[438]  
        Soviet proletarian-linguistics will always have a certain fascination for both expert and layman in the field of European linguistic theory. If, that is, we understand by Russian proletarian linguistics the new phase of language-study opened up as an analogy to the scheme of socialist proletarian science as a whole. It will prove interesting for its relationship both to Marxist-Leninist dialectical materialism and to modern linguistic trends.
        In this paper I wish to express my opinion on modern Soviet linguistics, and to show how this very field, in so far as it contains anything typically Soviet without parallel elsewhere in Europe, is bound up with the theses and theories propounded by Nikolai Jakovlevič Marr.
        Marr's linguistic teaching is now regarded throughout Russia as the Soviet theory, and its representatives are the linguistic theorists, since the rest are specialists, either in Russian or in non-Russian fields, i.e., people who do not dare to aspire to broader linguistic conceptions, or, if they do, follow in Marr's footsteps. This they do even if at times they fail to understand Marr, or do not want to do so.
        Marr, then, is so utterly the representative of Russian linguistic theory by reason of his doctrine that we must give him our first attention, and thus discover the key to Russia's present attitude to linguistics. Without Marr, we cannot appreciate subsequent generations of linguists.
        Marr's teaching, in its latest phase at least, is a sharp reaction against all traditional theory, against the "bourgeois" tradition of pre-revolutionary days as carried on unbrokenly elsewhere on the foundations of neogrammarian doctrine — unless we except the recent structuralist formula of the Prague-Copenhagen schools. This is so even when Marr's theories originate in, and build upon, neogrammarian principles, as I shall show.
        In his attitude to language development he attempts a new solution of fundamental neogrammarian problems as a whole, and by a radical redirection of his own linguistic aims he shows, or attempts to show, that this clearly arises out of similar or analogous basic reactions as new linguistic trends, whether they be the sociological trend of the Meillet-De Saussure type or the idealistic
[ 439]  
linguistics of Vossler, quite apart from the peculiar attitude adopted by Bloomfield and Sapir based on materialistic monism and standing close to the Soviet notion of linguistic processes in its theoretical basis.[1] The attempt to subject to criticism the results of the positivist and naturalistic trends of the neogrammarians and to ground itself on philosophical premises is the thing that links Soviet linguistics with all these trends in its interpretation of linguistic processes and the way it sets out linguistic facts. Recent Russian linguistics has been developing within the frontiers set by the philosophy of dialectic materialism and the Marxist-Leninist conception of language and linguistic history. This runs counter to the synchronistic school which stresses the logic of grammatical structure, like the school of De Saussure and all analogous trends[2] that disregard the parallel aspect of language as a system of differentiated and coordinated signals based on Bloomfield's theory of materialistic monism.[3]
        The theoretical face of Soviet linguistics changed radically after the Revolution, and its new programme was formulated most eloquently of all by N. J. Marr[4] in my opinion when, referring to the setting up of the socialist State, he emphasised on behalf of Soviet linguistics as a whole that the Regime stood for cultural autonomy and the free evolution of all Soviet nationalities and their languages as embodied in Soviet government policy. This takes into the field of specialist linguistic study all the languages of the Soviet Union, languages which were studied under the Tsars rather for their folklore interest than as the substance of linguistic interpretation, e.g., the Caucasian and the Palæo-asianic languages. This is what the new Soviet linguistics regards as its aim both in theory and plan, as this, like other scientific disciplines, bases itself on the single philosophic ground-plan of Marx-Leninist dialectic materialism.
        In this way Soviet linguistics has placed itself on an entirely different theoretical plane from anything previously known. The motley linguistic pattern of the Soviet Union is tending more than in other countries to encourage the comparative study of the most heterogeneous language-types and structures. The narcissistic concentration on the significance of one language-type as fostered in non-Soviet European linguistics is steadily falling out of favour, and is being replaced by the synchronistic study of the motley Soviet pattern of languages. The study of Russian and other Indo-European languages is losing its hold, whereas the Mongolian, Palæo-asianic, Caucasian and other groups are coming into their own.
[ 440]  
        Along with problems related to the framing of spelling systems for languages without literary traditions, as well as practical problems of other kinds, linguistic phenomena of special importance are being studied, as, for example, the passive structure of the verb, etc.[5]
        The year 1931 witnessed a debate on the intentions and the tasks facing Marxist linguistics — Marr's theory of relationship between language and society ; language and thought; the uniformity of language-formative processes as evidenced in Marr's theory of stadiality in language evolution ; yet it cannot be said that this debate cleared up the problems fundamental to Soviet linguistics, as its political colouring was discredited. To some, e.g., Danilov, Marr was not Left enough nor drastic enough; to others, e.g., Polivanov and even Bubrikhov, the objection was that Marr did not keep strictly to linguistic problems.
        It is true to say that the polemical attacks served to popularise Marr's theories, but failed to make Marr's doctrine more easily understood, especially outside the Soviet Union. Even when Marr finally succeeded in refuting all the charges made against him, whether from Right or Left, his teaching did not become the official doctrine of Soviet linguistics as has often been assumed. Nor is it true to say, as Europe had said before the Second World War, and as it may still be assumed elsewhere, that the last word had been said on Marxist linguistics in the true sense of the word when the said debate ended. The leading Soviet linguists like Meshchaninov share the credit, too. The latter declare Marr to be their preceptor even if they frequently interpret Marr in their own way. But the younger generations down to the youngest share in it, too. To them the bold picture of language evolution in its bearings on dialectic materialism and the growth of society is impressive by reason of its sheer boldness of perspective and its amazing breadth of vision.
        Together with this new trend of modern Russian linguistics headed by Marr there emerges a sociological plan based on class stratification. "Japhetidology" seeks a solution of the problem of linguistic history in relation to this. There is nothing typically Russian in this idea, unless it is the element of class antagonism implied in it, since the sociological interpretation of the essence of language merges with the theoretical complex of modern linguistics, and unlike neogrammarian and idealistic doctrine, sociology is the essence of the new linguistics, which is subjected to far more searching tests by theory and philosophy than ever before.[6]
        If non-Russian linguistics co-operates rather more in theory
[ 441]  
than in fact with sociology to explain phenomena, though a relationship between the community and its institutions and the hierarchical linguistic pattern — collateral as well as historical — has been common knowledge for a long time, it would seem that Marr[7] was the first to emphasise the sociological aspect on definite programmatic lines, basing it on historical materialism in Russia. The evolution of Russian linguistics on purely revolutionary lines is likewise immanent in the Japhetic Doctrine,[8] which in the hands of Meshchaninov after Marr's death in 1934 became the "new" language doctrine reflecting Marr's latest doctrinal phase known as stadiality, a theory which Meshchaninov conceived on different lines from Marr.[9]
       Meanwhile the Japhetic Doctrine stands in much the same relationship to the theoretical premises of historical materialism as, say, Meillet's conception of language (especially in his Linguistique historique et linguistique générale) regarded diachronistically stands to Durkheim's broad sociological theses in his Règles de la méthode historique.
       Among parallel linguistic trends the Japhetic Doctrine is outstanding by its departure from the general beliefs of the neogrammarians, and, in part, from those of Meillet and De Saussure, by its greater emphasis on speculation in face of linguistic facts, which it not merely describes and classifies but also tries to explain. The procedure of Japhetic Theory depends in a large measure on the application of the results of speculation to concrete linguistic material, e.g., in the case of structuralism. But herein lies a certain danger of dogmatism, the shadier side of which stands out characteristically in several of Marr's interpretations.
        Marr's Japhetic Theory, then, is based on the premises of Marx-Leninism in its interpretation of the historical evolution of society. In its daring application to the historical evolution of language, monistic and diachronistic, it seeks a way out of the crisis into which historical linguistics has landed by dealing rather with problems of language formation than with sets of laws applicable to semantic change as is done in contemporary European linguistics, and this is an aspect which traditional linguistics, even when grounded in the old comparative method, has always cautiously, and wisely, avoided.
        Nor is this the only point. In the personal interpretation of historical materialism as applied to linguistic evolution Marr's teaching has aroused lively controversy in Russia at times.[10] It was clear that the identical theoretical premises of historical materialism were capable of various interpretations as applied to concrete pro-
[ 442 ] 
blems of living speech. This is where the "new" science of language emerges, I believe, in its attempt to get closer to the official Marxist doctrine. I have already dealt with this and the manner in which it was worked out by Meshchaninov, who used Marr's adaptation of Marxist doctrine on stadial evolution in class stratification as his point of departure, and tried to evolve a theory of language of his own, avoiding Marr's language typology, and replacing Japhetic historical interpretation by a general grammatical system — not uninfluenced by contemporary European linguistics and the synchronistic approach—as applied to the problems of syntax and verbal expression in the sentence.[11]
       In this way Meshchaninov links the activity of contemporary European linguistics with the partial isolation of Russian linguistics, as is seen in the recent formulation by J. Vendryes,[12] and the broad character of outstanding linguistic works by well-known linguists like N. S. Trubetskoy, R. Jakobson, Brondal, Hjelmslev, etc., though each of these have an individuality of their own.
        As to the monistic perspective and its bearing on linguistic evolution, Marr sets out by broadening the field of Japhetic research to include mainly non-Indo-European languages, all of which he attempts to co-ordinate under one interpretative formula. It should not, however, be forgotten that by his Japhetic theory he was a pioneer in constructing the semantic edifices of the Caucasian lexical systems on a basis of mere theory, and it was with these languages that Marr began, being both a native Georgian and an outstanding expert on the Kharthveli (Gruzinian) group of languages and others within the same area. Neither should we forget how he resolved these languages in terms of the Semitic, Hamitic and Basque systems, at the very outset of his linguistic career.
        Starting from neogrammarian ideology and progressing to linguistic history, Marr attempts to establish their relationship to the ancient languages of Asia Minor, Sumeria, Akkadia, Lydia. and the language of the Vannic cuneiform inscriptions. From now on he builds upon a Caucasian theory based on linguistic substrata, and, following his thesis about the Semitic typology of Caucasian-type languages, he formulates a theory that these languages are in genetic relationship to the substrate languages of the Mediterranean,[13] including among the latter such well-known languages as Etruscan and modern Basque. In its broad outlines this theory represents nothing new ; it was put forward by European linguists before Marr's day, and is almost universally recognised by linguists, including the Italian Trombetti, who is Marr's equal in breadth of
[ 443 ] 
vision and interest, and the representatives of traditional linguistic movements, particularly in so far as the theory applies to the striking parallels, both lexical and toponymical, to be found in these languages. Critical voices still warn against this so-called " Caucasian " method of linguistic study.
        This substratic group of languages, made up of Caucasian, Etruscan, Basque and a whole series of unknown languages, is called by Marr "Japhetic" in contra-distinction to the "Prometheid" languages, by which he means Indo-European. It would be interesting at this point to ascertain the connections of the Mediterranean Japhetic substrate as presumed by Marr with the so-called Alarodian substrate formulated in outline by Oshtir. We should recognise how Marr and Oshtir both reacted similarly at this period to the incentives arising out of the crisis in Indo-European studies and their relation to the substratic Mediterranean problem. It would be of further interest to compare how far Marr and Oshtir resemble each other in the general formulation of their respective theses, and how far they differ. I think that, unlike Marr, Oshtir does at least pay lip-service to the laws of phonology, though his adherence is at times very arbitrary, and that he does not deliberately brush aside — as Marr does — the theoretical results achieved by the traditional Indo-European linguists.
        It was not until the declining phase of his scientific career, about 1925-1926, that Marr reassessed his Japhetic doctrine in the light of historic materialism. It was a period, as I have already emphasised, when theoretical basic doctrine was undergoing revaluation on the basis of dialectical materialism. The Japhetic languages were to him, and in defiance of the traditional view, not an independent linguistic group in "substratic" relation — if we may use the term — with the existing languages of Europe, i.e., the Prometheid languages as Marr calls them, but rather a group of languages at a particular stage of development, i.e., the Japhetic stage, and Marr regards it as the immediate task of Japhetic studies to find traces of this stage of evolution in the Prometheid (Indo-European) languages. Marr's new theoretical attitude arises as a matter of course out of his earlier views on the Semitic lexical stratum present in several Caucasian languages. In the latest phase of Marr's highly chequered linguistic career he reformed himself somewhat rapidly, but only, as I have emphasised, towards the end of his life, as he was doubtless under the pressure of Marxist-Leninist sociology, then under reassessment in an all-round attempt to reduce new Soviet science to a uniform Marxist level, i.e., evolution by stages, and
[ 444 ] 
the crystallisation of the class structure of society. According to this new hypothesis, the Japhetic languages, or rather the languages in the Japhetic stage of evolution, are the key to the understanding of the lexical structure of the languages of Europe, their semantic connections being characteristic of language systems of whole areas, not even excluding those of the Far East, America and Africa.
        To state the point concisely, the theory is one which is by its very nature a glottogonic (language-evolutionary) theory, resting on premises which have not been explored sufficiently, nor interpreted properly. This, in spite of the number of existing possibilities of interpretation.
        Hence Marr sets out by assuming that speech, i.e., the so-called phonic stage of language, is a derivative of gesture and mimicry. This seems to imply a language of arbitrary signs used semantically as a means of communication and understanding. This stage, which is alleged to evolve out of amorphic phenomena into agglutinating or inflected forms is called by Marr the Japhetic stage. The content of speech thus broadened out into a pre-logical phase. This phase then falls under the domination of certain laws of semantic change towards the formation of a lexical system in close combination with social phenomena, i.e., the evolution of a language system. This is the transitional phase through which every language must go before reaching the present stage represented by languages contemporary with Indo-European, whose structure is regarded by Marr as the final, and highest, stage ever achieved by language. On the face of it the thesis seems to come somewhat dangerously near to German theories of comparative linguistics, which presumed a sort of superiority in Indo-European languages and in the Indo-European ethnic whole. Though it is not the aim of the present study to determine how far the gesture-and-mimic theory as a primary factor in language evolution was derived by Marr from the teachings of positivist neogrammarians and psychologists (as, for example, Wundt, whom he seems to have known), and though it is not our purpose to ascertain connections between prelogical linguistic phases and the views of contemporary French linguists (as Levy-Bruhl), it should nevertheless be pointed out that Marr may be accepted and interpreted as being heir to the neogrammarian theory in this sense. It should be remembered that Marr's personality was being formed in the heyday of neogrammarianism, and that Marr took his scientific beginnings from the neogrammarians, proceeding thence to linguistic dynamism, though he differed from his preceptors in many respects towards the end of
[ 445 ] 
his career. Language evolution is to him, as to the neogrammarians, the specific and crucial problem of language research. That this view was alien to emergent Soviet linguistics is clear from the fact that it became the focus of debate on the tasks and purpose of Marxist linguistics. It has been pointed out that the purpose of Russian linguistics is not necessarily to be sought in the study of the prehistoric period of language, as Marr claims, but that Soviet linguistics should rather take stock of present-day language problems in all their scope and breadth, to discover how it bears on specific practical considerations.
        If it was the neogrammarians that came forward with their theory of divergency from a primitive language core, a theory that is generally held for Indo-European and to a lesser degree for Uralian, Semitic, etc., it was Marr who came forward with his notion of primeval variety which is continually shrinking. After observing that small linguistic units tend to disappear after a lapse of time, whereas larger linguistic units tend to come into being, he emphasised the point that linguistic evolution gravitates towards the formation of one language as an instrument of understanding in a classless society. This is a form of evolutionary convergence. It should be remembered that Meillet himself had already come forward with a convergent theory of his own, and that N. S. Trubetskoy and R. O. Jakobson, with their formulation of the convergence of languages within a given unifying linguistic framework, are also parallel with Marr. This means that Marr is by no means alone in his hypotheses as a casual observer might imagine. He is at all times part and parcel of European linguistics in the way he develops. All he does is to solve the problems of contemporary linguistic science by over-elaborating the existing techniques, and it is in this that Western European linguistics fails to understand him, or understands him imperfectly. A few more general remarks should be added to Marr's fundamental glottogonic theory.
        It is not clear, for instance, on what basis Marr formulates his hypothesis, i.e., that gesture language is older than a structural system of articulate sounds.[14] The probability is stronger if we accept the hypothesis of parallel evolution, if, that is, we are determined to grapple with this problem of language creation without necessarily settling it once and for all.[15] The apparent fact that gesture is resorted to in certain circumstances during cult practices as aids to understanding in Caucasia can in no way justify the dogmatic claim of greater age for this form of speech.
        In devising his theory of language evolution Marr markedly
[ 446 ]
deviates from tradition. He does not understand languages as related systems even in terms of the traditional theory of genetic kinship and divergent evolution from a primitive core, a theory held by linguistics ever since the days of Bopp and Rask, nor does he regard them in the achronistic light of many modern linguists who, by taking a synchronistic cross-section of a language group, aim at the correlation of secondary tendencies therein. Rather does he study the relationships of language systems as wholes in monistic perspective as Hegelian elements of dialectical materialism applied to the various stages of human evolution.[16] Here Marr concentrates to an unusual degree on the possibilities of "mixed" languages. He was led to such interpretations by the linguistic pattern of Caucasia, where the languages appear to be much more the result of compromise within the group than elsewhere. This is due to the varied language pattern within a small area where multiple convergence is in evidence.
        By breaking down the rigidity of the problem-complex in dealing with genetic relationships on traditional lines and substituting free association on a horizontal plane with neighbouring language systems and a historical perspective linked with social hierarchy, Marr opened up a vista, uncritically in the eyes of other linguists, which was capable of the most multifarious interpretation, though one in which his lack of scientific training was strikingly obvious.
        The logical consequence of his theorising about the effect of pre-logical phenomena on semantic change and the evolution of lexical systems is also evident. Nobody is likely to deny that there is something in the idea, but the problem will always be how to determine such changes. It will be difficult to devise general principles for explaining semantic changes and transitions in primitive languages. It is no use proceeding dogmatically, as Marr does, from a priori theories, even if, as may be admitted, we do know something about the semantic structure of primitive languages ; no use, at any rate, if the hypothesis is assumed that, for example, the name of a national unit can be linked with that of some totemistic animal. According to his dogma the two are identified, citing other a priori remarks of his in other studies, and quoting even the name of a god as being that of some eponymic hero.
        Equally unintelligible and inadequately founded in fact by normal linguistic methods are the semantic mutation-series linking, for example, the name for "cereal" not merely with the name for "bread", but also with the term for "tree" and " acorn." Similarly the term for "horse" is linked with names for "water-
[ 447 ]
 course." Yet the fact is overlooked that terms for watercourses may contain names of animal demons or totems even in European place-names.
        Marr often puts forward an etymological idea — highly intuitively, it would seem — which turns up later in Western European linguistics in a somewhat different guise. Thus, some time ago[17] he connected Georgian tha-v (tha-m, "head") as a de-sibillated form of an original sa-m with Russian samŭ, etc. Van Ginneken also[18] explains Slavonic samŭ, together with Gk. ’autos, Alb. vetë and Rum. îns(u) from Georgian thavi, "head."
        All this does not mean that I am seeking to replace the seriously-thought-out comparative work of most European linguists by Marr's etymological intuitions. In spite of the improbability of some of his etymologisings, however, and disregarding their frequently amateurish slickness, the basic principle of many of Marr's hypotheses cannot be denied, at any rate so long as they are based on his own linguistic material. Thus it is highly probable, though the idea is not Marr's own, that in the problem-complex of semantic change in linguistic evolution the pressure of cult and pre-logical thought played, or may have played, a part. The problem, however, remains unsettled, and it is by no means certain how far he was able to prove his point from concrete language material, nor what he was able to extract from it in order to facilitate the evaluation of linguistic facts. In other words : how far has a theoretically and logically sound argument been made out for the linguistic material under review ?
        It cannot be claimed, and this justifies Marr though the fact is well known, that language changes of any kind, including semantics which is the exclusive field in which Marr worked, have taken place according to some theoretical predevised scheme, or indeed according to any general trends or laws. There is always something to upset hard-and-fast laws of meaning, and this is characteristic of any evolution of spiritual values as distinct from the law-like rigour of natural phenomena.
        Marr sees in these deviations from common and often logically rationalised tendencies the residue of a pre-logical phase in language evolution. It is problematical, of course, how far he is right in this. He sees the confirmation of his hypotheses, and to prove his point within the framework of unrelated languages he is forced to frame a further Japhetic classification and a linguistic typology, This is surprising in view of Marr's repeated emphasis on the sociological aspect of language, and on a class hierarchy in its mechanical
[ 448 ]
formalistic evolution by a combination of morphemes (sound-patterns) and semantemes (sense-patterns). He contrasts spirant with sibilant, combining the former with e-, o- and a- variants and setting up etymologies in such a way that he has an unending fund of linguistic material with which to justify his hypotheses. Thus he frequently fails to study the semantic stratification of the lexical plan in its relation to the bizarreness of his word-types such as his own classification renders them.
        As I see it, this point would be made clear by a comparison with the semantic structure of the lexical system in a genuinely "primitive" language such as that of the Australian aborigines or certain of the Indonesian languages,[19] provided we admit of the possibility of comparison of such heterogeneous groupings. The semantic system of lexical co-ordination reveals a linguistic structure which is clearly foreign to the logic of present-day language, or, to use Marr's own terms, the contemporary stage of evolution, but it is sharply distinct from the hierarchy of semantic differentiation for the pre-logical Japhetic stage, as Marr dogmatically states it, so that in the end any common problem features are lost. Like most of Marr's theories, they bear on historical folklore, or, for that matter, anything but linguistics.
        Marr may have been aware of the structural complexity of linguistic phenomena when facing this problem-complex. He nevertheless failed to include this in his Japhetology in all the inclusiveness of its later phase. This may have been because of a defective comprehension of language systems.
        Though Marr may have been right here and there, his theoretical attitude to the problem-complex of semantic change in the older stages of evolution fails, it seems, in his unfortunate choice of material — in so far as one can call the material ill-chosen, for it is very hard to justify most of Marr's hypotheses by linguistic facts of a range required by the sweeping terms of his theories. All objections to Marr's Japhetology do in fact come from this direction, whether they are the casual comments of Meillet in Bulletin de la société de linguistique, or those of Hermann, H. Skold and others.[20] This is quite apart from the discussion aroused in Russia by Marr's attitude to historical materialism as expressed in his New Doctrine of Language.
       At the same time a very important point, underlined even by some contemporary Russian linguists, is that Marr avoids problems of morphological and phonetic relationship. In his Japhetology he has built almost exclusively upon relationships arising out of
[ 449 ] 
semantic analogy, but has failed to settle the question of semantic movement by any revision of his attitude. It is dangerous to found a new theory of linguistic evolution and assume a peculiar brand of kinship which might almost be called genetic purely on semantic analogy, since the greater part of lexical vocabulary is a civilised superstructure capable of cutting across genetic kinship. The validity of Marr's Japhetic theory should be ascertained in its bearing on the problems of kinship arising out of analogy and morphology as being one of the dominant criteria of kinship, in synchronic and diachronic perspective, whether kinship implies divergency from a primal core, or convergency towards union. Phenomena, in other words, which in their synchronistic guise cannot always be easily differentiated.[21]
       Marr's Japhetology may be summarised as follows :
        It is a trend which calls for a broad treatment of linguistic questions of kinship, the re-examination of old theories, and the bold statement of new ideas. This is to be done within the broad framework of linguistic and non-linguistic fact based on a uniform philosophy — in the Russian case Marxist-Leninist dialectical materialism. This holds good, of course, so long as we pursue the problem in terms of linguistic diachronism and regard it as a crucial problem of Marr's doctrine and of Soviet linguistics as a whole.
        This does not, of course, mean that we could exhaust all the problems of general linguistics solved by Marr. Yet I do not consider that questions of kinship between language and society, or of language and thought, are sufficiently typical of Marr's doctrine in all its originality. By this I do not mean that the solution proposed by Marr is devoid of interest. It is, however, closely bound up with the theories of dialectical materialism and in his hypothesis of uniformity in the process of language formation we do not discover in these relationships any solution that deviates from results obtained by modern European linguistics. Even so we find some interesting departures worthy of special study. Marr and his new language doctrine deserve further study in this connection for the set of problems brought forward by him; for surely a modicum of his fanaticism and dogmatism might be retrieved for European linguistics, even if I am convinced, and say so in so many words, that he overstated the demands of the historical point of view and the historical interpretation in terms of materialism, and thereby entered a field that was alien to his own linguistic problems.
        In a bold scheme he outlined rather than elaborated a brilliant fresco of historical linguistic evolution and of human society, doing
[ 450 ]
so with such zest that a great deal of fanaticism and dogma entered into his interpretations which were sociological and materialistic in trend, and this reacted to the detriment of the whole doctrine.

         Václav Polák (Prague).

         English version by stuart E. mann.