Thüringer, einer selbständigen mitteldeutschen Mundart, die manche Berührungen mit den benachbarten hessischen und fränkischen Dialekten aufweist,

zweitens der Sprache der Zuwanderer aus dem Westen, namentlich der heutigen Rheinprovinz und den angrenzenden Teilen der Niederlande. Die Einwanderung aus diesen Gebie

ten, für die ich der Kürze halber nur auf Bremers Ethnographie der german. Stämme §§ 185-191 (= Pauls Grundriß III2 S. 895-901) verweise, macht sich namentlich seit der Mitte des zwölften Jahrhunderts geltend und hält mit kaum verminderter Kraft über die zweite Hälfte des vierzehnten Jahrhunderts hinaus an. Wie die Einwanderer das wirtschaftliche Leben und den Charakter der Bevölkerung im mittleren und östlichen Deutschland-vorwiegend auf ehemals slavischem Bodenbeeinflußten, so tragen sowohl die mitteldeutschen und ostdeutschen Mundarten wie die neuhochdeutsche Schriftsprache -die ja in vielen Fällen zum Flämischen und Holländischen näher stimmt als zum Mittelhochdeutschen-die Spuren ihrer Sprache.

Erinnern wir uns, daß das Wort für 'Wand' im Mittelniederländischen und heutigen Flämischen weeg (weech) lautet, so werden wir nicht zögern, unser weg in der Erfurter Sprache des 13. Jahrh. auf Rechnung der niederfränkischen Kolonisten in Thüringen zu setzen.

Vermutlich ist unsere Urkunde nicht das einzige mitteldeutsche Schriftstück, in welchem damals das Wort weg als Ausdruck für 'Wand' gebraucht ist. Es lohnt sich wohl, nach weiteren Spuren desselben Umschau zu halten. Einstweilen wird der hier beigebrachte Beleg des alten Wortes, das man z. B. in den beiden mhd. Wörterbüchern von Lexer vergebens sucht, willkommen sein. Man findet es jetzt, in Grimms Wörterbuch s. v. Weeg, jedoch ohne die hier besprochene Stelle. JOHNS HOPKINS UNIVERSITY

THE HYPOTHESIS OF A PRE-GERMANIC

SUBSTRATUM

By E. PROKOSCH

“Mehrere, wenn nicht alle der zahlreichen Völker und Volksstämme, die in der Morgendämmerung der geschichtlichen Zeit uns im indogermanischen Europa begegnen, sind mehr oder weniger wahrscheinlich aus einer Mischung von Indogermanen und anderssprachigen Autochthonen hervorgegangen. Bei diesem ethnischen Verschmelzungsprozeß haben die indogermanischen Volkselemente den Sieg davongetragen: Die Ureinwohner wurden indogermanisiert, doch so, daß ein Teil der unterscheidenden Merkmale ihrer älteren Sprache auf die neue indogermanische übertragen wurde." T. E. Karsten, Mém. soc. néo-phil. de Helsingfors, VII, 321.

I. SUBSTRATUM AND PATTERN. Karsten's article "Zur Erklärung der germanischen Lautverschiebung," from which this passage is quoted, is the latest re-statement of the hypothesis of a pre-Germanic substratum through which the 'drift' of the Indo-European language was deflected, the result being the development of the Germanic language. The present article is not in any way a reaction to Karsten's theory. It was submitted (in synopsis) to the editors of the GERMANIC Review early in 1925, months before Karsten's article appeared in print. In fact, the quoted statement formulates the fundamental idea of the substratum question more or less in the same terms that I have used on various occasions. So I am far from finding fault with Karsten's clear and concise statement of the principle. After all, the thought as such is rather obvious. It has been cropping up for generations, seductive but elusive, and has often been toyed with in such a careless way that it is quite apt to be dismissed with a shrug by some of the leaders of linguistic science; Karl Brugmann, for instance, used to classify it with what he termed 'glottogonische Spekulationen.'

1 Thus in a paper read before the Anthropological Society of the University of Pennsylvania in autumn 1924, where I characterized Slavic as "Indo-European acquired and modified by a Finnish-speaking population of Northeastern Europe" and suggested analogous definitions for some other Indo-European languages.

As an a priori statement, the theory seems acceptable: A conquering nation mingles with the subject nation to the extent of ethnic amalgamation, but preserves its language; or, immigrants acquire the language of their new home; in either case it would seem that traces of the abandoned language could not fail to be incorporated in the dominant one, and linguistic literature teems with attempts to analyse the varying results of the process. Nevertheless the application of the theory to living languages has proven discouraging. Take modern English or French: In theory, the Celtic substratum appears inescapable, but there is hardly a single fact of importance in these languages that can with any degree of certainty be claimed as Celtic. Where are the concrete facts of an assumed Celtic or Slavic substratum in southern or eastern German? Of an Etruscan layer in Latin or Italian, of an Iberian basis in Spanish? I do not deny the existence of such facts, but their investigation has, with a few brilliant exceptions,2 been a curiously haphazard one, abundant in clever suggestions, but devoid of methodical principles.

Methodically, a valuable contribution to the problem is Sapir's assumption (or discovery, I should prefer to say) of linguistic 'patterns." "Every language is characterized as much by its ideal system of sounds and by the underlying phonetic pattern (system, one might term it, of phonetic atoms) as by a definite grammatical structure" (p. 58). . . . “One of the most curious facts that linguistics has to note is the occurrence of striking phonetic parallels in totally unrelated languages of a geographically restricted area” (p. 211). . . . "However we envisage the process in detail, we cannot avoid the inference that there is a tendency for speech sounds or certain distinctive manners of articulation to spread over a continuous area in somewhat the same way that elements of culture ray out from a geographical center” (p. 213). . . . In any given language, then, there exists an ideal conception of a relatively permanent sound system; an individual sound may

2 Schuchardt, Slavo-Deutsches und Slavo-Italienisches; Bröndal, Laan og Substrater.

3 Language, New York, 1921.

4

change, e.g., t becomes p in Germanic, but its place is filled again by d becoming t. The system "may shrink or expand or change its functional complexion, but the rate of change is infinitely less rapid than that of the sounds as such" (p. 58). As an illustration, let us apply the pattern theory to a particularly clear case: The Sanscrit consonant system has been preserved relatively intact in the modern Hindoo languages; it differs markedly from the consonant systems of all European languages, but is in good accord with those of the ThibetanBurmese group; at the same time, it is virtually identical with the 'reconstructed' consonant system of Indo-European as tabulated in our comparative grammars. The pattern theory, as a working hypothesis, leads unavoidably to one of the following conclusions: Either, Sanscrit has preserved the Indo-European pattern of consonants, while all other IndoEuropean languages have given it up. This would, from the point of view of the consonant pattern, group Indo-European with the Thibetan-Burmese family, postulating its Asiatic origin. In that case, the consonants of all Indo-European languages with the exception of Sanscrit must have been modified by the superimposition of various patterns of prehistoric languages of Europe or Asia Minor. Or, the Indo-European consonant pattern differed radically from that of Sanscrit. It may, or may not, have found its continuation in one or several of the other languages—say, in the Slavic, or Germanic, or Romance groups. If so, Sanscrit has adopted the pattern of northern India or a neighboring territory, and our reconstructed consonant tables of Indo-European are fundamentally wrong.

I am convinced that a consistent application of the pattern hypothesis to morphology as well as to phonology will bring us close to an understanding of this irritating substratum question. The present article, however, does not lay claim to any constructive result in this direction. It merely tries to examine systematically the numerous efforts that have been made to prove a non-Indo-European substratum as the basis of the distinguishing features of the Germanic languages. It is a surprising fact that more claims of that sort have been made 'Prokosch, MPh., XVI, 103 ff.; XVIII, 52.

for the Germanic group than for any other. Sporadic references to Iberian elements in Spanish, or Caucasian traces in Armenian, etc., are found here and there, but sweeping statements of a far-reaching effect of an ethnic substratum are virtually restricted to the Germanic group. Some of the arguments advanced border closely on the pattern principle, without, however, applying it with any consistency.

2. THE PRE-GERMANIC SUBSTRATUM is asserted most definitely by A. Meillet in the following sentence: "Le germanique est l'un des groups indo-européens où le 'substrat' (ou plutôt la série de substrats) a introduit les innovations les plus grandes." 5 To Meillet we are also indebted for a number of lucid and rather complete statements of the stock arguments for this view. The whole problem is (unfortunately and absurdly) so much tainted with national bias of one sort or another that the opinion of a scholar like Meillet is of double value. For in his Langues dans l'Europe nouvelle, published at a time when war feeling was still running high (1919), he showed such admirable freedom from chauvinism and such judicious control of a wide array of facts that in his case there is no room whatever for any insinuation of prejudice. He may justly be acknowledged the leader and spokesman of those who argue for a pre-Germanic (non-Indo-European) substratum, and thereby as the most distinguished advocate of the substratum theory in general. Primarily through Meillet's sponsorship, this theory is raised from a medley of random guesses to the dignity of a scientific theory that must be taken seriously."

The arguments for this theory, stated more or less collectively for the whole group of scholars who advocate it, are chiefly the following:

Langues de l'Europe nouvelle, p. 117; cp. also 110 ff., 121, etc.

• Caractères généraux des langues germaniques, p. 19, 74, etc.

7 Cp. also Penka, Origines Ariacae; Feist, Die Ausbreitung und Herkunft der Indogermanen (1913) and Indogermanen und Germanen (31924); Kauffmann, Deutsche Altertumskunde, S. 63: "Wir bezeichnen mit Urgermanen die von der einzel-europäischen Heimat, von den Donauländern her, nach dem nördlichen Europa sich ausbreitenden Indogermanen. Diese Einwanderer müssen die an der Ostsee sitzenden Prä-Neolithiker indogermanisiert haben, werden aber auch ihrerseits in einer neuen landschaftlichen und klimatischen Umgebung durch diese Ureinwohner beeinflußt worden sein." Similarly Sweet, History of Language, Chapter VII.