, gn-ú . . . gr_ieeb -ót, pl_ieeti-ót, V_ieed_{_i-}-ót, n_iees_{_i-}-ót, v_ieez_{_i-}-ót, b_ieer_{_i-}-ót, gn_{_i-}-ót. . .). All these alternations can be reduced to the single correlation:

{where P indicates palatalization, softening, Ø ֊ absence (in this case, of palatalization), and Y presence (in this case, of palatalization)}.

If one considers this process of substitution of the alternation k \\č,g\\žby the alternation k || k _i,g ||g_i superficially, one might conclude that a correlation, or psychophonetic alternation has been lost. Such a conclusion would, however, be misleading. One could speak of the loss of a correlation only if, instead of the pairs k || k _{i g} || g _i,we found the pairs k || k, g || g, i.e., if all forms of the present tense were leveled. But such a leveling should also have affected the other consonants, so that we would find not only the forms p_iek-ót, t_iek-ót, b_ir_ig-ót, st_ier_ieg-ót, . . . but also gr ₁eb-ót, pl_iet-ót, Vied-ót, n_ies-ót, Viez-ót, b_ier-ót, gn-ót, .... Actually the situation is quite different, and the development of the forms p_iek_i-ót,t_iek-ót, b_ier_iegi-ót, st_ier_ieg-ót . . . testifies to the great vitality of the correlation PØ || PY in the present tense of this type of verbs.

Characteristic Features of Correlations and Correlatives

1. From the viewpoint of anthropophonic causality—the alternating features concern the various points of articulation or the phonemes, individually, autonomously, and independently.

2. At any given stage of a language, the cause of the phonetic alternation lies only in tradition (transmission), in social intercourse; in usage. We have learned to speak in a certain way from our environment and our ancestors; such an explanation is completely sufficient.

3. The anthropophonic causes of an alternation, its anthropophonic causal connections, lie in the history of the language and can be established only through historical-linguistic studies. At one time an anthropophonic cause was at work, but later it ceased to operate, and now it is absent.

(These three characteristic features are common to correlatives with simple traditional alternations.)

4. As a result of the ever-recurring process of associating representations, each alternation of this sort was endowed with psychological differences of a formal morphological or semasiological, meaning-carrying nature. <...>

Thus we see that the correlatives, or psychophonetic alternants, carry corresponding psychological differences. Anthropophonic distinctions and nuances are here always accompanied by psychological, morphological, or semasiological nuances and distinctions.

5. From what has been said above, it follows that it is the nature of the psychological, morphological, or semasiological correlatives to be general and without exceptions.

A particular correlation encompasses all words of a given category without exception, e.g., within the conjugation all verbs of a particular type, within the declension all nouns of a particular type, a particular type of derivatives, etc.

6. The degree of phonetic similarity of the alternating phonemes is in such cases completely immaterial. It is only necessary that there be a psychophonetic association of the representations of particular anthropophonic activities with the corresponding psychological distinctions. <...>

7.The seeming phonetic changes which occur in the area of correlations are as a rule not gradations, not a movement in a particular anthropophonic direction, but leaps which from an anthropophonic point of view are totally incomprehensible and which are often in contradiction with the general course of historical-phonetic changes.

A striking example of this type is the previously mentioned replacement of Russian p _ieeč-ót, b_ieer_ieez-ót... by p _i eek_{_i -} -ót,b _ie-r _ieg_{_i -} -ót... or the use of the Russian imperative p _ieek_{_ií}í, b_ieer_ieeg_{_ií}í, pomog _i-í . . . instead of the earlier р _{_iе}есí, b_ieer_{_iе}еzí, pomozí. .., or the Polish piecz, pomóż ... in place of the earlier *piec(piec-y), *pomódz֊ (pomodz-y). . . .

8. Neither the divergents nor the traditional alternations tolerate a change of the alternational correlation or innovations of a particular type. However, one of the characteristic features of the correlatives is the creation of innovations according to a certain model, the possibility of transferring a ready made correlative relation to new words, the possibility of the incessant reestablishment of the relation.

The transference of a correlative relation can occur

a) within a group of words of a certain type (semantic or lexical transferability), or

b) within a specific morphological category.

9. [The feature which arises from the generalization of a whole series of correlatives.] In the case of divergents, some causal relations are universal. In the case of correlatives, only the ability to form correlative relations is universal; the specific types of correiatives are conditioned temporally and spatially.

10. [The genetic feature, the feature which characterizes the formation of correlatives in individuals.] Each member of a speech community arrives at the divergents by himself directly, by means of immediate, physiological accommodation, whereas thecorreiatives are acquired step-by-step, gradually, in proportion to the strength and intensity of psychophonetic associations already formed in ones mind.

In this context it is necessary to indicate that the correlations developed from traditional alternations within one and the same language differ often in their semantic function from those correiations in which at least one member has been borrowed from another related language.

In correlations of the second kind the borrowed member of the alternation generally has a more abstract, loftier, more literary or more solemn meaning than the member formed within the native language whose meaning is usually more concrete and colloquial. Thus, for example, Polish h || g in hańba | ganić, hardy | gardzić . . . (if, indeed, these pairs can be treated as examples of a developed correlation);

Russian ra || oro, ła ||0ł0, re || ere, le ||oło, ra|| ro-, šč || č, žd ||ž, о ||Ø in grad, graždanin | górod, gorožánin; glavá, glávnyj | golová golovnój; prédok | peredók; plen, plenítי | polón, polonií;rázum | rózysk; rávnyj | róvnyj; rab, rábskij | róbkij, robét/ osveščát, prosveščénie | svečá , prosvéčivat; čúždyj | čužój; roždát | rožát...;

French k || š in cause | chose, caniculaire | chien, camp | champ ... (if again, these represent at all a developed correlation).

As a very rare example of conscious and arbitrary interference with the spoken language in forming a correlative, or psychophonetic alternation, I shall report the following case which I myself observed. In a Slovenian school in the province of Goricia, in the Karst region, the teacher required the children to replace the wordand syllable-final consonantal u () characteristic of that dialect (as of almost all Slovenian dialects) with I in conformity with the spelling, not only in reading but even in speaking; hence, dàl, bíl, prósil . . . instead of dà ,bí, prósi,. . . . But since this dialect has an ł (resembling the Polish, Russian, or Lithuanian ł) in wrd-initial and word-medial position, the ־ł־ || - correlation peculiar to the native dialect was in the speech of the school children replaced by the correlation ł || -l-, i.e., the pairs dàł a | dà, biła | bi, prosíła | prósi, déłała | déla were replaced by dała | dàl, bíła | bil, prosíła | prosíl, déłała | déłal. . . .

A correlation introduced so artificially could obviously not endure and was in time replaced by the normal correlation -ł|| -. However if one did not know the local dialect, he might have coneluded on the basis of the speech of these children that the dialect had, in fact, the correlation ł || l It is almost certain that if the teachers had continued to demand that the children pronounce I in place of for a number of generations, the alternation ł || l would have eventually become an established fact of that language.

Finally, we might mention the various layers of alternations, which go back to different periods in the history of a language and which replaced each other as psychologically mobile, psychophonetic alternations, or correlations. Several such layers can be distinguished in Polish: (a) alternations of proto-Indo-European origin, (b) alternations of Common Slavic origin, and (c) properly Polish alternations of more recent origin which have, in part, retained their full psychological vitality.

For with time, correlations may lose their psychophonetic value and become purely traditional alternations that could be compared with extinct volcanoes.

Analytic languages, whose morphology shows a tendency toward decentralization, lack morphological correlations.

CHAPTER V

_________________________________

TRADITIONAL ALTERNATIONS.

The phonetic and psychological causes of alternations are, like those of other linguistic phenomena, both individual and social. They are social in that they are not individually phonetic or individually psychological, but collectively phonetic and collectively psychological. Their social character is, nevertheless, secondary, since they are not the result of imitation and repetition, but of psychological organization, and serve the psychological needs of each individual belonging to a given speech community.

Although both the divergents and correlatives must be examined from the viewpoint of traditional and social causality, the characteristic feature of divergents is their anthropophonic causality, and of correlatives, their psychological causality, whereas in the case of purely traditional alternants we can speak only of traditional and social causality.

As has been mentioned above, the original cause for the emergence of all alternants is always purely anthropophonic.

Hence it follows that purely traditional alternants are never original but are the historical continuation of neophonetic alternants or divergents, whose anthropophonetic cause has ceased to function as a living factor and belongs now only to the past.

Consequently, all traditional alternations are simultaneously paleophonetic.

Let us illustrate this otherwise simple statement with a few examples:

Polish ród | rod-u, in which -t(d) || -d is a neophonetic alternation, or divergence, but u(ó) || о is a traditional, paleophonetic alternation.

mróz | mroz-u: divergence -s(z) || -z-; traditional alternation u(ó)\\o;

mąż | męż-a: divergence -š(ž) || -ž-\ traditional alternation ą || ę;

plot-ę| pleś-ć: о \\ e and t || ś are both traditional alternations;

płaci| płacę, rodzi| rodzę: only one traditional alternation, ćí || с, dźi || dz;

Literary German geb-en | gab : two divergents, g _i|| g and -b|| -p (b); one traditional alternation, e ā a;

lad-en | Las-t : two traditional alternations, ā || ā, d || s (d as occlusive, s as spirant), and one divergence d || s (if one considers these phonemes from another point of view, viz. d as voiced, and s as voiceless);

Fros-t | frier-en, Ver-lus-t | ver-lier-en, ver-lor-en: the traditional alternation s || r.

Fros-t| frier-en: the traditional alternation ǒ || ī (ie);

Ver-lus-t| ver-lor-en: the traditional alternation u || o;

verlier-en | Ver-lus-t, ver-lor-en: the traditional alternation

Sanskrit nāmn| nāma: the traditional alternation n || a.

From the point of view of their origin and historical evolution, the traditional alternations fall into two main groups:

1) those that developed directly from divergents, or neophonetic alternations that have lost their live anthropophonic cause;

2) those that have gone through several stages; at first the divergence became a traditional sic! the author obviously means neophonetico alternation which was, in turn, utilized as a correiation that subsequently lost the ability to express psychological distinctions and became a plain traditional alternation.

1) Examples of traditional, paleophonetic alternations, which are derived directly from divergents of neophonetic alternations:

a) alternations which had developed as a result of palatalizing or dispalatalizing accommodation, i.e., softening or hardening-

Polish š|| x (ch) [szed-ł chodz-i. . .]

ć \\ t [ciek-, ciec | tok; cież-ki, ciąż-a | tęg-i, ws-tąż-ka . . .], b \\b’ [br-ać \ bior-e, bierz-e . . .],

о \\ e [bior-ę | bierz-e, nios-e niesi-e ...],

a ||e [świat | świec-i . . .];

b) the alternation found, among others, in Latin, in some German-speaking areas, and Čuvaš:

c) the alternation of a phoneme with zero, i.e., with the absence of a phoneme; e.g., Polish e \\ Ø [pies | ps-a, sen | sn-u . . .],

2) Examples of traditional alternations which have gone through three stages of evolution (from divergents to traditional alternants, from traditional alternants to correlations, and from correlations back again to traditional alternants).

The Indo-European alternation e || о which for some time marked the difference between primary verbs and certain types of nouns and which is preserved until now, for example, in Polish as

The Polish and other Slavic alternations Ø || i, Ø || у [pn-ę | pin-a, czt-ę| czyt-a. . . , tk-a | ty k-a, tch-nę | dych—a. . .].

The Polish alternation o || u (ό) [chod| chód. . .].

The alternation in the conjugation of the literary German strong verbs: e.g., i, e || a [bind-e | band, geb-e | gab. . .],

[bind-e, band | ge-bund-en; werf-e, warf | ge-worf-en. . .].

In view of the fact, emphasized above, that the correlations characteristic of a given speech community are acquired through the individual effort of its speakers who often fail to develop the feeling (awareness) of a correlation, it would appear that all the correlative or psychophonetic alternations mentioned in Chapter IV lack, for such speakers, a correlative character and fall into the category of purely traditional alternations. (. . ...

Features of Traditional Alternations

1. The alternating properties are associated with certain points of articulation, i.e., with specific phonemes, individuality and autonomously. <. . .>

2. It is then apparent that at any given stage of a language, tradition (transmission), social intercourse, or usage are the sole causes of an alternation. We have learned to speak in a certain manner from our environment and our predecessors; this type of explanation is completely sufficient. The preservation of a particular alternation cannot be ascribed to any indivdual factor.

3. The anthropophonic causes, the anthropophonic causal relations of an alternation lie in the past history of a language and can be discovered through historical-linguistic studies. An anthropophonic cause was operative for a certain time, after which it ceased to act.

The above three features apply to simple traditional alternants and the correlatives.

4. Psychological associations, which support the preservation of traditional alternations, are in constant conflict with the tendency to eliminate phonetic distinctions that are justified neither by individual anthropophonic tendencies nor by individual psychological needs. Such a conflict leads either to the introduction of a correiative meaning in traditional alternations <. . .> or to the elimination of differences, to leveling <. . .>

CHAPTER VI

_________________________________

FOREIGN ALTERNATIONS, i.e,

ALTERNATIONS WHICH ARE DUE TO THE

INFLUENCE OF ANOTHER LANGUAGE.

The correspondences or formal equivalents of different languages are not used by the same speakers and lack a common psychological substratum, except for the special case of bilingualism, when one and the same person speaks two (or more) languages. But even outside such cases, a correspondence can become not only a historical but a live, psychophonetic phenomenon. This happens when a language has borrowed words from historically related neighboring languages as, for example, when Polish has borrowings from Russian dialects (and vice versa), Lithuanian from Slavic (and vice versa), or Latin from Italic. These borrowings may be due not only to geographical proximity, but also to literary or cultural proximity. Such are, for example, the borrowings from Latin into French and other Romance languages or from Old Church Slavonic into Russian and other Slavic languages.

In this way there arises a live phonetic correspondence, which is perceived as such by the speakers. But in addition to producing an awareness of correspondences, borrowings may also account for introducing:

a) paleophonetic alternations from a closely related language;

b) a live relationship between native and borrowed forms of etymologically cognate morphemes; such a relationship might be called an alternation of correspondences (or correspondencesimultaneity).

It is clear that such borrowings and exchanges are ordinarily found only between closely related languages, such as Polish and Czech, Polish and Ukrainian, Serbian and Bulgarian, Russian and Church Slavonic, French and Latin, Latin and Italic dialects, etc. For if there is no close linguistic relationship, we can hardly expect to find a sufficient number of morphemes having a constant etymological or psychophonetic relationship, which is required to produce in the borrowing speech community an awareness of a thorough-going correspondence between the respective languages. And the existence of such an awareness is an absolute prerequisite for recognizing this type of alternation.

An example of a totally foreign alternation, i.e., of an alternation in which both members are of foreign origin is the Polish alternation h || z [in błah-y | błuz-en. . .], which was borrowed from Czech.

As examples of mixed alternations, i.e., of alternations in which one member is of native and the other of foreign origin, we may cite the Polish g || h [gan-ić | hań ֊ba, gardz֊ić | hard-y, błog-| błah-y], Polish ł0 || ła [błog-i | błah-y, błaz-en. . .], Latin b|| ƒ [ rub-er, ruf-us. . .], French š || k [chose | cause, champs | camp. . .] and other such doublets which are partially felt by the speakers themselves to be etymologically related and which can partially be established only through scientific, theoretical study.

Different types of alternations which are due to borrowing can be studied to great advantage in literary Russian, since this language was for a long time subject to strong Church Slavonic influence, to which it owes a significant number of alternations.

In accordance with what was said above, we must distinguish first of all two basic types of alternations occurring in Russian:

1) alternations which are entirely of Church Slavonic origin, and

2) mixed alternations, which are half Church Slavonic and half Russian.

The alternations of the second type are far more numerous than those of the first type. The phonetic quality of a mixed alternation in which one of the members is of Church Slavonic and the other of Russian origin may sometimes give rise to uncertainty as to whether one member of the alternation or both members are of Church Slavonic origin.

Russian alternations which are entirely borrowed from Church Slavonic are:

a) ti|| šč, di|| žd in o-svetít, | o-svešč -át, rodit, | rožd-át\ ....

The corresponding native alternations are ti || č, di || ž in sveti-t|| sveč ú, rodít | rož -dt. . . .

In both types of alternations, the first members are identical, but the second ones are different. The properly Russian šč, žd come from another source and participate in another type of alternation: sk|| šč [isk-át | íšč -et, pisk| pišč -ít. . .], sti || šč [pustí-t |pušč -ú, svist֊ít| svišč ֊ú.. .]; žĭd žd,žd || žid [žd-át | ožid-át ... vražd-á vraž d-a. . .].

The alternations borrowed from Old Church Slavonic also include:³

b) sk|| st in blesk | blist-át_yblest֊ét. . . ;

c) er || ra in mérz-kij | mraz,smerd-ét | smrad, vert-ét | vrat-á, vrat-it, vrašč -át. . . ;

le|| ła in vlek-ú, vleč , -vlek-át | vlač-it ób-lak-o. . . .

Examples of mixed alternations in Russian, i.e., in which one of the members is of native origin and the other a borrowing from Church Slavonic are:

a) g || γ (h) in gospod-in | уospòd, gosudár | γ osudár, bogátyj | bóγa. . . .

One should not conclude from this example that γ (i.e., a voiced x (сh) or spirant (g) was the original Church Slavonic pronunciation. On the contrary, since g is used by the Slavs of the Balkan peninsula where the Church Slavonic literary language originated, we may assume that this pronunciation was also characteristic of Old Church Slavonic. But Church Slavonic penetrated into Great Russia through Kiev, or more generally through the Ukraine, and the Ukranian clergy and scholars have in this case imposed their pronunciation upon Church Slavonic, i.e, they pronounced the original letter γ as γ (h). Sanctioned by the ecclesiastical academy of Kiev, the one-time center of Orthodoxy, this pronunciation became the standard obligatory in all Russia, including the Ukraine and Great Russia. This is why the Russian Orthodox clergy adopted this γ pronunciation in reading, and why it is still maintained in Russian words of Church Slavonic origin that are still perceived as borrowings.

The g || γ alternation can, of course, be treated as an alternation only in the speech of those Great Russians who use g in their native pronunciation. But in Byelorussia and in the many other areas of the Great Russian linguistic territory where only (h) is pronounced, the alternation is lost without a trace.

b) ó || é in those cases where an ó should have developed in Great Russian and where only spelling and a desire to pronounce the letters in accordance with their names in the alphabet account for the é pronunciation in words of Church Slavonic origin, and for the alternation ó || é (where ó is of Russian and è of Church Slávonic origin); e.g.,

in suffixes: -óž || -éž [grab-ëž, pad-ëž | mjat-éž, pad-éž. . .]; óv-a || év-a [proper name Korol-ëva | -korol-éva]՛, -ónn|| -énn[počt-ënn-yj, soverš-ënn-yj, osvjašč-ënnyj, vljubl-ënn-yj. . . | počt-énn-yj, soverš-énn-yj, preosvjašč-énn-yj, nezabv-énn-yj... ] : -ó || -é -e [֊tjó || -t ĭjé and -njo, ֊enjó | ֊nie, -en ie: žitë-bytë ži-tié, by-tié, vra-në, žra-rië lga-nё pisá-nie, poslá-nie, zaklánie, vved-erë (dialectal) | vved-énie, javl-énie, voznes-énie, vozdvĭž-enie. . .].

In the last case, -ó|| -è, -e [-tjó -ti jé, -njo, -enjó-nje, -énje], -0 appeared in Russian not by way of a phonetic process, but by way of morphological assimilation (analogy) to other nouns of this type. In all other cases the Russian -ó developed from a short e (corresponding to Common Slavic ĕ or ĭ) before a hard or nonpalatalized consonant.

c) č || šč, ž || žd:

in the suffixes: -uč-, -ač| -ušč-, -ašč-: drem-úč-ij, plov-úč-ij, pax-úč-ij, kip-úč-ij \ dréml-jušč-ij, plyv-úšč-ij, kip-jdšč-ij . . . , gor-júč-ij, kol-júč֊ij, von-júč-ij \ kól-jušč-ij, vonjá-jušč-ij . . . , gor-jáč-ij, vis-jáč-ij, sto-jáč-ij | gor-jášč-ij, vis-jášč-ij, stojášč-ij . . . ;

d) -oro\\ -ra-, oło|| -ła-, -ere|| -re-, -oło-(-ele֊) || -le-, ro\\ ra-: vorot\\ vrat-, gorod\\ grad-, norov\\ nrav-, storon\\ stran-, porox\\ prax. . . , golov|| glav-, molod|| mlad-, volok\\ vlak-, solod\\ slad..., berem\\ brem-, bereg\\ breg-, vered|| vred-, sered|| sred-, pered|| pred-, pere|| pre-, čered|| čred. . . , volok|| vlek-, molok|| mlek-, polon|| plen. . ., rab\\ rob-, rov\\ rav-, roz\\ raz-, rost\\ rast. . . ;

e) Ø \\ о in s-bor \ so-bór, v golové, v glavé | vo glavé.

The mixed alternations could have formed by virtue of a mental process reminiscent of the mathematical formula: two entities which are equal to a third entity are equal to each other, or two entities which are similar to a third entity are similar to each other.

Let us compare, for example, the native Russian alternation ti || č [vărót ǐ \\ vȳ-răč-ú... (voroti| voro č-u)] and the alternation ti \\ šč of Church Slavonic origin [sȳvrăt-í| sȳ-vră šč-ú . . . (sovrati| sovra šču)]. Both of these alternations share a common member, ti, which alternates with č in Russian and with šč in Church Slavonic. In accordance with the above mathematical formula we may say: two psychophonetic entities (that is, two phonemes or groups of phonemes) alternating with a third entity, alternate with each other. In the given case we obtain the bilingual alternation č || šč of which Russian makes, in fact, extensive use.

On the basis of the above analysis of alternations from the viewpoint of the history of inter-ethnic relations we may conclude that the paleophonetic alternations, whether they be purely traditional or correlative (i.e., psychologically mobile) form two types:

1) they develop within a particular language without any foreign influences, they are the result of the linguistic activity of one speech community; or

2) they arise under the impact of a closely related speech community from which a whole category of words has been borrowed containing phonemes that are part of a particular alternation.

The process of borrowing takes place (a) through oral transmission, or (b) under the influence of a foreign literature.

The alternations resulting from the influence of another language can be of two types: (1) either the entire alternation, i.e., both of its members are borrowed, or (2) only one of its parts (members) is borrowed and the other part is of native origin.

In a homogeneous (with respect to its alternations) speech community, all alternations owe their origin to an internal anthropophonic impulse. But when a speech community is mixed (with respect to its alternations), we can distinguish in terms of the original anthropophonic impulse (1) monolingual and (2) bilingual alternations. Monolingual alternations may owe both parts (members) (a) to the native language, or (b) to the foreign language from which they were borrowed. The following possibilities obtain:

1) monolingual native, or internal;

2) monolingual foreign;

3) bilingual, or internal-external.

Only in the case of the first type can it be asked whether the original anthropophonic cause is still active in the given state of the language, or whether it belongs to the past. In the case of the second type, such a question is totally irrelevant.

Paleophonetic alternations, whether internal or external, can either

1) have a purely traditional causality, or

2) be utilized for a psychological, morphological, or semasiological purpose.

Characteristic Features of Foreign Alternations

1, 2, 3. The first three characteristic features are the same as in traditional alternations and correlations. In addition, they have a fourth and fifth feature:

4. The original anthropophonic causes for the transition of an originally homogeneous phoneme into an incipient (embryonic) alternation and then into a neophonetic alternation (divergence) as well as the anthropophonic causes for the further transformation of a divergence into a traditional alternation do not belong to the given speech community, but belong either to a related speech community (more precisely, to the speech community from which the particular alternation was borrowed), or in part to the borrowing speech community, and in part to the speech community from which one of the members of the alternation was borrowed. Thus the Russian alternations di \\ žd, ti \\ šč arose not within the confines of Russian, but among the Slavs who settled the Balkan peninsula; one of the members of the Russian alternations ž \\ žd, č \\ šč is rooted, however, in the past of the Russian language itself, while its other member was formed in the past among the Balkan Slavs, since it was they who originally formed the Church Slavonic language.

5. Fully or partially borrowed alternations cannot be explained by anthropophonic causes. But as far as the neophonetic aspect of the borrowed alternations is concerned, it should be noticed that the phonemes making up the alternations must obey the requirements of the native language of a given period. The sound laws governing the language of a given period apply equally to the native phonemes and alternations and to the naturalized or borrowed phonemes and alternations.

The fourth characteristic feature of traditional alternations pertains only to those borrowed alternations which remain within the sphere of purely traditional alternations, without being utilized for psychological purposes. When such a utilization takes place features 4, 5, 6, 7, 8, 9, and 10, which characterize the correlations or psychophonetic alternations, apply also to the borrowed alternations.

CHAPTER VII

_________________________________

INCIPIENT (EMBRYONIC) ALTERNATIONS.

As in nature in general, we must also distinguish in language two kinds of phenomena: macroscopic phonomena which are immediately apparent and accessible to observation without any special effort, and microscopic phenomena which can be detected only as a result of a concentrated effort.

This distinction also holds with respect to observable entities and their differences. In the first case the entities, whether they are bodies, impressions, or representations, are macroscopic or become accessible to our senses by means of magnifying devices. In the second case however the entities in question and their differences may at first glance appear to be clear and transparent, or they may be infinitely small and, as it were, unreachable, unless we apply magnifying instruments and exercise an effort of maximal attention and concentration.

For this reason we should consider, in addition to the clear and easily definable alternations, alternations involving minimal differences, embryonic alternations.

What is in question here is not merely the desire to satisfy an idle curiosity or the fruitless exercise of our powers of discrimination, but the discovery of the germinal beginnings of anthropophonic causality. For it is precisely at this stage that we find the original influence of different phonetic conditions which produce those incipient differences that, as they increase with time, lead to a split of an originally homogeneous unit into two or more clearly distinct units.

On the basis of the positive results obtained in the study of phenomena due to diverse anthropophonic conditions, we should examine such conditions even though at first sight they do not seem to account for any palpable effects.

And, if for no other purpose, the embryonic alternations should be studied for the simple reason that they may point up the possibilities of linguistic change and provide a stimulus for further objective microscopic linguistic investigations, where we need no longer depend on our subjective impressions but on physical, acoustically and optically reliable equipment.

Thus we may investigate the factors of anthropophonic change and the emergence of phonemic divergence in morphemes containing the respective phonemes independently of their etymological relationship. What we are concerned with is the pronunciation of divergent phonemes in phonetic words or in any phonetic combinations of the given language, regardless of their meaning. Thus the embryonic divergences of the phoneme k in the combinations ka | ke \ ki \ ko \ ku \ k ą\ k ę| kr (kraj) | kł (kłaść) | kl (kl ąć) | kš (krzywy) | ks (k sobie) \ km (kmin) | kń (kniaź) | kt (kto) | kp (kpić) ... do not depend on the meaning of the morphemes and words containing these combinations. Compare the similar independence of meaning of the phoneme a in the combinations ka | ta | pa , where the phonemic divergence of the vowel is determined only by the preceding consonantal phonemes. These embryonic differences in seemingly identical phonemes may further depend on their occurrence in word-initial, word-final, or word-medial position [x|| -x ||-x-], on the degree of intensity (accent), on their occurrence in autophthongal or symphthongal position {u | [a] u, i | [a] i . . ., i.e., u \\ ṷ, i \\ ḭ. . .}, etc.

But if we are to consider this type of divergence within the framework of alternations, we should concentrate only on related morphemes. In the pair dn-0 | den-ko, the phoneme d is only seemingly identical. While the main point of articulation, the moments of pause and closure produced by the respective speech organs are indentical in both ds, they differ in the transition to the following phonemes so that their pronunciation is, in effect, different both acoustically and physiologically.

In the Polish words ród | rod֊u, mróz | mroz-u, mąż | męż -a we have, in addition to the transparent traditional, paleophonetic alternations u(ó) || o, ą || ę ,and the equally clear neophonetic alternations or divergences t (-d) || -d-, -s (֊z) || -z -š(- ż) || -ž, the following embryonic neophonetic alternations:

a) r [ód] || r[odu], [m] r [óz] || [m] r [ozu],

b) m [ąż] || m [ęża],

c) [ró] t \\ [ro] d [u], [mró] s || [mró] z [u], [mą] š || [mę] ž [а]. ...

These alternations, t || -d  s || -z-, š || ž, require further comment. We are not here concerned with such transparent divergenees which stem from the different activities of vocal chords in the larynx, or the weaker quality of t,—s, š with respect to -d,-z-,-ž-, but rather with those that are due to the influence of the preceding vowels u(ó), ą and о, ę upon the consonants. uuuuu

In Russian góda | gădá there are two embryonic alternations, g || g, and d || d, since the vowels following these consonants carry in each case a different stress.

In short, an enormous number of phonetic facts are caused by such phenomena. One could affirmatively claim that there is not a single group of etymologically related words in any language that does not display a whole series of such embryonic alternations or, to put it differently, that there is not a single phoneme in any language that always occurs in the same anthropophonic environment.

Embryonic alternations are to be interpreted as alternations of imperceptible differences in the respective phonemes. As these differences are infinitely small, we may designate them by Ø:

d(x׳ ֊ x) = Ø

where d indicates the difference, and x, x—any phoneme in different anthropophonic environments, which may in time produce a split of the phoneme into two or more phonemes.

The very fact that the words containing the respective phonemes differ on the one hand anthropophonically, i.e., in their phonetic combinations and structure (e.g., the different place of stress), and on the other hand psychologically, i.e., semasiologically or morphologically, introduces a difference between the seemingly identical phonemes that may eventually become perceptible.

Thus the anthropophonic difference between a || a in the Polish words matka | macierz (where the first a occurs in a closed syllable before a nonpalatal consonant, and the second a in an open syllable before a palatal consonant) may in time lead to a transformation (degeneration) of the a in two different directions; for example, to the change of a in macierz into an *-type vowel.

Such a change, we may hypothesize, would in turn entail an alternation m || m, since the m before a in matka is in a different anthropophonic position than the m before e in our hypothetical *mecierz.

A similar anthropophonic difference may affect the vowel a in matka | mateczka (the first in a closed, the second in an open syllable) which may eventually split into two distinct phonemes.

On the other hand, we must keep in mind that because of psychological reasons the phonemes belonging to the morpheme mat in matka, may undergo different changes than the same phonemes of that morpheme in mateczka, since the word mateczka is a diminutive, affectionate, hypocoristic term, which the word matka is not. The morpheme mat and its phonemes m ... a ... t are, consequently, subject to quite different conditions in the two words.

But the potential changes of phonemes caused by psychological differences are not to be confused with anthropophonic causality and must not be subsumed under the concepts of either embryonic or other kinds of alternation.

It is, nonetheless, a fact that each phoneme (sound) is subject to different kinds of influence depending on whether it is treated as a simple sound or a phonetic constituent of a morphological unit. In a similar fashion, each human is subject to different influences as a psychological individual, as a member of a family, of a society, of a state, etc., just as each body is exposed, in turn, to physical conditions, chemical conditions, and so forth.

Embryonic alternations involve two degrees of differences depending on the particular combinations of the phoneme:

1) Truly embryonic, potential alternations in which the difference between the phonemes equals zero (ø).

For example, the perceptual difference to the ear between s || s in the Polish kos | kosa is ø, as long as our perception is not sharpened by means of optical and acoustical equipment.

2) Overt and discernable alternations. The difference between the phonemes can here be designated as a—> ø, i.e., as a discernable difference with the limit ø.

The latter alternation is transitional to divergents.

The embryonic alternations include, however, only alternations with minimal phonemic differences which are not immediately perceived but can be discovered through the conscious effort of analysis.

In the chain of historical development these alternations form only an intermediary link between embryonic alternations proper and neophonetic alternations or divergents which can be detected by minimal perception.

That such minimal perception is at work is confirmed by analogical formations, such as Polish z wusa , instead of z wozu down from the wagon, which appears not only in the language of children, but also in the language of adults, due to the influence of the nom. sg wus (wóz).

CHAPTER VIII

_________________________________

THE GENETIC RELATIONSHIP
BETWEEN DIFFERENT TYPES OF ALTERNATIONS.

GRADUAL TRANSITION FROM ONE TYPE TO ANOTHER.

Let us take as an example a pair of etymologically related Polish words: plotę || plecie (now pronounced piote | pieće). The constituent phonemes of these words form the following alternations:

The present state of alternations in these words has developed from earlier stages in which the constituent phonemes belonged to different classes of alternations.

Historical and comparative studies enable us to posit the following succession of alternations dating back to the Indo-European period:

1)*plet-o\*plet-e-.

This period was characterized only by the embryonic alternations t [о] t [e], e [to] || e [te], in which t and e were differentiated in the same way in all anthropophonic sequences of eto | ete regardless of etymology.

2)* piet _o –o-|*piet _i –e-

with the neophonetic alternation or divergence t ₀ || t _i and the embryonic alternation e[t ₀]] || e[tj.

3)*ple ₀tt₀--o\ ple_itt_i—ee-

with two divergences, t ₀ || t _i and e ₀ || e_i

or *ple ₀tt|*ple_it t-

with the traditional alternation t || t

and the divergence e ₀ || e_i.

This period was marked by a gradual accumulation of anthropophonic tendencies. It was also a period of individual oscillations, when an individual speaker, following earlier usage, might retain the alternation plet֊ || plet-, while another speaker would yield to the strong anthropophonic tendencies and introduce a new alternation plot11 plet into his speech. Moreover, the same individual might use plet and plot interchangeably, that is, plet|| plet or plot|| plet. The language of the children of the speech community had an important role in this process.

4) The established norm

with the paleophonetic or traditional alternations t || t (  t \\ ć ), о || e, and the embryonic alternation l[o] || l[e].

5) The state of conflict between a feeling for the unity of the morpheme and the impression made by its outer form. One consequence of this is the tendency to eliminate external differences and unify the phonetic form of a morpheme which is felt to be a psychological unit, or else to utilize the phonetic differences for psychological purposes, so that phonetic divergence comes to be associated with psychological divergence.

In the case we are discussing, the alternation t || ć is associated with certain verb forms, plot-ę, plot-ą | pleci-e | pleci-esz, pleciemy, pleci-ecie; the other alternation o || e is also to some extent associated with the psychological alternation of verb forms, but it is psychologically much weaker, leading to analogical formations such as plete, pleto instead of plote, ploto (plotę, plotą).

Let us consider more closely the history of the different alternations in the morphemes plot| pleć -.

Let us take first the alternation

In the first stage this was an embryonic alternation t[o] || t [e]; then it became a divergence t[o] || t_i [e); in the third stage it became a paleophonetic alternation t || t_i  t || t (t_i denotes a t with palatalization caused by a following palatal phoneme, while t is the same phoneme t with concomitant independent palatality). In the fourth stage palatalization increases, so that finally the original phoneme splits in two, t || t, and eventually t becomes ć In the fifth stage the traditional alternation t || t t || ć) becomes a correlation, and has remained so until the present.

Now let us take the alternation

In the first stage it was merely embryonic and remained so in the second stage. In the third stage it became a divergence, i.e., e [t] || e [t_i] 0[t] || e [t’] In the fourth stage this divergence became a paleophonetic or traditional alternation. Finally, in either the fourth or the fifth stage, the alternation о || e became psychophonetic, i.e., a correlation; today, however, it seems to have retreated, so to speak, to the status of a traditional alternation which, as a resuit of the tendency to level the phonetic form of identical or etymologically closely related morphemes, is gradually being eliminated.

As for the embryonic alternation

this appeared in the third or fourth stage, but has remained embryonic.

Many other examples could be cited to illustrate this gradual development.

We can make two general conclusions from our study of these alternations:

a) first we can give a survey of the genetic development of alternations in the history of a particular language;

b) then we can examine how this process is implemented in the language of individual speakers of a given speech community, especially in the language of children.

I. Historical sequence of Various Alternations in the National Language (the Language of a Speech Community).

All alternations owe their origin to a splitting of a unity into variety as a result of special circumstances and general causes.

The original cause or impulse of any alternation will always be found in the various anthropophonic conditions in which an originally unified phoneme is bound to appear.

As a consequence of anthropophonic differences , the emerging embryonic distinctions of a phoneme give rise to distinctions which are minimal both in terms of their own objectively observed magnitude and in terms of the impression of them perceived by members of the speech community; and as is well known, the greater or lesser force of the observable image depends on the force of the perceived impression.

Later in the historical succession, the minimal distinctions cease to be minimal; they increase, become intensified and apparent to the naked eye.

This increase of distinctions is first brought about by the same conditions which provided the first impulse for the development of minimal distinctions; but it lasts only a limited period of time, for the phonemes which have now been endowed with new properties begin to change spontaneously in a certain direction. Thus, originally combinatorial changes are replaced by later spontaneous changes, and a conditioned neophonetic alternation or divergence becomes an anthropophonically independent, paleophonetic, traditional alternation.

However, in addition to this possibility of development, i.e., the possibility of increase and ultimate independence of the embryonic minimal distinctions which are at first due to the role of the environment, there also exists another possibility, the possibility of the disappearance of minimal distinctions, namely at a point when the conditions responsible for these distinctions cease to operate.

Thus, for the elimination of neophonetic alternations, irrespective of whether they are embryonic or already plainly divergent, it is sufficient that there no longer be a causal relationship between the anthropophonic modifications of the given phoneme and the conditions responsible for them. Subsequently, however, a neophonetic alternation can be eliminated in either of two ways:

1) the anthropophonic distinctions between the alternating variants of a given proto-phoneme (Urphonem) are totally dependent on the influence of the environment, owing to anthropophonic factors, but by themselves they lack individual traits and are committed to memory neither by way of tradition nor through linguistic intercourse in general;

2) or else, in addition to dependence on the anthropophonic environment, the new features developing in this manner acquire a sort of individual independence and eo ipso are transmitted by tradition and linguistic intercourse in general. The phoneme in question is subject to such a strong anthropophonetic influence that the new features which are thus being formed force themselves upon our perception as independent and autonomous.

1) In the first case anthropophonic variants (variations, modifications) of the phonemes disappear along with the loss of the functional dependence, and the embryonic variants return to a state of absolute identity. When the causes are eliminated their results, too, disappear.

2) In the second case the loss of dependence on anthropophonic conditions does not entail the loss of the features which have been acquired by way of anthropophonic adaptation; on the contrary, having ceased to be functionally dependent, the new features become individualized and autonomous; as such they are capable of further strengthening and growth and are able to impress themselves ever more strongly on the memory of the speakers. In this way the properties of the phonemes which are acquired by way of anthropophonic adaptation become individual, independent properties of a given phonetic segment of a given morpeme in a given word.

This unconscious remembering of individual features, insofar as it affects etymologically related morphemes, provides the point of transition between the category of divergents and that of traditional alternations.

Thus we enter the field of traditional alternations, which are maintained by force of linguistic intercourse between the members of a given speech community. However, although tradition is a force capable of preserving alternational distinctions inherited from earlier generations, it is not sufficient to endow these distinctions with a causal relationship to serve for the expression of specific needs. But it is precisely in terms of these needs that the individual speaker, and eventually the entire speech community becomes aware of the lack (absence) of a raison detre for the distinctionst and this awareness cannot but lead to a conflict with tradition.

Nevertheless, the distinctions between phonemic alternants which have already lost their raison detre can be retained for two reasons:

1) one, psychological—when individually unmotivated distinctions are committed to memory not as types, but precisely because of their specific, individual character. This is especially true of frequently used, everyday verbs such as be, eat, know, give . . ., nouns like father, mother . . . , e eyes, e ears, h hands, feet . . . , pronouns like I,you . . . ;

2) another, sociological—when the preservation of individually unmotivated distinctions facilitates communication between members of the same and different generations. Here tradition is an important factor of conservatism in maintaining the cohesion of a given speech community.

However, there may arise an individual need to liquidate distinctions that are motivated neither anthropophonically nor psychologically, i.e., alternations that are neither divergences nor correiations; superfluous traditional distinctions may be eliminated in one of three ways:

1) The feeling for an etymological connection between the morphemes containing the alternating phonemes may bring about a phonetic assimilation (leveling) of their form by generalizing one of their variants. For example, in Polish

and in Russian, on the contrary,

Another example of partial leveling in Polish is:

Compare further z-bór, po-bór, na֊bór, wy֊bór . . . , -bor-u. . . .

2) The feeling for the etymological connection of the words in which the morphemes containing the alternating phonemes occur is lost, causing lexical differentiation between words that were previously felt to be etymologically related; this differentiation, however, enriches the vocabulary of a given language; e.g.:

Even jes—(jes-t. . .) >< s-(s- ą) or chodz-i ć szed- ł, sz-ła are now phonetically speaking two separate roots which maintain their connection only through their semasiological association.

The complexity of alternating relations of a given morpheme facilitates its mixing with morphemes, its attraction to a different class of morphemes. Pertinent Polish examples are:

od-po֊czn-ę j od-po-czą-ć | od-po-czyn-ek . . . from od-poczy-nę \ od-po-czy-nąć | od-po-czyn-ek, and attracted, at least morphophonemically, to the class of za-cznę, po-cznę | -czać | -czyn-ać \za-czyn-, roz-czyn. . .

3) The third way of eliminating unmotivated alternations is the utilization of these alternations for psychological purposes, that is, for the association of phonetic differences (nuances) with morphological or semasiological ones. This is the source of psychophonetic alternations , or correlations, which every member of a given speech community is bound to acquire by his own mental effort through accumulating and generalizing individual associations.

Regardless of the subjective feeling of the speaker, traditional alternations may regularly occur in certain types of forms which are semasiologically distinct; e.g., Polish:

nog-a, ręk-a | nóz-ka, rącz֊ka . . . ;

pi-ć, gni- ć | poj-ić | gnoj-ić . . . ;

tok-, bok | tocz-yć, bocz-yć sie . . . ;

po-mog-ę, chodz-ę, mocz-ę, robi-ę | po –mag-am, chadz֊am,

macz-am, rabi-am . . . ;

gniot-ę, miotę [ gniat-am, miat-am . . . ;

sł-аć, tk֊ać | sył-аć, tyk-ać. . . ;

u-br-ać, wy-pr-ać | u-bier-ać, wy-pier-ać . . . ;

u-mrz-eć, wy-prz-eć, wy-trz-eć | u-mier-ać, wy-pier-ać, wy-cier֊ać. . . .

However, if the speaker becomes aware of a constant relationship, a fixed association between the phonetic form of such words and their meaning, then the traditional alternation eo ipso becomes a correlation, and the native speaker acquires a richer repertory of psychophonetic devices for the expression of morphological and semasiological distinctions.

But despite repeated efforts to eliminate irrational, i.e., neither anthropophonically nor psychologically motivated phonetic distinctions, there can always remain a considerable substratum of distinctions which do not affect the semantic relationship between the individual variants, i.e., distinctions constituting the paleophonetic or traditional alternations, such as are found in (Polish):

ś-mier-ć |mar-twy | -mor-u | -mór | u-mier-a . . . ; br-ać

wierc-i | wart֊ki.. . .

This substratum constantly decreases as a result of the abovementioned changes in three directions; on the other hand, it constantly increases at the expense of former divergences, as well as of former correlations.

The tendency to eliminate distinctions lacking individual motivation (causality) affects alternations of both divergent and correlative origin. In the latter case, i.e., when the previously live psychophonetic connection disappears, only the first two ways of eliminating the alternation manifest themselves: either the morphemes assimiliate (level, make uniform) their phonetic form or the feeling for the etymological relationship of the words and morphemes containing the alternating phonemes is lost. The third way, i.e., the restoration of the psychophonetic character of traditional alternations of correlative origin, is a very rare phenomenon altogether.

Thus, traditional alternations can be classified according to their origin as:

1) paleophonetic alternations_ג which have lost their anthropophonetic causality hmotivationm;

2) paleopsychological alternations which have lost their psychophonetic causality motivationm.

In time both of these alternations either can be eliminated by analogical leveling of the constituent phonemes (either x׳ || x  x || x׳ or x ||׳ x״  x״) or can lose their alternational character altogether when what were at first phonetic variants of one basic morpheme split into two morphemes which are no longer felt to be related, so that the alternations become merely residual (rudimentary):

Since we are dealing with the psychological value of phonetic phenomena, and the feeling for this value ebbs and wanes with the individual characteristics of all the members of a given speech community, we must assume a whole scale of indeterminate transitional states which are reminiscent of the flickering of a flame about to die out.

In this gamut of countless indeterminate transitional states three classes of alternations stand out conspicuously, and have been discussed in Chapters III, IV, and V:

traditional alternations which are motivated purely socially (being due only to tradition and communication); these alternations are either paleophonetic or paleopsychological (but the latter are likewise ultimately of a paleophonetic origin; Cf. Chapter V);

neophonetic alternations, or divergences _ג which have an individual anthropophonic motivation absent in the other two classes (Cf. Chapter III);

psychophonetic alternations, or correlations _ג having an (individual) psychological motivation absent in the other two classes (Cf. Chapter IV).

From the viewpoint of their historical development these classes can be presented in the following sequence:

1) divergences, characterized by a live physiological process and individual anthropophonic relationship, preserving the original motivation of all alternations . . .. (examples: d || t in wod-a | wód-ka , ś || ź in pro ź-i ć| pro ź֊ba, i₂ ((y) || i₁ ((i) in s łom-y | ziem-i);

2) paleophonetic, or traditional alternations, characterized by active social factors, with loss of the originally anthropophonic motivation of the alternations . . .. (Examples: p || p’, t || ć, ar || er in na-parst-ek | pierści֊eń ,ł || l, o || e in czo ł-o | czel-e);

3) correlations characterized by vitality of the psychological factor, individual psychological correlation, secondarily developed psychological motivation . . .. (Examples: o || a in chodz-i ć| chad z -ać, trwoż-yć| trwa ż-ać, s || ch in w łos | w łoch, klus-ki | kluch-y).

From the standpoint of space, i.e., of place of origin, the traditional alternations fall into two classes:

a) native formations developed within a given speech community, and

b) foreign formations adopted from another speech community by interlingual oral communication or through writing.

From the viewpoint of transition from an earlier to a later state, the traditional alternations in turn form two classes:

a) those that developed directly from divergences , and

b) those which are a continuation of correlations _ג and have switched to the category of traditional alternations.

In the former alternations we can, in turn, distinguish three classes:

a) those directly derived from divergences;

b) those which have passed through only two stages of alternational evolution—that of divergences and that of traditional alternations; and

c) those which have passed through four stages of evolution before their disappearance—divergences, traditional alternations, correlations related to tradition, and purely traditional alternations.

In the correlations we can distinguish two principal levels of psychological intensity.

a higher level which involves an active and creative association of the phonetic variants with psychological differences (nuances), and which makes possible the formation of new alternating pairs on the model of an existing productive type (for example, the alternation о || a in Polish verb stems: nos-i- ć| nasz- ć); and

a lower level on which the association between the phonetic variants and psychological differences, though present, is not strong enough to create new analogical formations; this is then the level of psychophonetic weakness and passivity (for example, i ||oj in pi- ć, gni-ć| poj-i ć , gnoj-i ć. . .).

Between these two levels there is a whole range of intermediate stages.

We have shown above that the traditional character of simple traditional alternations always finds itself in conflict with the tendency to eliminate phonetic distinctions which are not motivated physiologically or psychologically, and that when these individual tendencies prevail, the traditional alternations assume a psychophonetic character and become correlations, or else there is phonetic leveling of the alternating pair.

Similar conflicts can also develop in the case of correlations. For example, in the Russian correlations k || č in verbs such as pekú | pe čóš, tolkú | tolčóš . . . we can detect a conflict which leads ultimately to the substitution of the k || č correlation by a new k || k correlation (pekú | pekó š, tolkú | tolkóš. . .).

If we were to regard the alternating phonemes k || č as indivisible entities, we would have to deny the presence of any conflict and expect the correlation k || č which forms a closed unity to remain without change until its psychophonetic character is finally replaced by a purely traditional one. However, if we consider the articulatory components of the phonemes and the verbal type to which they belong, the conclusion imposes itself that the correiation k || č disrupts the harmony of the dverbalv type and that only the alternation k || k does it justice, since it is in full agreement with such other partial alternations as b || b (grebú | grebot), d || d (vedú | vedót), s || ś(nesú | nesót), etc., which can be subsumed under the general formula

PØ || PY

(where PØ designates the absence of palatalization, and PY the presence of palatalization in the consonant).

This generalization of Russian correlations in the above-mentioned verb type is taking place at present in various Russian dialects. The same phenomenon occurred at a much earlier time, for example, in possessive adjectives which have come to serve primarily as patronymics and family names, such as súkin syn, Súk-in, Sobák-in, Kó šk-in, Sipjág-in, Múx-in . . . ; the old correlation k ||č , g ||ž, x || š (súk-a | suč—in, sipjdg-a | sipja ž-in, múx-a | mú š-in . . .) has been replaced by the new correlation k | k, g | g, x | x on the model of a large number of partial alternations of the same general type, b || b, d || d, s || s (rýb-a | Rýb-in, Marúd-a | Marúd-in, pl áks֊a | Pláks-in . . . )

Thus we see how correlations can grow, how partial correlations can be absorbed by more general ones, and how the generalization of correlations may make them more efficacious and supple.

However, as is the case with all linguistic forms, the alternations too must ultimately be weakened and disappear, no matter how transparent and productive [mächtig] they may be at first. Having absorbed a series of partial correlations and having reached the highest degree of strength and clarity, the general alternations in time lose their force and psychophonetic character, become plain traditional alternations, and are finally lost in the mass of unproductive extinguished alternations.

II. The Development of Alternations in Individual Language, especially in that of Children

Language can not be inherited; only the faculty of speech and the tendencies toward a certain direction of changes incipient in the structure of the language are inherited. Heredity is a biological factor, while the language of any individual develops through social intercourse. Nevertheless, we must fall back on heredity in order to explain the persistence of repeated historical changes.

The most radical changes take place in the language of children. The most far-reaching are phonetic changes and morphological leveling. Thereafter the children gradually acquire the speech habits of the adult members of the community, but a certain number of changes inherited from childrens language may remain in their individual speech, and, more importantly, the tendencies toward such changes, even if they arise again spontaneously in later generations, become a part of their inheritance. Insofar as they accrue in successive generations, these changes eventually become ingrained in the language.

In alternations we are dealing first of all with the inherited accumulation of phonetic tendencies and with their gradual intensification.

Since every individual acquires on his own the language of his speech community, he necessarily also acquires the alternations by himself.

During the gradual acquisition of his native language, every child proceeds through several stages: at first he does not understand anything; then he begins to understand the language of his environment, although he is still unable to speak, i.e., he is still in a state of audition and passive linguistic perception; finally, he begins to speak, adding to audition and perception the factor of phonation. Linguistic cerebration or thinking, once set in motion, now finds a solid foundation in the acquisition of individual language.

To be sure, in the earliest stages of this process, when the child is just beginning to understand the language of his environment, there can be no question of alternations. Alternations develop only later. But in any case we can regard pairs of phonemes which enter into words and morphemes understood by the child, and which form alternational pairs in the language of his environment, as embryonic, or germinal (incipient) in the speech of the child.

At this stage, when the child has not yet begun to talk but is already aware of the properties of the native language and can understand it within certain limits, that is, when the child has reached the state of advanced audition and perception, but without phonation, there naturally cannot be any question of neophonetic alternations or divergences, since these depend on individual pronunciation. Whether there also develop at this timea correlations or psychophonetic alternations depends on the personality of the given child, or his psychological capacity and agility. In any case, there are no correlations in the very first stages of this development; they emerge only by way of associations from traditional alternations. The simple traditional alternations apparently exist in childrens speech in the stage of audition and perception; the ability to distinguish phonetic modifications (variants) of morphemes which are felt to be etymologically related then develops spontaneously. That is why the clear distinctions between divergents or neophonetic alternants also leave a definite imprint on the memory by dint of their own traditional quality. Thus, for example, before the child begins to distinguish in his own pronunciation such a divergence as d || ţ in Polish broda | bródka, he has already observed the anthropophonic difference in the pronunciation of the two phonemes. And this is how the element of tradition becomes added to the neophonetic alternations or divergences . . .

When the child begins to talk, imitating the speech of his environment, he also imitates the alternations, although the neophonetic alternations or divergences may develop independently, inasmuch as they are determined on the one hand by the shape of the speech organs and on the other hand by inherited tendencies. This allows for wide individual variations: some children develop divergences to a much greater degree than others. However, the tendency to develop new divergences is, on the whole, much more typical of the speech of children than that of adults, whose Ianguage represents the linguistic norm. What is only embryonic in the latter can become apparent and palpable in the speech of children. The various conditions implicit in the combinations of a single phoneme quickly yield definite results here when the phoneme splits into two or more phonemes; the anthropophonic causes in the language of children are far more prominent than in that of adults, and their effects are, consequently, more immediate.

In short, as in the realm of phonology and morphology, children anticipate the normal development of language in the area of alternations.

At any rate, childrens speech has far more neophonetic alternations or divergences and fewer correlations or psychophonetic alternations than the normal language of a given speech community. However, we cannot ignore one correlation which is highly developed in childrens speech: the alternation of palatal with nonpalatal consonants for the expression of endearment (d || d, ś\\s,l\\ ł...).

As childrens language comes to resemble that of adults, the child regresses, so to speak, in the field of alternations; he loses the most innovative variants (modifications) and replaces the stage of transparent divergence with that of an embryonic alternation.

Now that we have examined, although only in rough outline, the development and transformation of various types of alternation in the language of children, we shall briefly and schematically present in tables and formulas the historical sequence of the various stages of alternation, and the transition from one stage to another in the language of an entire speech community.

I. Divergences encompass the whole language, but arise spontaneously, through anthropophonic adaptation. Every period in the history of a language is characterized by its own divergences, although there are some which are external and universal.

II. Correlations develop from traditional alternations.

III. Traditional alternations develop either from divergences in the native language or from traditional alternations of foreign origin.

II-III. Phonemes, which express traditional alternations, whether merely traditional or traditional-psychophonetic (correiations), must at the same time fulfill the requirements of divergences or anthropophonic alternations of the particular period. (. . .)

While the alternations, after completing a certain evolutionary cycle finally disappear, the sources for new alternations never do. Consequently, there is an unending process of reconstruction of alternational relationships resulting in the accumulation of new layers of alternations.

Some kind of anthropophonic change is always taking place at every stage of the language, some kind of adaptation of the phonemes to anthropophonic conditions. The effects of these adaptations are transmitted by tradition from one generation to another, until they are, in turn, replaced by new changes.

NOTES

1. Cf. Brugmanns review (Literar. Centralblatt, 1882, No. 12, p. 401), where in any case Brugmann erroneously states that the author defines as phonetic interchange what is usually called phonetic transformation or phonetic change.

2. However, at that time I understood phoneme to mean something different than now; i.e., I interpreted it as the sum of the phonetic properties representing an indivisible unit within either a single language or a group of languages. (The proposal to use the term ״phoneme instead of sound came from Kruszewski.)

3. In the case of this alternation, there is a striking parallelism between šč and žd. Why dont we have št like žd, or ždž like šč? It is not too difficult to answer this. In the Old Church Slavonic documents there was an abbreviated symbol instead of . This sign, which had the form of a single letter, was pronounced like šč in words of Russian origin that were not influenced by Church Slavonic such as pušč—ú, í-ščet, pišč- ít.... It was then quite natural for this pronunciation to spread as well to words with the letter , borrowed directly from Church Slavonic texts such as osveščát, prosveščénie, sovraščát, prekraščát, pišča, umerščuljat . . . , and not osveštát, prosvešténie although this second pronunciation, with št, was characteristic of Old Church Slavonic itself. The matter stands somewhat differently with žd. The Church Slavonic texts use no abbreviated sign for žd of the type in place of , and therefore it did not occur to any Russian to pronounce the words roždát, roždestvó, ograždát, ubeždát’, iždivénie, voždelénnyj in any way other than with žd.

A Baudouin de Courtenay Anthology