1. The rise of structural linguistics in the twentieth century has been associated in the West with the seminal book of Ferdiliand de Saussure, Cours de linguistique générale, which was published posthumously (in 1916) by his students A. Sechehaye and Ch. Bally on the basis of his lectures at the University of Geneva. During llis lifetime de Saussure had published (at the age of 19) only one book, the Mémoire sur le système primitif des voyelles dans les langues indo-européennes (1878), which was perhaps as important for the further development of linguistics as the Cours, although it was primarily a diachronic study, while the Cours tried to formulate the general, synchronic principles of language. The implications of the Mémoire were stated succinctly by A. Meillet: le Mémoire, he writes, apportait, par une innovation, un système cohérent qui embrassait tout les faits, mettait à leur place les faits connus. . . . Des hors il nétait plus permis dignorer à propos daucune question, que chaque langue forme un système où tout se tient.1
De Saussures monograph was highly evaluated also by Baudouin de Courtenay, who was at the time of its publication professor at the University of Kazan: De Saussures discovery of the vowel alternations is crucial for the understanding of Indo-European morphology. . . . It is the great merit of de Saussure to have emphasized the role of phonetic relations within the morphological structure of words, a phenomenon which Baudouin preferred to call, in the terminology of the Kazan school, morphologization of phonetic differences (1909 (2), 219).*
While Baudouin de Courtenay was in Kazan working out his theory of semasiologization and morphologization of phonetic differences, i.e., of the function of sounds in the lexical and grammatical system of a language, de Saussure was occupied with the formulation of his phonétique sémiologique, a discipline he defined as dealing with des sons dans chaque idiome en tant quayant une valeur pour lidée.2 De Saussure did not, however, publish his findings, for he was ridden with doubts about the validity of his new method and was deeply troubled about the consistency and cohesion of his new, synchronic approach to language.3
Unlike de Saussure, who lived in the hub of linguistic activity, which was at that time dominated by a genetic approach to Ianguage, Baudouin de Courtenay moved boldly ahead with a small following of students to elaborate and to publish his findings, for in the remote Russian city of Kazan he was more or less immune to the criticisms and attacks of his Western and Russian colleagues.
Both men were thus essentially set on the same course of inquiry, which amounted to no more nor less than a complete overhaul of contemporary linguistics. Their principal goal was a description of language as it is, and not as it was, or how it came about—a goal that, in the words of Baudouin de Courtenay, could only be accomplished through a radical revision of [linguistic] principles and methods that would be predicated in the first place on the coining of a completely new terminology and, in the second place, on a reformulation of the relations of linguistic elements (1909, 261). De Saussure echoes these statements: En matière de la langue on sest toujours contenté dopérer sur des unités mal définies, and la linguistique travaille sans cesse sur de concepts forgés par les grammairiens, et dont on ne sait sils correspondent réellement a des facteurs constitutifs du système de la langue.4
The striking similarities of formulation and outlook between the Polish and French scholars were not the result of mere convergence of ideas. Baudouin de Courtenay followed the linguistic literature of his time closely (as attested by the long biographical lists attached to his programs of lectures and books) and, as we have seen, he regarded de Saussure as a true innovator in linguistic theory, whereas de Saussures writings reveal an intimate familiarity with Baudouins thought, to the extent that some passages of the Cours echo almost verbatim Baudouins formulations. Baudouin and his pupil, Kruszewski, were also singled out by de Saussure as the only European scholars who truly contributed to linguistic theory: Baudouin de Courtenay et Kruszewski ont été plus prés que personne dune vue théorique de langue; cela sans sortir de considérations linguistiques pures; ils sont dailleurs ignorés de la généralité de savants occidentaux.5 The intellectual affinity between Baudouin de Courtenay and de Saussure was recognized by other linguists, for example by de Saussures disciple, Meillet, who wrote that between the Russian and French schools of linguistics on ne constate du reste . . . aucune différence essentielle de principes et de méthodes6 whereas some of Baudouins disciples are known to have claimed that de Saussures Cours contained nothing that they had not already learned from Baudouin.7 Neither of these claims were quite accurate, for although Baudouin and de Saussure were guided by a common objective and principles, they differed in their emphasis on and the solutions given to various linguistic problems.
It would, however, be an oversimplification to credit only Baudouin de Courtenay and de Saussure with the revival of general linguistics and the broadening of its scientific horizon. As early as in the seventies and eighties of the last century, i.e., at the very height of the Neogrammarian movement, a reaction to the onesided historicism and blunt empiricism of the comparativists had already set in. This reaction started mostly outside of Germany, since the Leipzig school of linguists had, according to H. Schuchardt, succeeded in silencing all other linguists in Germany.8 Some of these linguistic reformers propounded ideas which were clearly at variance with Neogrammarian dogma; thus the American W. D. Whitney emphasized, like de Saussure, the social character of language and its importance as a system of signs; in France, M. Bréal and A. Meillet were raising the questions of meaning and of semantic change; G. I. Ascoli and H. Schuchardt (of Graz), like Baudouin de Courtenay, undermined the notion of linear development of languages through the study of linguistic geography and language mixture. Schuchardt stressed, like Baudouin de Courtenay, the creative role of the speaker in linguistic change; G. von der Gabelentz, like de Saussure and Baudouin de Courtenay, focused attention on the general and synchronic aspects of Ianguage; the Swiss J. Winteler, the Englishman H. Sweet and the Russian Caucasianist P. Uslar were about to arrive, independently of Baudouin de Courtenay, at the concept of the phoneme. This ferment in European and American linguistics did not, however, procede in a straight line, but, like all novel intellectual currents, followed a somewhat meandering course. The boldest and most systematic among these innovators, de Saussure, Baudouin de Courtenay, and Kruszewski, were not themselves consistent in their views and shared, in fact, many tenets of the Neogrammarians, together with the scientific preconceptions of their time. Thus de Saussure believed for some time that a diachronic approach must precede the synchronic study of language, and that historical change is not subject to systematization, whereas Kruszewski and Baudouin were convinced that one of their main achievements was the discovery of the laws of analogy, in which they merely preceded the Neogrammarians. Baudouin shared with the latter, furthermore, a dualistic conception of language as a complex of physiological and psychological phenomena, and especially in his early writings he showed faith in the strict laws of causality and in the march of progress, leading to the greater humanization of language and to its greater simplicity determined by the principle of least effort. These positivistic and mechanistic elements that are ingrained in the thought of the Swiss and Polish linguists do not, however, diminish the novelty of their ideas, which formed the foundation of modern linguistics.
It may be pointless and unrewarding to compare the respective merits of Baudouin and de Saussure for the subsequent direction and development of twentieth-century linguistics. Neither of them produced a unified system or a mature work which would weave the various strands of their thought into one fabric. The impact of the Cours (whose elliptical and inconsistent statements can only in part be blamed on the notes of de Saussures students) on twentieth-century linguistics is well known, and it has remained until now the most programmatic and most cited work of that linguistic trend which has become known in Europe and America as structural linguistics״ (a term that was actually coined by R. Jakobson). The work of Baudouin, on the other hand, has remained virtually unknown in the West, first because most of it was written in Ianguages that are inaccessible to Western scholars (Polish and Russian), and second because it is scattered in journals and publications that are not readily available even in Eastern Europe (the 1963 Soviet edition of Baudouins writings has now made most of them accessible in Russian). And yet it now seems possible to assert that the ideas of Baudouin have been more seminal and have withstood the test of time better than the structuralist ideas advanced by Ferdinand de Saussure. It will be remembered that the linguistic conception of de Saussure is organized around a set of dichotomies, or rather irreconcilable antinomies, which were apparently designed to overcome the blunt empiricism and mechanism of late nineteenth-century linguistics, and which include such well-known terms as synchrony/diachrony, langue/parole, social/individual, paradigmatic/syntagmatic, external/internal (approaches to Ianguage). Some of these terms and concepts were first formulated by the Kazan linguists, who, as Baudouin tells us, developed at one time a veritable mania for coining new linguistic terminology (1895 ; 1903, 54ff.). To Baudouin and his followers we owe in addition such technical terms as phoneme, morpheme, grapheme, syntagm, distinctive feature, alternation, which have become the stock-in-trade of any practicing linguist. The polarities established by Baudouin were, on the other hand, less extreme than those introduced by de Saussure, and in consequence more realistic and fruitful. In place of an ordered synchrony, which is opposed to a disordered diachrony, Baudouin speaks of the laws of statics and dynamics that apply to history and to synchrony; and in place of a collective langue that is opposed to an individual parole, Baudouin recognizes the role of the speaker and the interaction between the conservative social norm and the creative, innovative power of the speaker in the speech act. For de Saussure language is only form, not substance, whereas for Baudouin it is both form and substance. That is why for de Saussure dans la langue il ny a que des différences sans termes positifs,9 and why de Saussure defines the phonemes as having une valeur purement oppositive, rélative, négative,10 while for Baudouin the phonemes are oppositional and relative, but at the same time consist of positive, phonetically definable properties. For de Saussure language is, like a game of chess, a tightly ordered, algebraic system of relations, whereas Baudouin admits fluctuations and free choice, as well as the constant interference of social, spatial, and temporal factors. De Saussures Cours has, furthermore, remained a theoretical program, a guidepost for modern linguistics, whereas Baudouins work has from the beginning had a more pragmatic orientation and has pointed the way to concrete, empirical investigations. Thus there is hardly a topic in modern linguistic research that has not in one way or another been inspired by Baudouin and for which he has not himself done the spadework. This is of course partly due to Baudouins long and productive life, and partly to the fascination he holds for generations of linguists, beginning with his Kazan disciples and continuing with his students in Europe and in America. The vitality of his thought, however, is mostly due to the activity of those linguists who were directly or indirectly associated with the Linguistic Circle of Prague, who have consciously continued in the course he set and who have more than any single structuralist group or school defined the landscape of modern linguistics. R. Jakobsons work on phonology and distinctive features; N. Trubetzkoys research in Russian morphophonemics; L. ščerbas, R. Jakobsons, and U. Weinreichs studies of mixed Ianguages and areal diffusion; Jakovlevs and J. Vacheks contributions to graphemics; E. Polivanovs and L. ščerbas work on the various levels of language and types of scientific notation; A. Martinets and R. Jakobsöns investigations in diachronie phonology; R. Jakobsons analysis of child-language—all are, in one way or another, the outgrowth and further elaboration of Baudouins ideas and program. And what is even more striking is that Baudouin himself has clearly delineated (and indicated as the impending tasks of twentieth-century linguistics) a number of linguistic areas such as typology, language universals, sociology of Ianguage, and the relation of linguistics to other sciences, which are only now beginning to move into the forefront of linguistic research.
2. Jan Baudouin de Courtenay (whose full Christian name was actually Jan Ignacy Niecislaw) was born on the first of March 1845 in Radzymin near Warsaw.
After the completion of high school (where his chief interest was mathematics), Baudouin entered the historical-philological faculty of the short-lived Polish University (Szkoła Główna) in Warsaw, where he received his masters degree in 1866. By ancestry a Frenchman and nominally a Catholic, he considered himself a Pole and an atheist. But having lived and taught most of his life in Russia, Austria, Latvia, and Poland, he was, in effect, a cosmopolitan who, in the words of Meillet, a eu le maleur de nappartenir tout à fait à aucun pays.11 He could trace his origin to a long line of French aristocrats that included Baldwin, count of Flanders, the founder of the Latin Empire in Constantinople in the thirteenth century. His impoverished great-grandfather migrated to Poland, where he became colonel of artillery and head of the court-guard of the Polish king, August II. His grandfather was Kammerherr (Chamberlain) of the last king of Poland, Stanisław August Poniatowski,. a man of letters, who translated Molière and wrote several books on mesmerism.
Baudouin continued his studies in comparative Indo-European, in Sanskrit, and in Slavic philology in Prague (under Schleicher), in Berlin (under Weber), in Jena, Leipzig, and St. Petersburg. He received a doctorate in Leipzig (in 1870) for previously published linguistic work, and a masters degree in St. Petersburg (where his Polish degree was not recognized) for his study on the Old Polish language before the fourteenth century.
An independent and searching spirit, he had respect neither for his Neogrammarian mentors in Leipzig (Leskien, Brugmann, Delbrück) nor for his Russian supervisor, I. I. Sreznevskij, whom he considered a dry and narrow-minded philologist. All his life he claimed to be self-taught, though he was obviously concerned (certainly in his early work) with similar problems as the former, and owed some of his interests (especially the one in Slovenian dialectology) to the latter. To Sreznevskij he was also indebted for his first position as docent (assistant professor) of comparative grammer at St. Petersburg University. In 1872 he was sent by the Russian Academy on a field trip to Southwestern Austria and Northern Italy to investigate the local Slovenian dialects. The main result of this field trip was his O pyt fonetiki rezjanskix govorov (Phonetic Outline of the Rezija Dialects) for which he was awarded the Russian doctorate in comparative Indo-European grammer. In conjunction with his Slovenian fieldwork, he came into personal contact with Ascoli, whose lectures he attended in Milan and who must have influenced his view on dialectology and on the mixed character of languages.
In the fall of 1875 Baudouin moved to Kazan, first as assistant professor, and, a year later, as full professor of comparative IndoEuropean linguistics and Sanskrit. The nine years spent in Kazan mark the Sturm and Drang period of Baudouins scholarly and pedagogical career. Although he tried his best to get out of this provincial, frontierlike Russian-Tartar city (as evidenced by his letters and pleas to Jagić), he found there a lively and exciting scholarly community, a group of brilliant disciples and followers (including N. Kruszewski, V. Bogorodickij, S. Bulič, and the Turcologist V. Radloff), and immediate publication opportunities (in the Zapiski of Kazan University). Baudouins linguistic program subsequently became known as the teachings of the Kazan school, whose primагу aim was—according to the testimony of one of his students—the application of a strict scientific method to linguistics and the study of the modern [linguistic] ideas of European scholarship.12
In 1883 Baudouin de Courtenay moved to Dorpat (Tartu) to occupy the newly founded Chair in Comparative Slavic Grammar. In addition to Slavic linguistics, he continued to study and teach general linguistics, Baltic dialectology, and Lithuanian texts (Donelaitis). From Dorpat he transferred in 1893 to Cracow as professor of comparative linguistics and Sanskrit. His delight in being finally at a Polish university and in the company of an eminent colleague, J. Rozwadowski (a general linguist and Indo-Europeanist), and within the fold of the Polish Academy of Sciences was, however, short-lived. As a result of his attacks against the bureaucratic and nationalistic Austro-Hungarian administration, his five-year contract was not renewed, upon which Baudouin returned to Russia, to St. Petersburg, where he remained until the end of World War I. He again attracted and raised a new generation of outstanding linguists and Slavists (E. Polivanov, L. Ščerba, I.. Jakubinskij, S. I. Bernštejn, M. Vasmer). But his commitment to social and political causes got him into personal difficulties there too. Because of a brochure in which he attacked the Czarist political suppression of national minorities (entitled The Territorial and National Mark in Autonomy and dedicated to all patriots), he received a twoyear prison sentence (in 1913), which forced him to suspend his teaching and to spend several months in jail.
He was actually freed by the outbreak of World War I and he resumed for a brief period his teaching at St. Petersburg University. With the formation of an independent Polish state, he was invited to the Chair of Indo-European Linguistics at the University of Warsaw. He died November 3, 1929, at the age of 85.
Baudouin de Courtenay combined a profound belief and passion for science with an equally strong faith in the rights of man and social justice. He would make no compromise for reigning fashions, and he chided some of his compatriots for their obsequiousness to German scholarship (a Russian or Pole, he said, may also have original ideas (1901 ). He had little patience with nihilism and apathy in science, with philological plodders and collectors of data. At the same time he believed that ideas are not the exclusive monopoly of specialists, and he viewed his teaching as a venture to be shared with his students. He defended his students from unfair attacks and credited them with their discoveries and terminological inventions. Kruszewski, he wrote on more than one occasion, enlivened the intellectual atmosphere of Kazan, has taught me many things, and has been able to develop 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 (1895 ).
His attitude toward Kruszewski was not, however, free of ambiguity and jealousy, as evidenced by the petty and ungenerous Necrology (1888) which he wrote upon the death, at the age of 37, of this brilliant Polish linguist.
Baudouin dedicated all of his long and active life to the study and understanding of language and languages. He embraced the nineteenth-century belief in generalizations and laws, though he conceived them in a spirit quite different from that of his Western colleagues. He could speak and write a large number of Ianguages (Polish, Russian, Slovenian, Czech, German, French, Italian, Lithuanian, Yiddish) and was a specialist in various IndoEuropean and other languages (Sanskrit, Latin, Slavic, Baltic; Turkic, Finno-Ugric; Ido, Esperanto).
During his life he was honored by many societies and universities, mostly for his work in historical linguistics. His more important work in general linguistics found almost no echo outside the circle of his immediate students and followers. Meillet himself, who paid him otherwise high tribute, considered him le dernier survivant du grand renouvellement de la linguistique, et en particulier de la linguistique historique.13 It is for this reason, and because of his deep modesty, that Baudouin could not divest himself of doubts concerning the value of his lifetime work or of the importance of his school. Any mention of the so-called Kazan school, he wrote, is bound to evoke an ironic smile, a playful operatic mood, if not downright hostility. . . . The Moscow linguistic school [of Fortunatov] is quite a different matter; its founder is a scholar of great knowledge and has the ability to concentrate on a few problems; Moscow is a great city, and the Moscow school can boast of many disciples (1903, 48).
He looked back with bitterness at the fragments and pieces in which he advanced his pioneering ideas, and he blamed his lack of training, his inability to concentrate, and his personal hardships for his failure to leave any major unifying works. Like the other bold innovators in linguistic theory at the turn of the century, he was acutely aware of the incompleteness of his results and of being out of step with his own generation. This feeling must have been particularly poignant toward the end of his life, for in his native Poland he had almost no followers. In a late letter to Ułaszyn, one of his few Polish disciples, he wrote: At every step I have met only blows and disappointments. . . . Laisse nous oublier que nous avons vécu.14 It was the task of the following generations to vindicate his name and outstanding achievements.
3. It is generally agreed that the emergence of a structural approach to language is intimately connected with the formulation of the two interrelated concepts of system (оù tout se tient) and of invariant (with the corrollary notion of variant) which modern linguistics has come to share with other sciences.15 It must not, however, be forgotten that these concepts were first advanced in early nineteenth-century linguistics, though they were then given a somewhat different interpretation.
The idea of an organic whole (which determines the function and properties of the parts) as opposed to a mechanical system (in which the whole is a sum of its parts) was the credo of the Romantic philosophers of language, and it was clearly stated by W. v. Humboldt: Es gibt nichts Einzelnes in der Sprache; jedes ihrer Elemente kündigt sich nur als Teil eines Ganzen an . . . [die Sprache] muss in jedem Augenblicke ihres Daseins dasjenige besitzen, was sie zu einem Ganzen macht.16 But, since the sign cannot, according to Humboldt, be separated from what it designates (for it is neither the product of reason nor of social agreement), it follows that the study of language would remain a chimerical enterprise if it were to ignore the dry mechanical analysis of its physical aspect. The analysis of language must recognize the mutual dependence of its external and internal form”: Die Verbindung der Lautform mit den inneren Sprachgesetzen bildet die Vollendung der Sprachen, und der höchste Punkt dieser ihrer Vollendung beruht darauf, dass diese Verbindung ... zu wahren und reinen Durchdringung werde.17 The degree of perfection and interpenetration, or of the synthesis of phonetic matter with grammatical form, was, however, soon to be measured (especially through the works of Schlegel and Steinthal) in terms of internal flexion (i.e. the vowel alternations within the stem), which was supposed to lend proof to the superiority of the synthetic languages over other types of linguistic structure.
Despite its reputed atomism, nineteenth-century linguistics also had a fairly clear conception of invariance which it applied historically and not synchronically, for the actual aim of the comparative method was to reduce the variant forms of the IndoEuropean languages to a common prototype, i.e. to original invariants .18 But the subsequent refinement of this method led to a stronger interest in the formulas of correspondence and in the processes of phonetic change than in the reconstruction of the proto-language. The success of the physical and biological sciences inspired, further, the belief that language develops according to natural and immutable laws that are independent of man and society. Imbued with the notion that causality and determinism are the only valid principles of science, linguists were no longer content with predicting the past on the basis of the present, but attempted to establish laws of change valid for all time. To admit any exceptions to the blind and necessary forces of phonetic change was, according to Leskien, to deny the scientific character of our discipline, and the first thesis of the Neogrammarian program enunciated by H. Osthoff and K. Brugmann reads: All sound changes follow laws that are without exception [ausnahmslos] . . . for all speakers of a given speech community and for all words in which the given sound occurs.19 But when Osthoff elaborated on these laws he felt it necessary to specify that the boundary of any given dialect must be defined as narrowly as possible [“so eng als möglich],20 while the second thesis of the above-mentioned program had already made allowance for forms that result from analogical levelling and thereby elude the effect of the phonetic laws. In the words of Baudouin, the vaunted phonetic laws turned out to have a predictive power similar to that of meteorological forecasts (1910 ). The concept of phonetic laws thus underwent, toward the end of the nineteenth century, the same transformations as the general concept of natural law: From a universally valid principle applicable to all cases, it has been demoted to a rule of uniformity and statistical regularity that is not exempt from the workings of chance.21 On a theoretical plane the new scientific mood is most clearly reflected in H. Pauls Prinzipien der Sprachgeschichte (1886). Linguistics, declares Paul, is not, like the natural sciences, a science of laws, but a historical science that deals with unique, nonrepeatable events, and its hero is not abstract society, nor the no less abstract Völkerpsychologie, but man and the individual speech-act (Sprachtätigkeit). For between abstractions there can be no causal relations; these obtain only between real objects and [concrete] facts.22 This attitude, in turn, prompted him to question the possibility of scientific hypotheses and generalizations about language and to deny the very existence of language as a social norm: In reality we have to recognize as many languages as there are individuals.23 The presumed aversion to theory and the call for plain fact gathering are also proclaimed by Osthoff and Brugmann, who urged the linguist to leave the murky circle of his workshop, which is beclouded with hypotheses, and to step out into the clear air of palpable reality.24 And since linguistic reality is made up in its entirety of speaking individuals, the linguist deals, in effect, with only two types of phenomena: the physiological production of speech-sounds which are subject to mechanical change, and the psychological processes of association which generate new linguistic forms. This methodological dualism did not, however, seem to involve any contradiction, for in the final analysis both types of phenomena could be reduced to a common psychophysical or physiological base. The constant interaction of phonetic laws and analogy formation, wrote Wundt, becomes more comprehensible when we consider them not as disparate and opposing forces, but as conditions both of which are in some way ultimately rooted in the unitary psychophysical organization of man.25 The fin de siècle was thus in linguistics (as in other sciences) a period of intellectual disarray in which its practitioners not only gave up faith in the integrating and explanatory function of science, but lost hold of the very object of their study, language. Having renounced the great explanatory schemes of history propounded by their predecessors, they clung to the positivistic belief in the bare fact which finds its explanation in a preceding fact, and to the psycho-physiological view of Ianguage as an activity that is shared by a given speech community by virtue of repetition and imitation. It was this mechanistic and atomistic conception of language that Baudouin de Courtenay set out to correct, even though he did not always succeed in shaking off its influence.
Baudouin de Courtenay has on various occasions summarized the main tenets of his school, but none of his statements of linguistic principles (including the one reprinted on p. 213) can serve as an accurate measure of his theory of language, for they mix polemic and topical pronouncements with statements of a more general and lasting import. As a true philosopher of language, Baudouin was always deeply absorbed in general linguistic questions, and utterances on these questions are scattered in all his books, articles, and even insignificant reviews. Baudouin saw the first and indispensable step toward a renewal of linguistics in its emancipation from the philological and archeological approach that had dominated it throughout the nineteenth century (1909, 237), and in its return to a synchronic analysis of language. The study of living languages, he wrote, must precede the study of extinct languages, as the study of any subject must proceed from the known to the unknown and not vice versa (1897 ). When one is too absorbed in things that happened in the past, he quoted Descartes as saying, one remains quite ignorant of things that happen in the present (1909, 254). He proposed that each subject be first of all treated in terms of its own, intrinsic properties without forcing upon it alien categories (1897 ), for only thus can one gain an understanding of the static laws and forces which act in the synchronic state of a language (1876/77 ) as opposed to the dynamic laws and forces which condition the development of a language. What he held most against so-called comparative grammar was that it works only with several dozens of roots and does not attempt to cover entire languages with fragments but to provide an insight into the entire structure of a given language (1901 ). From this, he said, stem the perverted methods of investigation which characterize philological linguistics (1889 (2) ). Although Baudouin distinguished and emphasized the two dimensions of synchrony and diachrony and the two types of laws and forces that govern their difference, he was never tempted to exaggerate the importance of one at the expense of the other, nor to say with de Saussure that tout dans la langue est histoire, . . . quelle se compose de faits, et non de lois, or that tout ce qui semble organique dans la langue est en réalité contingent et complètement accidental.26 On the contrary, even in his early Programs he pointed out the complementary aspects of permanence and change (1871 ; 1876/77 ), while in his later works he explores in detail the interplay of statics and dynamics and the presence of synchrony (i.e. of stability) in diachrony, and the penetration of diachrony (through the accumulation of historical layers) into synchrony.
Baudouins conception of the tasks of linguistics differed in most other respects from that of his positivistic colleagues, whose seientifie credo was zuerst sammeln, dann erklären. The goal of all science, he wrote, is explanation, because reality [is not] a heap of incoherent and disconnected phenomena (1871 ). Observation and interpretation must go hand in hand, and though it is an empirical science, linguistics must not be content with pure induction but must make the broadest possible use of the deductive method (1871 . He rejected as arbitrary the division between natural science as a science of laws and historical science (including linguistics) as a science of facts. All sciences dealing with comparison and generalization of details, he wrote, are natural sciences if they reveal the organization and regularity hidden in the phenomena, whereas they become historical seienees if they dwell only on accidents and exceptions (1881 (2)). The linguist must not be content with the registration of haphazard and sporadic facts (1877/78 ) but must attempt at each step to discover the genuinely distinctive properties of language and the laws [which] are hidden in the depth, in the intricate combination of various elements (1910 ).
The historical-comparative method, he insisted, is not the sole approach to language but constitutes only a single moment in the history of our science, since comparison is not an end but only a means and cannot be considered the monopoly of one or a few sciences (1871 [56ff.]). The so-called phonetic laws which aspire to the status of natural laws are, in effect, no more than observations about the uniformity of phonetic correspondences (1910 ). Although Baudouin himself appears to have been a believer in strict historical determinism—“given the knowledge of the past, he wrote, we can predict the further internal development of a Ianguage (1876/77 )—he later abandoned this optimistic nineteenth-century viewpoint: The assumption of sound laws operating without exceptions, of gradual, purely physiological change in a definite direction, he wrote, is not confirmed by the actual facts of language and is in contradiction with its socio-psychological character (1927 ). And when he contemplated the direction of morphological change, he formulated for himself—as he tells us—a theory which he labelled evolutiones linguarum terrestrium. According to this theory the formal expression of some grammatical categories keeps alternating cyclically, like a perpetuum mobile, between suffixation and prefixation. The result [of this change] is constant oscillation, vibration and eternal evolution which is reminiscent of the ebb and flow of the tides (1930 ).
In opposing the reigning concern with historical change, Baudouin pointed out that in spite of all the fluctuations and variations, we must note the presence of conservatism (1910 ) and that the concept of statics is applicable not only to synchrony but also to diachrony. The laws of development in time should therefore be viewed as a counterpart to the laws of resistance to historical change, i.e. the laws of dynamic stability (1877/78 ); certain features are exempt from change, for there is a limit to the changeability of sounds (1876/77 ). The formula that something is constantly being changed should therefore be complemented by the formula that something is constantly being preserved, something remains stable (1910 ). The mere registration of phonetic, lexical, and formal differences between languages does not explain the direction of historical change. In describing the development of a given language, the linguist should rather attempt to identify the general tendencies which constitute characteristic and invariant properties (1871 ) and enable us to identify languages or entire groups of languages.
The theory of gradual, uninterrupted, and unidirectional change found, in Baudouins opinion, a striking refutation in the phenomenon of language mixture which he had occasion to observe in vivo during his field work in Slovenia. Linguistic diffusion, he noted, took place both on the axis of spatial contiguity (between speakers of different languages and dialects living in adjacent areas), and on the axis of temporal succession (between speakers of different generations), though he did not fail to notice that it also operates at a distance (among nomads and migrants and between old and modern languages). Distrustful of genealogical reconstructions and bent on refuting the concept of linear development, Baudouin exaggerated the importance of language mixture (though probably no more than his contemporary, Schuchardt). Thus, he ascribed the phenomenon of mazurzenie in Polish dialects and the loss of free stress in West Slavic to the influence of a FinnoUgric substratum, and the formation of Armenian to the mixture of the Indo-European with a Japhetic language (1901 [219ff.]). At the same time he made some truly cogent observations, such as that in linguistic diffusion the victory generally goes to those forms that are grammatically simpler. In addition to explaining the history of certain linguistic facts, the problem of languages in contact was for Baudouin connected with the synchronic issues of coexisting systems and the speakers ability for code switching. In recognizing the coexistence of languages serving special social and expressive functions (e.g., argots, secret languages, languages of traders, and different styles of speech, such as slovenly, rapid, or solemn speech) he rejected the myth, so dear to the Neogrammarians, of a homogeneous society and an average national language (1871 ; 1910 ). He defended the usefulness of international artificial languages (against the attacks of Brugmann, Delbrück, and Leskien), arguing that artificiality in language or mans conscious participation in the fate of a language is only a matter of degree (1871 ; 1907 [256ff.]).
Baudouins position also differed from that of most of his contemporaries with regard to the status of territorial dialects. The phonetic laws were, according to prevailing opinion, most tangibly at work in dialectal speech, which was allegedly least affected by foreign or cultural influences. Dialects were consequently regarded as the most reliable source for the reconstruction of an original language. Only in the veins of dialect speech, wrote Osthoff, flows pure, unadultered blood of millennia.27 The belief that language varies from individual to individual led, on the other hand, to the conclusion (shared by such men as H. Paul, G. Paris, and J. Gilliéron) that dialect boundaries do not exist and are only an invention of the linguist.28 Similar considerations weakened the faith in the accepted divisions of Indo-European languages, a faith some linguists (e.g., Brugmann and Förstermann) tried to salvage by comparing the largest possible number of diverse features. One must view with utmost scepticism, wrote Delbrück, all classifications of the Indo-European languages except the Asiatic group, whereas Meillet was of the opinion that the only valid criterion for defining a linguistic unity was le sentiment des sujets pariants.29 Baudouin, on the other hand, who was an experienced dialectologist and who came into contact with mixed German-Slovenian and Italo-Slovenian dialects, recognized that dialects could be described as independent linguistic entities, от what he called linguistic individuals (1875, IV). With respect to the Indo-European languages Baudouin also questioned the established classificatory schemes, quipping that A. Schleicher built his classification (the tree theory) on wood, whereas J. Schmidt built his (the wave theory) on water (1909, 227), and arguing that the genealogy of languages is not as simple as the pedigrees of nobility (1884 ). Genealogical classification was for him only one of the possible, and by no means the most important, way of grouping languages. At the present state of our science . . . , he wrote, one cannot speak (in most cases) of a genetic classification in the strict sense of the word, but only of a scientific characterization of various groups of languages and dialects (1877/78 ), for what is important is to identify characteristic traits that are of general significance that pervade the phonetic and morphological structure of language and not disparate details of a secondary nature, or facts which are mere surrogates of the general features (1884 ; 1871 ). He implemented this program by providing the first typological description of the Slavic phonemic systems and a brilliant outline of the phonological history of a Slavic language (Polish). What he tried to emphasize in his diachronic studies was not so much the phenomena of change as the phenomenon of dynamic stability, and the common tendencies which define the limits of historical change and determine the unique structure or composition of a given language or group of languages (1877/78 ).
Just as he opposed the reduction of language to an aggregate of disparate facts, he resisted the efforts to treat the language of a community as a sum of individual speech-acts. Language as a complex of constituent parts and categories, he wrote, exists only in potentia (1871 [68, 78]; 1889 (2) ), and must be distinguished from the continuously recurrent process of social communication and knowledge and understanding of languages differ from [their] command more or less as knowledge of physiological processes differs from their performance (1870 ). But the distinction langue/parole was not, in his opinion, reducible to the opposition social product vs. individual activity, for he recognized that the speech-act itself is a social act, and that language opens to the speakers possibilities of choice which presume an active and creative process. Language, he said, is throughout a psychological-social phenomenon (1889 (1) ; 1889 (2) ; it is both a tool and an activity, (1907 ) because thinking and membership in society are the two basic conditions for language in the proper sense of the word (1889 (2) ). The opposition between the individual and the collective will be overcome if we adopt the concept of collective individuality՛ (1910 ). The feeling for language is not merely a clever formula or will-o-thewisp but a real and objective phenomenon of language (1902 ). Thus he returned to Humboldts conception of language as a dialectical unity in which the I and the Thou, the individual and the social, the inherited and the acquired constantly define and presuppose each other.
Even the most ordinary and least scientifically trained mind performs tentative, unconscious, and partly conscious [linguistic] operations, Baudouin wrote in 1889; the science of linguistics introduces nothing new in this respect; it only perfects and refines our thinking [about these processes], frees it from the accidents of chance, and substitutes a chain of conscious and clearly defined concepts for a mass of vague and diffuse ideas (). Linguistic activity is not, in other words, mere repetition and imitation but a process of integration and systematization, of uninterrupted semantization and classification in which every word (be it even a nonsense word) is interpreted in terms of its phonetic and structural (i.e., morphological) affinity to other words. The phenomena of folk etymology and analogical levelling (which Baudouin called morphological assimilation), the grammatical mistakes made by children and the reinterpretation of foreign words are not, therefore, to be treated as a sign of linguistic pathology, but rather as a sign of the creative ability of the speaker to group isolated words into semantically well-defined patterns (1915, 188ff.). This grouping, according to Baudouin, involves two principles of association that were first formulated by Kruszewski (who adopted them, in turn, from the English philosophers): association by similarity and association by contiguity. These two laws of association make the mass of words into a harmonious whole of coordinate systems (or nests) and ordered series (or rows) (1889 (1), 184ff.). But unlike Kruszewski, who applied these principles only to morphology, Baudouin recognized their validity for phonology as well, because, as he pointed out, the laws of grouping and equilibrium apply to all the elements and levels of a language (1889(1) 185).
4. Baudouin de Courtenays principal contribution to modern linguistics is his elaboration of a theory of phonology and morphophonemics which has served as a point of departure for all subsequent research in these twin branches of linguistics. His work on the role of sounds in the structure of language was, however, a direct outgrowth of the nineteenth-century development of phonetics and the phonetic orientation of comparative grammar, which in the seventies had become bogged down in practical difficulties and internal contradictions.
The phoneticians of that time (mostly students of the physiological and physical aspects of speech) were inspired by a quest for maximal accuracy and objectivity, which could be reached, they believed, by recording and classifying the minutest, microscopic nuances of the speech-sounds and by analyzing them into ultimate articulatory activities. Like the chemist, writes Techmer, who decomposes cells into molecules and atoms, the phonetician is interested in the simplest physical properties of sound matter,30 whereas Osthoff exhorted the linguist to catch the smallest nuances of live language with photographic faithfulness and precision . . . and to put into fetters the fleeting, changing sound [den beweglichen, lebendigen Sprachlaut in Fesseln zu schlagen],31 a program which is reminiscent of the impressionist ideal of the time to capture on canvas all the transient visual impressions with an innocent eye (Monet, wrote Cézanne, nest quun oeil, mais quel oeil).
The realization of this ideal turned out to be elusive when the phonetician was confronted with the practical task of describing previously unanalyzed dialects or primitive languages, which imposed the need for a unified system of notation and which prompted research on international, rational alphabets that would employ a manageable number of distinctive signs. This dilemma between theory and practice was clearly stated by Ascoli: In describing dialects [the linguist faces] only two real difficulties: to reconcile the requirements of physical precision which are by nature [le quali sono di lor natura] infinite with the needs of history and practice, to eliminate the uncertainty of acoustic perceptions.32 Skepticism in the value of physical measurements was also expressed by Sievers: Even the most subtle acoustic investigations will not yield to the linguist any more useful material than the simple subjective evaluation by ear.33 Because of the constant fluctuation of the sound, the researcher, he concluded, must select only that nuance of sound which is its typical representative.34 The isolation of the typical representative of a sound, the distinction between its essential and accidental properties (Winteler) or, finally, the quest for sounds which are in gegensätzlicher Verwendung (Sievers) or independently significant (Sweet) imposed itself in the seventies of the last century as an inescapable problem that demanded theoretical solution, and prepared the ground for Baudouins phonological theory.
The more immediate impulses for this theory came, however, from the contemporary debate concerning the role of analogy in phonetic change, or more generally, the interrelation of the psychological (i.e., morphological) and physiological (i.e., phonetic) aspects of language. The purely historical and physical approach of the comparativists to phonetic change was first most seriously challenged by de Saussures monograph, in that it (1) treated proto-Indo-European as a synchronic system; (2) discussed the vowel changes in terms of morphological alternations; and (3) posited the existence of purely relational, unspecified phonetic units. After the appearance of de Saussures monograph, it was no longer possible to treat phonetic change apart from morphology, or to claim that analogy is an exception to the regularity of phonetic laws. The opposite seemed now rather to be the case, namely, that phonetic change disturbs the regular relations of the grammatical forms, while analogical levelling counteracts the blind force of phonetic change or, as de Saussure put it:
Le phénomène phonètique est un facteur de trouble. Partout оù il ne crée pas des alternances, il contribue à relâcher les liens grammaticaux qui unissent les mots entre eux ... le mécanisme linguistique sobscurcit et se complique dans la mesure оù les irregularités nées du changement phonètique lemportent sur les formes groupées sous des types généraux.35
De Saussures ideas fell on fertile ground in Kazan, for they converged on many points with those of Baudouin and Kruszewski (the latter wrote an enthusiastic review of de Saussures work in 1880). Kruszewskis study, Über die Lautabwechslung (Kazan, 1881),36 provides the most lucid summary of Baudouins and Kruszewskis solution to the question of analogy and phonetic change and of the phonological theory to which it gave rise.
In the synchronic state of a language, Kruszewski argues, there are two types of sound-changes, or to be more precise, of alternations of sounds. The first type (which Kruszewski calls divergents) involves imperceptible, microscopic variations of one and the same sound and is due to different anthropophonetic (i.e., purely physical) conditions, whereas alternations of the second type (the so-called correlatives) involve phonetically different sounds and are due to different morphonetically conditions. An example of the first type is the change of s to z in intervocalic position in the German Haus: Häuser (where s is the primary and z the secondary divergent), and an example of the second type the alternation of s with r in such German forms as gewesen: war (where s is the primary and r the secondary correlative). The mistake of the historical linguist lies, according to Kruszewski, in that he applies the criteria of gradual phonetic change, which concern the alternations of the first type, to those which belong to the psychological sphere and are phonetically dissimilar and discontinuous. The correiatives are neither exceptions to the regular (or pure) phonetic laws, as argued by the Neogrammarians, nor are these laws less regular at the present than they were in the remote past, as claimed by Curtius. Both types must be viewed as being equally regular and without exceptions, their difference consisting only in that the variations of the first type are conditioned (or as we would say now, predictable) phonetically, whereas the modifications of the second type are conditioned morphologically and are thus associated with different grammatical functions. Between the alternations of the first and second type there is, indeed, a necessary relationship of a historical order, for a spontaneous process of degeneration inevitably transforms the variations of the first type into alternations of the second type. Kruszewski subdivides the latter alternations further into correlatives which express minor internal differences (i.e., different lexical items), and into correlatives which render broader internal differences, or those between entire families of words. Kruszewski thus draws a sharp distinction between the combinatorial, contextual variants which are incapable of carrying grammatical functions and independent phonemes which are associated with grammatical distinctions. Without the concept of the phoneme, he concludes, it is impossible to study either phonetics or morphology (45).
It is through the study of alternations that the perennial question of the relationship of sound and meaning, the signans and the signatum, was thus again brought into sharp focus, serving as a point of departure for a deeper and many-sided functional interpretation of the role of the sounds in language. Baudouin was the first in modern linguistics to recognize that sounds and their combinations, or the sensory, external, peripheral aspect of language, means nothing by itself (1889 (2) ), and that languages are neither phonetic nor acoustic in their nature (1910 ), but that the proper function of sound is to serve the central, psychological level of language, i.e. that distinctions of sounds are connected with distinctions of meaning. According to Baudouin the semiotic function of sound was first understood by the Hindu grammarians, who by means of analysis (vyākara ṇa) and abstraetion tried to establish the sounds and sound complexes that serve to modify meanings, and who to a large extent surpassed all that was accomplished after so many centuries by European scholars (1909, 111). The role of sound in the mechanism of language (1871 ), and the interrelation of the meaningful and meaningdifferentiating elements of languages was to absorb Baudouins attention during his entire life, starting with his first paper on consonan tal alternations in Polish (in 1869) up to his late inquiries on the morphologization and semasiologization of phonological elements on the one hand, and the phonologization of morphological elements on the other (1915, 177-78).
In the same year, 1881, Baudouin published his lectures on the comparative grammar of the Slavic languages (Nekotorye otdely . . .), in which the interpretation of the alternations resembles closely the one offered by Kruszewski. The alternants (or comparents) are still treated from both a comparative and intralingual viewpoint with an equation established between the etymological correspondences and the alternating correlatives of homogeneous morphemes. The phoneme is still defined here as a mobile component of a morpheme and a marker of a given morphological category. However, the same study also makes an attempt to broaden the definition of the phoneme beyond its purely morphophonemic framework. Baudouin proposes here to limit the use of the term sound only to conditions of anthropophonetic dependence. The linguistically autonomous sound, or the phoneme, can on the other hand be arrived at only by purging it of the accidence of divergence which splits the unified sound into its anthropophonetic varieties. This unified, discrete sound is the result of abstraction and of phonetic generalization, or the sum of generalized anthropophonic properties. As such it must be distinguished from that other notion of the phoneme which treats it as an alternating component of a morpheme. In the future, Baudouin concludes, it will be necessary to have different terms and a different notation for these two aspects of the concept of the phoneme (1881(2), 122).
The theory of alternations received the fullest treatment in Baudouins An Attempt at a Theory of Phonetic Alternations (1895), through which he, no doubt, also tried to assert his claim to the authorship of the theory vis-à-vis Kruszewski. The principal emphasis of the essay is on the interdependence of the diachronic and synchronic aspects of alternations, for, as the author tells us in the Introduction, the former aspect had been ignored by the Hindu grammarians (who lacked a feeling for history and chronology), while the latter had been ignored by contemporary European grammarians. What Baudouin specifically examines is how the phenomena belonging to the axis of succession (Nacheinander) acquire different functions and coexist on the axis of simultaneity (Nebeneinander). For the periods of development, he had written in another context, do not replace each other, as one sentry replaces another, but each period creates something new, which in the imperceptible transition from one stage to another prepares the basis for subsequent development (1871 ). The synchronic state of a language is the result of a historical process in which the same means are used for different ends (1871 ) and in which, as in a geological cross-section, layers of different historical periods exist side by side as mobile and productive formations and as unproductive, inert residues that resemble extinct volcanoes.
According to Baudouin there is an unceasing tendency to endow sound matter with semasiological and morphological functions so that the divergents become in time converted into traditional alternations, and the latter into mobile correlatives. The correlatives themselves may eventually replace each other, as when the Sanskrit gu ṇa is replaced by ʋṛddhi, or when the Slavic k ~ č,k ~c alternation is replaced in Russian verbs by the more general k ~ k alternation (by analogy to verbs which employ the alternation of hard and soft consonants). A peculiar place is in Baudouins triadic scheme assigned to the traditional alternations (i.e., alternations which are maintained only by force of tradition), which hover between the anthropophonic divergents and the psychophonetic correlatives as an unmotivated irrational type that has no clear-cut morphological function and is doomed to be either eliminated or converted into productive correlatives. Intent on drawing a sharp demarcation line between the phonetically conditioned and morphologically motivated types, Baudouin failed to examine in any detail the complex, systematic relations between the various types of alternations and to discover that, like other elements of language, the alternations form a dynamic, hierarchical system which is correlated with the stratification of grammatical categories. A more serious methodological shortcoming is Baudouins tendency to list the alternants but to omit rules that would correlate and derive one set of alternants from another, more basic set. If it is claimed, Baudouin writes, that the č in pieczę, rączka is derived from k, then one could with equal right insist that, on the contrary, the k of piekę, ręka is derived from č. ... 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) (1895 ). In a synchronic description, he insisted, we must operate with purely phonetic difference between etymologically related morphemes, and not with change. Note that a similar approach to the alternations is taken by de Saussure: II est trés incorrect de dire, comme on le fait volontiers, que la a de Nacht se change en ä dans le pluriel Nächte : cela donne lillusion que de lun a lautre terme il intervient une transformation. ... En réalité nous avons affaire à une simple opposition de formes.37 It is only in his later work that Baudouin came to posit the question of the irreversibility of alternating relations and to emphasize that the Polish u or š can be predicted from о and x (as in the forms koza/kózka גmmucha/muszka) but not vice versa. Baudouins reluctance to operate with base forms and rules is understandable if we keep in mind the novelty of his theory and his apprehension that any reference to change or to prior and derived forms might be misinterpreted as meaning historical phonetic change (which happened anyway). Another reason why Baudouin might have rejected an approach that operates with primary and secondary alternants was that he found it wanting also with respect to primary and secondary divergents which were posited by Kruszewski and which, Baudouin must have felt, could not lead to a correct interpretation of the phoneme. That this was indeed the case is apparent from Baudouins critical review of Kruszewskis work written in 1889, which marks in many ways a new phase in the development of Baudouins phonological theory. In this review Baudouin takes exception to Kruszewskis one-sided approach to the phoneme as a member of alternations, and to the putative distinction set up between sounds as purely physical phenomena and sounds that possess, because of their morphological function, a psychological reality. Linguistics, he argues, is throughout a socio-psychological science. Kruszewskis anthropophonetic bias prevents him from recognizing the linguistic status of both the phonemes and the divergents. He makes an unjustified and illogical jump and switches from the central, psychological aspect of language to its periphery when he divides the morphemes into plain, bare sounds in the way he divides sentences into words, and words into morphemes (182 ;183). It is, perhaps, only in the language of a talking machine or phonograph that morphemes, words, and sentences are made up of complexes of physical, semantically irrelevant sounds. Kruszewskis acoustic approach causes him also to treat speech-sounds as ultimate indivisible units, whereas in the language of physiology and analysis, the sound is a complex of heterogeneous, but coordinated activities, which must, in turn, be reduced like all elements in language to the common denominator of their meaningful associations (184). Kruszewski precludes the possibility of precise analysis when he treats the phonemes of language as if they were species of animals and plants. Phonetic change is not the result of natural laws, but of the different complexity or clarity ( okreś lono ść) of the sounds themselves. Equally misleading is Kruszewskis attempt to group the phonetic divergents into a single unit on the basis of their phonetic similarity or anthropophonetic affinity. This criterion, according to Baudouin, is completely false and without any foundation, and is ultimately based on the subjective opinion of the investigator. Kruszewski himself indicates the inadequacy of this criterion when he states that it applies to most cases, for if it does not apply to all cases, it cannot serve as a distinctive feature (165). Kruszewski misses, further, the linguistic relevance of the divergents when he describes them in purely physical terms; for though the divergents are not associated, like the correlatives, with morphological functions, they serve as linguistic cement and as markers of the structure of the word or of the sentence (181; 166).
This review marks Baudouins departure from the naturalistic conception of the sounds of language, which was still rampant in the phonological theory of the Kazan period, and anticipates most clearly the tenets of modern phonological theory. Having abandoned, however, the physicalist approach to the speech-sounds, Baudouin came to embrace a psychological viewpoint which still left unresolved the tension between their psychological and physiological aspects, i.e. the relation between the phoneme as a psychological entity and the phoneme as a complex of articulary-acoustic properties (1889 (2) ).
Baudouin never tired of emphasizing that the phonemes are combinations of more basic articulatory-auditory activities of a definite type, that they are not like separate notes, but like chords composed of several elements (1910 ). The basic units of language are, for him, in a descending order, the syntagms, the morphemes, and the phonemes:
However, requirements of scientific analysis ... do not allow us to stop with the phonemes. The phonemes consist of ultimate psychological (articulatory and acoustic) elements which, from the point of vitw of pronunciation . . . , are kinemes, and from the point of view of perception are acousmemes. I consider these terms indispensable for the greater precision of the abstract concepts of our science. (1910 ).
He furthermore emphasizes that what is semasiologized and morphologized are not the phonemes, but their articulatory-auditory elements (1922 [324ff.]), which are bound and interconnected with each other (1910 ) and which form a system of oppositions (1876/77 ; 1922 ). In his treatment of the phoneme as a psychological entity, he imputes to it, however, properties which do not appear in the actual speech-chain, but are projected on the basis of fuller, nonelliptic forms (1929 (2) ), or because it alternates with other phonemes in grammatically related forms as, for example, the feature of voicing in final devoiced consonants, or the presence of zero phonemes in various derived and inflected words. Furthermore, he never explains why this posited capacious mental entity should encompass only the distinctive features of a sound, and not include all the other, nondistinctive elements that are present in the speech-chain.
Baudouins conception of the phoneme as a complex of articulatory-acoustic properties thus only superficially resembles the modern theory of the phoneme as a bundle of distinctive features which was introduced into linguistics by R. Jakobson. While for Jakobson the distinctive features are the invariant properties of a speech-sound which are present in all its manifestations on the motor, acoustical or auditory level,38 for Baudouin they remain representations of acoustic-articulatory attributes that are divorced from the actual speech-sounds and are unified into a sound-image only in the mind. Baudouin does, it is true, often assert that in considering the psychological aspect one must not forget the social aspect of language (everything constant belongs to the psychological world . . . but is possible only in society (1915, 157), but the twentieth-century dichotomy between an internal, psychological and an external, physiological (or physical) reality is not thereby overcome. That is why one also finds in Baudouins writings a tendency to identify the properties of the phoneme with those of its primary variant or imagined psychological unit (1895 ), or to group the variants in terms of their presumed psychological similarity (although the Russian i and у are pronounced differently, they are psychologically closer to each other than they are to other sounds (1903, 36-37)). This uncertainty and vacillation also explains why the phoneme was identified with the primary, contextual variant also in the work of some of Baudouins students (e.g., Ščerba). One must not forget, on the other hand, that a whole generation of linguists after Baudouin kept groping for an adequate definition of the phoneme, defining it either in physical terms (as a class or family of phonetically similar sounds), or in psychological terms (as an ideal sound), or as neither one nor the other, i.e., as a fiction.
As far as Baudouins psychological theory of the phoneme is concerned, we may discern in it a weaker and a stronger variant. In his earlier writings Baudouin merely states that an analysis from a psychological viewpoint (as opposed to a purely phonetic analysis) deals with the relationship between the psychological content of language and the psychological equivalents of sounds, i.e. with the sound-images (1884 ) ,or that the phoneme is a psychological equivalent of the sound complex (1899, 351; 1903, 39). The psychological significance of the phoneme lies here primarily in its connection with the meaningful, central, psychological level of language. Baudouin himself had on various occasions expressed reservations with regard to the psychological dogma that dominated late nineteenth-century linguistics. Linguistics, he wrote, must be regarded as an independent science, not to be confused with either physiology or psychology (1871 ); the question of what is psychological in the strict sense must be left to philosophers and psychologists (1884 ) because linguistics has its own psychology (which he proposed to call glotto-psychology (1909 (2), 267)), and, although linguistics may be considered a socio-psychological science, psychology and sociology are only its ancillary sciences (1889 (2) ).
In its stronger variant, Baudouin defines the phoneme (or phonemes) as a sound of the same intention, but of different anthropomorphic realization (1895 ; 1910 , as phonetic notions which exist objectively in mens souls (1910 ) and as a sound-image which is mobilized in the mind not being strong enough to reach the world of the sense or to become perceptible (1929 (2), ), as or the periphery of the speech apparatus (1929 (2), ). In refusing to view sound as a sum of physical data governed by natural laws, Baudouin thus came full circle to embrace a view of the phoneme as an incorporeal and inaccessible mental image, a noumenon which is prior to and richer than the world of phenomena. In this too, then, he found himself in accord with Humboldt and his idealistic conception of the speech-sound: Man muss die Sprachbildung überhaupt als eine Erzeugung ansehen, in welcher die innere Idee, um sich ze manifestieren, eine Schwierigkeit zu überwinden hat. Diese Schwierigkeit ist der Laut und die Überwindung gelingt nicht immer in gleichen Grade.39
In departing from the earlier morphophonemic interpretation of phonemes which he had shared with Kruszewski, Baudouin was free to examine the speech-sounds in their mutifarious aspects and functions. In the lectures for the year 1876/77 he had already broached the question of the dynamics and mobility of sounds, i.e., their role in phonomorphology as opposed to the statics of sounds, which deals with the parallelism of sounds, and the distinctive properties that give rise to certain oppositions (). He repeatedly underscores the distinctive capacity of sound, which goes beyond its morphological function: Psychophonetic alternations are one of the morphological means on a par with other inflectional and derivational devices and belong, in fact, to morphology, not to phonology, whereas it is the task of phonology to define those phonetic features that are associated with morphological nuances and differences (1899, 360) and morphologization affects only certain articulatory-auditory distinctions . . . while semasiologization affects all articulatory-auditory distinctions (1927 ; 1912, 174). Baudouin cites the use of rhyme, alliteration, and other poetic devices as evidence of the double affinity of sounds (1884), i.e., of their phonological-morphological and phonological-semasiological functions. The latter function is illustrated by such minimal pairs as Russian tom/dom, bal/pal, sdal/stal, den/ten (1912 (2), 174; 1927, 395). The morphophonemic processes of a language use the distinctive features selectively, as, for example, in Russian, where they involve primarily the features of palatalization and of stressed and unstressed vowels (1917, 254). Under statics of sounds Baudouin considers also the different utilization of sounds in various grammatical classes and forms, or the association of certain phonetic elements with certain morphological forms (1915, 205ff.) and the role of prohibitive analogy, which prevents phonetic change in some grammatical forms (1903, 41). Sounds are, further, distinguished in terms of their active or passive role in the system, which accounts for their different psychological intensity and strength (1884 ). The sounds that occur in grammatical morphemes carry a psychological stress; they are the strong phonemes which serve to differentiate morphological categories and are resistant to change (1884 ; 1899, 358; 1903, 40; 1910 ; 1929 (2) ).
The phonemes of a language, according to Baudouin, form groupings of paired and nonpaired sounds which are at the basis of the equilibrium of language (1877/78 ). Certain oppositions are basic and stable, whereas others vary and fluctuate, and are bound to disappear (1889, 188; 1910 ). The difference in the phonetic make-up of languages, according to Baudouin, can serve as a useful criterion for a typological characterization of languages (1876/77 ; 1902 ). Phonetic relations can also be studied statistically (1876/77 ; 1876/77 ). Thus one may compare percentages of various sounds in a given language, the ratio of single phonemes to clusters, of kinakemes to phonemes, of phonemes to syllables, or of syllables to phonetic words (1876/77 ; 1915, 171).
The sounds of language may further be utilized in a syntactic function, to mark boundaries between words and word-groups (syntagms), or as phonetic cement (1876/77 ; 1881 (1), 149; 1889 (1), 181; 1908, 171), i.e., as features which signalize units of the higher, semantic level of language and which integrate them into wholes. These, as we would say now, configurational and redundant functions of the speech-sounds Baudouin studied from the point of view of word and sentence morphology (morphology of the second degree). For example, stress, which in Russian is morphologized and semasiologized, is also utilized to indicate the unity of a phonetic word, whereas in Polish it performs only the latter function; a similar function is in the Turkish languages, performed by vowel harmony (1876/77 ; 1908, 171). Voicing, which is distinctive in Polish, can at the same time mark a word boundary in some of its dialects (e.g., gloz narodu vs. wiosna; 1908, 171). Baudouins concern with the stability and instability or fluetuation of sounds inevitably led him to the study of the facultative sounds of language (1929 (2) [ 291ff.]), though he obscured this problem by reference to latent psychological forces that strive to, but cannot, be realized in the speech-act. As was his wont, he examined the question of stylistic variation from a variety of viewpoints: from a paradigmatic and syntagmatic point of view (e.g., the varying stability of phonemes and the tendency to weaken some of them at the end of a word); with reference to morphology (e.g., the omission of phonemes in certain grammatical forms); with relation to writing and other conservative forces which contribute to the preservation or restoration of some phonemes. The existence of faculative sounds was to him further proof of the dynamic character of language and of the intrusion of the factors of time and space into synchrony.
Baudouins inquiries into phonology and morphophonemics are closely related to his work on writing and on visual language in general. Though he devoted most of his efforts to clarifying the role of sound in language, he was, like de Saussure, aware of the importance of signs that employ other physical means and that supplement or even replace the verbal sign (as in the case of deaf mutes). He was not affected by the mystique of the speech-sound that was the legacy of the Romantics, and he predicted the vindication of the rights of visual language, which is now held in contempt by linguists who still confuse letters with sounds (1909, 264). He envisaged two lines of approach to the study of visual language: one that would explore gestures that accompany the spoken word or are used in its stead; the other that would study systems of writing and various types of scientific notation. To the last two Baudouin devoted a number of penetrating studies that deal both with the orthographic systems of various Slavic languages (especially Russian) and with general theoretical questions. In considering the role of writing in the perception and use of spoken language, he urged attention to the three types of members of a speech community: language-learners who are in the process of acquiring spoken and written language; illiterates who are only in command of spoken language; and the educated who have a mastery of both spoken and written language (1909, 265). Even the simplest elements of writing, Baudouin declared, are the result of a deeper analysis of the complex linguistic representations (1910 ). Writing systems that refer to the phonemic level of language are phonemographic, whereas those that refer to the semantic level are morphemographic or syntagmographic. Systems that utilize only one principle of representation are monistic, whereas most systems incorporate more than one principle and are dual or pluralistic, 1917, 280; 1929 (1) ). Baudouins concern is directed primarily to phonemography, i.e. to the type in which the division of the sentence into syntagms or syntactic elements is not taken into consideration (1929 (1) ). He points out that the letters, or graphemes of conventional alphabets are not the equivalents of phonemes, for the latter consist of more elementary components, while the former are global, synthetic units. But in addition to synthetic graphemes, some languages make use of analytical, diacritic graphemes which correspend to the distinctive features; e.g. the Polish diacritic marks in ą , ę,ś etc. (1915, 211 ff.). The graphemes, like the phonemes, may also indicate morphological relations, as in the European alphabets which employ capital letters, or as in Arabic or Hebrew, where special letters are used to mark word boundaries. Of equal interest to Baudouin are those writing systems that omit certain features, as for example the Semitic system, which ignores vowels, or those of various European languages that do not mark the prosodie features. Baudouin constantly calls attention to the confusion of concepts which results from the identification of phonemes with letters. On the other hand, he emphasizes the conservative role of writing and the existence of double—written and spoken—norms, and he reminds us that without the representation of graphemes there would not have been formed in our minds the representations of phonemes as discrete articulatory-acoustic units (1915, 211). In this way he makes the point that as a derived, secondary system which selects and gives prominence to certain features of the verbal code, writing itself exerts an influence on that code.
5. We shall now consider Baudouin de Courtenays contribution to morphology, typology, and diachronie linguistics, which were no less pioneering than those he made in phonology and morphophonemics. In the nineteenth century, morphology was the only branch of linguistics that had a direct bearing on typology, as it provided the criteria for the classification of genetically unrelated languages. Historical linguistics, on the other hand, was a strictly genealogical, reconstructive science, and it was Baudouin who gave it a broader, typological outlook.
Baudouins ideas on grammar and morphology are intimately related to his synchronic approach to language and must be seen, like his phonology, as an attempt to outflank the difficulties which emerged within traditional comparative grammar. The morphological analysis of the Indo-European word, which was so brilliantly initiated by Bopp, gave rise to doubts by the sixties and seventies of the last century about the possibility of an objective reconstruction of the Indo-European inflection and the decomposition of the word into its ultimate components. Linguists like Fick and Sayce had come to question the empty clatter of stems and suffixes,40 which to B. Delbrück had become no more than useful operational terms without objective reality. In his disillusionment Delbrück called for a return to the grammatical practice of the Greeks, for whom the word, rather than its parts, was the ultimate unit of language, while adding that the Greeks were naive, whereas we are resigned.41 Far from rejecting the attainments of historical grammar in breaking up the word into its morphological components, Baudouin sought rather to put the analysis of the word on a realistic, synchronic basis. He had little patience with Delbrücks pessimism and criticized him for ignoring the objective-psychological factors which make morphological analysis possible (1909, 259). Baudouins main effort during his Kazan׳ period was, in fact, directed toward a deeper and general linguistic interpretation of the morphological structure of the word.
In his review of Kruszewskis work (in 1889), Baudouin points out that the proper title of Kruszewskis Outline of the Science of Language (1883) should have been The Word and its Components, for the problems we discussed in Kazan concerned mostly the structure of the word and not the totality of language (1889 (1), 176). He reiterates the idea in his remarks on Bogorodickijs work: Analysis of complex wholes into their components has been the central interest of the Kazan school . . .; analysis, decomposition into features is in all sciences the beginning of precise investigation (1903, 54). But in further elaboration of his ideas Baudouin progresses inevitably toward a more comprehensive view of morphology as a branch of linguistics that encompasses not only the external, formal side of language, but also its internal, conceptual aspect. Having thus begun as an adherent of Steinthal and under the shadow of Schleicher, who emphasized the material morphological differences between languages, Baudouin at the end of his career moves closely to the Humboldtian conception of inner form as the organizing force of the external elements of language, and thereby concludes the great cycle of nineteenth-century linguistics, which moved from the general and universal to the individual and particular, and then back to the universal and general properties of language. While Baudouins first paper (in 1869) dealt with the morphological analysis of the Slavic word, one of his last papers (1930) boldly resumes the question of the influence of language on the organization of our perceptions and knowledge. But in this paper, as elsewhere (1909, 187ff.), Baudouin rejected Humboldts idealistic interpretation of the relation of thought and language, treating language rather in the rationalist and modern spirit, as a system of signs, and linguistics as a pivotal science among other semiotic sciences. Linguistics, he wrote, is one of the sciences of mental, human phenomena which are expressed by means of signs and symbols drawn from the physical world (1909, 247-48), and in the future, he predicted, a linguistic mode of knowledge will take its place alongside the intuitive-artistic and the analytic-scientific modes of cognition (1909 (2), 270; 1904 ).
In his early work on morphology Baudouin sought first of all to identify the basic units of the sentence and the word. Words and fixed expressions are the syntactic atoms (or syntagms), whereas roots in the broad sense are the morphological atoms (1877/78 ; 1912, 19) of language. To the last he later assigned the term morpheme, which would cover any component of a word (Meillet found it to be a joli mot and borrowed it in his translation of Brugmanns comparative grammar). Baudouin questioned the Indo-Europeanist theory according to which only the endings are subject to change, whereas the stems are of a fixed, immutable nature. In a study submitted for publication to Schleicher in 1869 (which was turned down for its unorthodox views and was published only thirty years later), Baudouin tried to demonstrate the historical transformations that the Indo-European nominal stems have undergone in Slavic as a result of phonetic and morphological processes. In this study, which advances the theory of reinterpretation (pererazloženie) or morphological absorption (later developed in greater detail by Kruszewski), Baudouin is able to show that far from being opposite and antagonistic forces, phonetic and analogical developments do interact to produce a new type of (consonantal) stem and new types of endings. Through this study Baudouin posits, in addition, the question of the exchange of functions between the lexical and grammatical elements, a question that was broached only later by Meillet.42 He also puts into a new context the problem of analogy that had been raised by the Neogrammarians to the rank of a psychological law on a par with the phonetic laws (though they used it only as a last resort), and which they treated as a kind of quantitative proportion between disparate forms selected at random. For Baudouin analogical levelling is primarily a qualitative, grammatical process (1903, 47) which is possible thanks to the existence of fixed yet mobile systems (1908, 164), i.e., morphological patterns that allow the speaker to interpret and to regroup the elements of a word. Reinterpretation, analogy, and folk-etymology are for him the three types of processes that prove the reality of the morphemes and show that linguistic creativity takes place not only in the field of syntax, i.e. in the combination of ready-made words into phrases and sentences, but also in the combination of morphemes into words (1917, 281). The live and constant process of semantic interpretation and reinterpretation is continually confirmed by the use of puns, the creation of new roots, and the mistakes in the language of children (1876/77 ; 1903, 44ft.; 1917, 2î4à©¤© Baudouin examines and classifies the morphemes of a language according to various criteria: their central vs. peripheral position in the word; their degree of independence; and their lexical, syntactic, or morphological functions (1909 (1), 183ff.; 1912, 85ft.). Morphology in the broad sense is defined by Baudouin as including both the study of word structure (i.e. morphology proper) and the study of sentence structure (syntax). Morphology together with semasiology and lexicology pertain to the central aspects of language, to its psychological content (1884 ; 1889 (1), 200). Of these three fields only morphology is truly a part of linguistics, since it exists exclusively in language and nowhere outside Ian֊ guage (1909, 182), whereas semasiology (which includes the study of grammatical categories and the parts of speech) deals with extralinguistic representations. This distinction, which is drawn in most of Baudouins early writings, must be seen as his tribute to the nineteenth-century tradition (by no means dead even now), which identified the study of structure with that of its external form, relegating the study of meaning to psychology or to other sciences. In his later works Baudouin abandons this antinomy, emphasizing the linguistic reality of grammatical concepts (1922; 1930), and defines morphology more generally as the combination of objective phonology and objective semasiology. These two, he writes, constitute the basis of linguistics as a separate seience (1915, 166). The question of the correspondence between these two qualitatively different (różnogatunkowe) representations (1915, 219) and of its expression in different languages is one of Baudouins central preoccupations, one that, as we have seen, guides his entire approach to phonology and morphophonemics. The formal and meaningful elements of language are, according to Baudouin, in constant interplay, yielding either complete harmonу or disorder and chaos which languages perpetually strive to correct. The optimal type of correspondence is achieved in the Uralo-Altaic languages, which are for Baudouin (as they were for Sapir) sober and orderly, whereas the Indo-European languages that were so glorified by the Romantics as the realization of the synthesis of matter and form present a picture of disorderly linguistic thinking (1909 (1), 185; 1922, 82). Despite his bias in favor of the agglutinative languages, Baudouin succeeds in unravelling the systematic, though far more complex, harmony in such a typically synthetic language as Polish, where he shows that the difference between various parts of speech and grammatical categories is matched by a corresponding difference in the structure of their stems, by their different types of endings and morphophonemic processes (1915; 1922). In his study on the Influence of Language on World-View and Mood (1929), Baudouin posits the question of the relation of language to thought, and examines the nature of various grammatical categories. Language, he points out, must not be identified with the world of outside reality, for it is a system of abstract ideas, a mythology which, like Midas converting all he touched into gold, converts all our experience into linguistic form. The grammatical categories of a language are obligatory and affect every lexical item, like the obligatory designations of religion which are written in passports even of professed atheists (234). Baudouin illustrates the obligatoriness of grammatical meaning with various types of gender that give rise to etymological myths and put their stamp on religious beliefs, on poetry, on folklore, and on the arts. The grammatical categories are further discussed in terms of their mutual, binary relations, within which the particular categories are opposed to each other as specific and less specific, i.e., as marked and unmarked terms (1922, 122ff.). Within the category of person, for example, only the first and second persons designate persons proper, while the third person is opposed to these as a non-person, which explains why the third person also figures in impersonal constructions. Within the system of tenses the present is opposed to the past and the future as a tempus aeternum that may designate either, while in extralinguistic thought the present does not even exist, for it is an infinitely small point that moves continuously from the past to the future. In his relentless quest for generalizations, Baudouin subsumes under a common heading grammatical categories that at first glance seem to be of a totally different order. Thus he sees in the formation of a new class of numerals, in the high frequency of zeroes, in the use of arithmetical or geometrical aspects (such as the semelfactive, iterative or punctual and durative), and in the use of argumentatives and diminutives, the reflection of increased quantitative or mathematical thinking in Polish (1922, 142ff.). These, and many other such novel ideas, are scattered in all of Baudouins mature writings.
Equally timely is Baudouins work in the field of linguistic typology. Beginning with his earliest lectures and programs, he raises the question of a scientific characterization of various groups of languages and dialects (1877/78 ). Under close scrutiny, he argues, all the genealogical schemes which have absorbed so much energy appear to be inadequate (1884 ). At the same time, he came to recognize the one-sided and antiquated character of the prevailing morphological classification of languages which was based on a single, external principle and according to which the inflected, synthetic languages (with their internal flexion) represented the apex of development (1876/ 77 ). It is important to remember that typological comparisons, which enjoyed some vogue at the beginning of the nineteenth century, had fallen into disrepute at the time of Baudouins activity, when linguistics was declared (e.g. by Paul) to be a strictly historical science and when Brugmann denied, on theoretical grounds, the value of such comparisons.43 Even Meillet, who expressed a vivid interest in typology, believed that it could not be done without a prior historical description of a maximal number of languages,44 a requirement that would have postponed the enterprise indefinitely. Baudouin, on the other hand, saw in the structural comparison of languages simply an extension of the internal analysis of a single language, because he was convinced that the two approaches were complementary aspects of the same inquiry into the general properties of language. A morphological or structural classification of languages, he wrote, is based upon the similarity of their actual states independent of their historical relations . . . and depends on universal conditions, and properties common to all people regardless of their origin (1889 (2) ). But such a classification, he stipulated, presupposes the existence of descriptive grammar which states the facts of language in their mutual relation at a given period of history (1909, 189). In his early lectures and programs he believed only in the fruitfulness of a morphological classification which would define first the psychological character of language, and then its external form as a reflection of internal activity (1877/78 ; 1871 ; 1884 ). But later he came to acknowledge that the current state of our science hardly permits such a characterization of the IndoEuropean languages or of the Slavic languages either. Such a characterization is, at any rate, far more difficult than a characterization of phonetic features and of their historical development (1884 ). The success of such a classification, he warned, would depend entirely on the selection of features of general significance, and not of disparate facts (1884 ). In his program of lectures for the year 1877/78 (published in 1881), he himself offered the first phonological typology of the Slavic languages which is based on two such features of general significance for the Slavic world, stress and quantity. The presence of one or both of these features enables Baudouin to single out five different Slavic areas: (1) Serbo-Croatian, which combines stress and quantity; (2) Slovenian, which admits the long/short opposition only under stress; (3) East Slavic and Bulgarian, which lack the opposition of quantity; (4) Czech and Slovak, which lack free stress; and (5) the other West Slavic languages, which lack a free stress (where the stress is of an anthropophonic nature). With some minor corrections, this classification is still valid today and is, in effect, the neatest classification that can be made for the modern Slavic languages. In comparing systems, Baudouin wrote, we should not compare individual sounds, but the various elements of sounds, their phonic components (1902 ). Thus he raised the question of the different utilization of the feature of nasality or stress in the various Slavic languages, the relation of stress to the quality of the vowels (1884 ), and the different character of the palatal č in Polish and in Russian. In traditional comparative grammars, he complained, such comparisons have [hitherto had] almost no place (1902 ). From a typological point of view he also examined the Slavic categories of gender (1929, 223ff.) and distancing (1912 (1)), and he offered a typological comparison of the Indo-European with the Turanian and Uralo-Altaic languages that makes reference at the same time to their phonological, morphological, and syntactic features (1876/77; 1903). Among the salient facts of the Turkic languages, he mentions the configurational function of stress and the morphologized character of vowelharmony, the progressive character of its alternations, the autonomy of syllables as morphological units, the nominal character of these languages, their lack of gender and of syntactic congruence. At the same time he emphasizes the contradictions in their word and sentence structure, i.e., the initial position of the dominant morpheme or the root within the word, and the final position of the dominant syntagm, the subject, within the sentence (1930 ).
In addition to comparing languages in terms of their phonological and morphological properties, Baudouin also recognized the need for classifications based on phenomena of linguistic diffusion showing the role of geographic contact and cultural influence in the formation of mixed languages (1902, 1909; 1930).
Baudouins approach to diachronic linguistics, like his work on typology, rests on the premise that any comparison of systems in time presupposes the description of their discrete, synchronic states. This approach, he declares, attempts to establish two actually given periods in the development of a language and to determine the direction of change of various categories of sounds and of the sound system as a whole (1884 ). Phonological change is, in Baudouins conception, not merely a replacement of sounds and of their inventories, but a rearrangement of . . . the articulatory-auditory combination of the phonemes (1910 ). In his early formulations, Baudouin identifies synchrony with statics, and diachrony with dynamics (1877/78 ); but the laws of the former, in his opinion, are connected with the laws of the latter (1870, ), since the beginnings of dynamic sound changes [take place] in the synchronic state of a language (1876/77 ). The possibility of change or of the dynamics of sounds is, in other words, conditioned by the stability or variability implicit in the sounds of language, i.e., by the internal stratification of the phonemic oppositions. Through all the fluctuations and deviations the linguistic facts and their causal relations exhibit remarkable uniformity and regularity . . . which characterize both the stability of the combinations of the articulatory and auditory elements and their fluctuations and changes (1910 ). The various sounds of a language, writes Baudouin, have a different potential for change, and they carry in their very nature the germ of their relative stability and variability (1889 (1), 187). The proper description of historical development must, like typological comparison, consider the two interconnected aspects of invariance and variation, and can be accomplished only by generalizing the entire line of historical development (1903, 35) through the study of linguistic tendencies that pervade the history of a particular language (1876/77 ).
Baudouins historical treatment of Polish phonology (1922,323ff.) is an illustration of such a description that focuses both on the permanence of a given phonological system and on its drift, or its movement along a more or less determined path. The direction of change, Baudouin reminds us, is not an exact formula of phonetic law, but rather a statistical constant (1910 ). Looking at the history of the Polish phonological system as a whole beginning with Indo-European until its modern dialectal differentiation, Baudouin underscores first of all the conservatism of some phonemes (or, rather, of some distinctive features and their combinations) that have remained intact throughout the history of the language. Second, he distinguishes in this history two main phases of development, one of which extends from Indo-European to Common Slavic, and the other from Common Slavic to modern Polish. Each of these stages is characterized by a different trend which lends meaning and unity to each of its particular changes. In the first stage there is a tendency to strengthen the vowel system at the expense of the consonants, whereas in the second stage the trend is reversed. The first tendency is manifested in the loss of Indo-European aspirates and palatals, in the elimination of final consonants and clusters, in the formation of nasal vowels, in the influence of vowels upon consonants, etc., whereas the second tendency leads to the loss of diphthongs and syllabic sonants, to the closing of the syllable, to the introduction of palatalized and palatal consonants, and to the influence of consonants on the distribution of the vowels. This interpretation of the phonological development of Polish is no doubt oversimplified and is open to criticism on several points. First, the Polish dialects do not conform to this dyadic scheme, for some of them belong rather to the vocalic type, having preserved (or transcoded) certain vocalic distinctions of Old Polish and having relinquished some of the consonant differenees of Common Polish (e.g., the loss of palatalized consonants and of the palatals ć, , ś, ź); second, it is questionable whether the paradigmatic and syntagmatic (distributional) changes of Polish (or of any other lagnuage) can be reduced to a single formula, as either of them may occur independently of the other. But the importance of Baudouins study lies in its insistence on the limits of the variability of sounds, i.e., on the constraints that are imposed on the range of change which account for the recurrent character of certain kinds of change within the history of a given language. This approach must not be confused with teleological explanation, which Baudouin specifically rejects: Between the starting and ending point of historical change . . . there is no relationship that could be interpreted as a law of evolution. On the contrary, the path of evolution taken by a series of generations presents an infinite number of discrete points, such that each successive stage depends directly on the conditions of the . . . articulatoryauditory representations of the preceding stage (1910 ). What Baudouin tries to emphasize is rather that comparison is not a goal in itself, and that the quest for general laws must guide the linguist both in vertical (i.e., historical) and horizontal (i.e., typological) comparisons of languages.
6. This introduction is not the place to discuss Baudouins significance in Slavic linguistics, which was the main field of his professional activities. I shall indicate here only the main areas of his interest and his major achievements. As is apparent from the foregoing, it is almost impossible to separate Baudouins contributions to Slavistics from his work in general linguistics, for the study of empirical facts was for him always a springboard for broad theoretical generalizations. In all of his general linguistic studies he draws on his vast knowledge of the Slavic languages and dialects, with which he had first-hand familiarity both as a student of Slavic texts and as a dialectologist. He was an extremely talented organizer of field work (as, for example, in Slovenia, where he secured the cooperation of scholars, teachers, farmers, and newspapers), and was an indefatigable collector of linguistic and ethnographic data (which include Serbo-Croatian texts from Southern Italy, Lithuanian folksongs, and a vast store of Slovenian material, most of which remains to be published). His main fields of interest, besides comparative Slavic, were Polish, Slovenian, and Russian. In each of these areas he left a large body of work dealing with the history, structure and/or dialectology of these languages. In Slavic historical linguistics his name is associated with the socalled third palatalization of velars, with the euphonic n in the oblique cases of personal pronouns, and with Lindens law (the treatment of initial wr in Slavic). His major comparative works deal with the relationship of Polish, Russian, and Church Slavonic, and with Slavic phonology and morphology.
Boudouins book on Old Polish Before the Fourteenth Century is still unsurpassed. On the basis of toponymie and anthroponymic material culled from Latin medieval texts, Baudouin succeeded in reconstructing the picture of Old Polish phonology and in defining its principal phonological development (the change of the palatalized consonants t d, s, z, r into the palatals ć, , ś, ź and ř). His Psychological Description of the Polish Language and his Outline of the History of the Polish Language not only give a compact and profound analysis of the synchronic and diachronic aspects of Polish, but are among his most mature works on general linguistics. The formerly much-disputed question of the position of Kashubian among the Slavic languages was solved once and for all by Baudouins penetrating analysis, which proved it to be a member of the Lekhitic group of the Slavic languages most closely related to Polish (in Baudouins catching phrase, it is plus polonais que Ie polonais même).
Baudouin was the first Slavic scholar who put the study of the Slovenian dialects on a scientific foundation. He was also the first to identify the fundamental character of the phonological development of Slovenian, which consisted in transforming the opposition between the long, accented and short, unaccented vowels into a maximally differentiated qualitative opposition. Slovenian linguistics is only now beginning to recognize the significance of Baudouins insights, after years of haggling over his obviously misguided theory about the Turanian origin of the Rezija dialects. Baudouin also emphasized (contrary to someauthoritative opinion of his day) the basic historical unity of Slovenian and Serbo-Croatian as opposed to Macedo-Bulgarian.
Baudouins contributions in Russian are scattered in all of his general linguistic works, as he believed that only live languages provide a solid basis for the verification of linguistic hypotheses. Among his specifically Russian works are his analysis of Russian orthography with relation to the sounds of Russian, his léxicographical additions and introduction to Dais dictionary, and his study on thief-slang (in the introduction to Trachtenbergs monograph). His lectures on Russian clarified for the first time (after the efforts of Böhtlingk, Grot, and Tulov) the basic principles of Russian phonology, and prepared the ground for an outstanding school of Russian phonologists.
Such, then, was the scope of this remarkable scholar and teacher.
The present translation is the first rendering into English of Baudouin de Courtenays works, which were written originally in Russian, Polish, German, and French. I have profited mostly from the Russian edition of these writings, which was published in Moscow in 1963 under the editorship of V. P. Grigorev and A. A. Leontev. I have consulted the original publications for ambiguous or difficult passages and have translated items VII, XII, XVII, and XXI directly from the German and Polish.
This selection of articles and books is intended to give a diversified, though fairly compact, picture of Baudouins representative works. The Statement of Linguistic Principles (VIII) has a purely historical value and is included here because it is well known and has been frequently cited. The first eighteen studies are entirely within the field of general linguistics, and the last four deal with Slavic topics. The material is otherwise presented in a chronological order.
In addition to the omissions in the Soviet edition (marked in this book by . . .), I have omitted a number of passages and phrases, most of them in the nature of introductory or concluding remarks, plus some metaphors, and the mathematical tables which would now appear obscure and have no bearing on the arguments (these omissions are marked by(. . . ). In a number of cases I have added (in brackets aa) explanatory notes and minor corrections.
Baudouins writings present a serious challenge to the translator. Succinct, epigrammatic, and almost classical formulations are embedded in a stream of redundant statements, piled-up metaphors, and poorly organized sentences which are bound (especially in their German form) to test the patience of any reader. This translator can only concur with Meillets verdict on Baudouins style: Sa manière de sexprimer était gauche, реи idiomatique; on dirait que jamais il ne se sert dune langue maternelle.45 Total fidelity to the original text would have been as much a disservice to the modern reader as to Baudouins basically lucid ideas. I have therefore chosen to be faithful first of all to Baudouins thought, without departing too drastically from his wording. On the other hand, I have tried to adhere to Baudouins linguistic terminology, although he himself was not always consistent. For the benefit of the English reader I have included a glossary of Polish and Russian words which he did not translate (in his German and French texts).
It is my pleasant duty to thank Professor Thomas A. Sebeok of Indiana University for his encouragement for me to undertake this onerous though rewarding task of bringing Baudouin to the English reader, and Mr. Bernard Perry of Indiana University Press for his support and patience with this endeavor. I also wish to thank my students, David Henderson and Terry Merz, for helping with the translation of certain chapters.
* For the references to Baudouin’s works see p. 48.
1. A. Meillet (1), Introduction a l’étude comparative des tangues indoeuropéennes (8th ed.), Paris, 1937, p. 475.
2. R. Jakobson, “De Saussure’s Unpublished Reflections on Phonemes” (quoted from manuscript by courtesy of the author; forthcoming in Cahiers Ferdinand de Saussure, 1971).
3. R. Godel, Les sources manuscrites du Cours de linguistique générale de F. de Saussure (henceforth quoted as SM), Paris, 1957, p. 2gff.
4. F. de Saussure, Cours de linguistique générale (4th ed.), Paris, 1949, pp. 153, 154.
5. SM, p. 51.
6. op. cit. (1), p. 478.
7. A. A. Leont’ev, “Boduèn de Kurtenè i francuzskaja lingvistika,” Izves.tija AN, Serija literatury i jazyka, 1966, 25, 4, pp. 329-32.
8. See F. M. Berezin, Ocerki po istorii jazykoznanija v Rossii (Konec XIX-nacalo XX vekov), Moscow, 1968, p. 259.
9. Cours, p. 172.
10. SM, p. 65.
11. A. Meillet (2), (Necrology), Revue des études slaves, 10, 1930, p. 175.
12. V. Bogorodickji, “Kazanskij period professorskoj dejatel’nosti I. A. Boduèna-de-Kurtenè,” Prace fdologiczne, 15, 2, 1931, p. 467.
13. op. cit. (2), p. 174.
14. H. Uîaszyn, “Jan Baudouin de Courtenay. Charakterystyka ogólna uczonego i cztowieka (1845-1929),” Biblioteczka Kola Slawistów imienia Baudouina de Courtenay, 1, Poznan, 1934, pp. 1-43.
15. R. Jakobson, Kazanska szkola polskiej lingwistyki i jej miejsce w éwiatowym rozwoju fonologii, Bulletin de la Société polonaise de linguistiquey 19, i960, p. 20.
16. W. von Humboldt (1), “Über das vergleichende Sprachstudium in Beziehung auf die verschiedenen Epochen der Sprachentwicklung,” Werke, 3, Schriften zur Sprachphilosophie (ed. A. Flitner, K. Giel), Darmstadt, 1969, p. 2.
17. idem (2), “Über die Verschiedenheit des menschlichen Sprachbaues und ihren Einfluss auf die geistige Entwicklung des Menschengeschlechts,” Werke, p. 473 (VII, 95).
18. Note that according to F. Schlegel “comparative grammar” was intended to reveal the “internal structure” of the related languages [“die innere Struktur der Sprachen oder die vergleichende Grammatik], “Über die Sprache und Weisheit der Indier, F. v. Schlegels sämmtliche Werke, 7, Vienna, 1846, p. 291.
19. H. Osthoff, K. Brugmann, Morphologische Untersuchungen auf dem Gebiete der indogermanischen Sprachen, Leipzig, 1878, p. XIII; see also K. Wechssler, Gibt es Lautgesetze?, Forschungen zur romanischen Philologie, 1900, p. 422ff.
20. H. Osthoff, Schriftsprache und Volksmundarten, Berlin, 1883, p. 17.
21. For further comments on this subject, see E. Cassirer, The Philosophy of Symbolic Forms, 1, Language (7th printing, New Haven, Conn., 1968, p. 170ff.
22. H. Paul, Prinzipien der Sprachgeschichte (4th ed.), Tübingen, 1920, p. 24.
23. op. eit., p. 37.
24. op. eit., p. IX.
25. Quoted from Cassirer, p. 173.
26. SM, p. 38.
27. op. cit., p. 15.
28. For an account of the controversy concerning the reality of dialects, see L. Gauchat, Gibt es Mundartgrenzen?, Archiv für das Studium der neueren Sprachen und Literaturen, 111, 1903, pp. 365-403.
29. Meillet (3), Linguistique historique linguistique genérale, Paris, 1 (1929), 1948, p. 107.
30. Fr. Techmer, Phonetik, Leipzig, 1880, p. 59.
31. op. cit., p. 20.
32. A. Camilli, Il sistema ascoliano di grafia fonética, Città di Castello, 1913, p. 22.
33. E. Sievers, Grundzüge der Lautphysiologie, Leipzig, 1876, p. 37.
34. ibidem, p. 42.
35. Cours, p. 221.
36. The following quotations are from the Polish translation of Kruszewskis works Wybór pism (translated by J. Kurylowocz, K. Pomorska), WroclawWarsaw-Cracow, 1967.
37. Cours, p. 218.
38. R. Jakobson, M. Halle, Fundamentals of Language, The Hague, 1956, p. 8.
39. op. cit. (2), p. 459 (VII, 82).
40. A. H. Sayce, Introduction to the Science of Language (3rd ed.), London, 1890, p. XVI. There also the reference to Fick.
41. В. Delbrück, Einleitung in das Sprachstudium, 1880, p. 75. Cf. the references to Delbrück in Baudouin, 1909, p. 258ff. and in Kruszewski, p. 50.
42. op. cit. (3), p. 130ff.
43. K. Brugmann, Zum heutigen Stand der Sprachwissenschaft, Strassburg, 1885, p. 11 ff.
44. La linguistique générale quon obtient en faisant abstraction de lhistoire est encore une science реи faite, difficile à faire, et qui pour se faire suppose du reste quon ait déjà décrit aussi complètement que possible 1histoire du plus grand nombre de langues possible (3), p. 101.
45. A. Meillet (2), p. 174.
WORKS BY BAUDOUIN DE COURTENAY CITED IN THE INTRODUCTION
For the full titles of Baudouin de Courtenays works see the Table of Contents and the Bibliography.
Page numbers referring to the English translation are given in square brackets; otherwise they refer to the Soviet edition, with the exception of the following, which come from the original publications: 1881 (2), 1888, 1909 (2), 1912 (2), 1915, 1922 (Ch. II).
|1871||General Remarks. . .|
|1875||Opyt Fonetyki. . .|
|1875/76||Program of Readings. . .|
|1876/77||Program of Lectures. . .|
|1877/78||Program of Lectures. . .|
|1881 (1)||Nekotorye Otdely. ..|
|1881 (2)||Neskolko slov 0 sravnitelnoj grammatike indoevropejskix jazykov, zmnp, 213.|
|1884||The Slavic Linguistic World. . .|
|1888||Zametka к nekrologu N. V. Krusevskogo, RFV, 20, 297-302.|
|1889 (1)||Mikotaj Kruszewski|
|1889 (2)||On the Tasks of Linguistics|
|1895||An Attempt at a Theory. . .|
|1897||Statement. . .|
|1901||On the Mixed Character. . .|
|1903||Lingvističeskie Zametki. . .|
|1904||Linguistics of the Nineteenth Century|
|1908||О związku wyobrażeń. . .|
|1909 (1)||Zametki na poljax soč. V. V. Radlova|
|1909 (2)||Zarys historii jezykoznawstwa...|
|1912 (1)||Ob otnosenii russkogo pisma. . .|
|1912 (2)||Sbornik zadač. . . |
|1915||Charakterystyka Psychologiczna. . .|
|1917||Vvedenie v jazykovedenie|
|1922||Zarys historii jɛzyka polskiego|
|1927||The Difference between Phonetics. . .|
|1929 (1)||The Influence of Language on World-View.|
|1929 (2)||Facultative Sounds. . .|
|1930||Problems of Linguistic Affinity.. .|