For One Democratic State
in the whole of Palestine (Israel)


FOR One Man, One Vote




by Paul Wexler


“What Yiddish teachs us about the role of the Khazars in the Ashkenazic ethnogenesis”


Students of Khazar history have long been interested in the fate of the Khazars after the collapse of their empire in the 10th century. This is a difficult topic to research since the ethnonym in its native habitats fell into disuse in the 11th century, after losing its raison d’ętre as a political label. While it has been widely assumed that the non-Judaized populations merged with coterritorial and contiguous groups according to linguistic, ethnic and/or religious bonds, there remains considerable disagreement over whether the Judaized Khazars merged with non-Khazar Jews (adopting the latter’s epithet) or were assimilated to non-Jewish populations, or both. Scholars seem willing to accept the merger of Khazar Jews with other Jews in the Caucasus and possibly even Central Asia and the Fertile Crescent, but there is widespread disagreement over the possibility that Khazar Jews (i) may have augmented the Jewish communities in the East Slavic lands (the very existence of the latter has most recently been challenged by Pereswetoff-Morath 2002) or (ii) actually constituted the major component of the Ashkenazic Yiddish-speaking communities of Poland and the East Slavic lands after the 13-14th centuries. Not unexpectedly, this topic is highly charged, since a primary Khazar origin for the Ashkenazic Jews would invalidate the Zionist thesis that the contemporary Jews are largely of Palestinian Semitic origin and are more deserving of Palestine than the indigenous Palestinian Arab population.

Most scholars are justifiably sceptical about the hypothesis that the Khazars became a major component in the ethnogenesis of the Ashkenazic Jews when the latter emigrated from Germany to Poland (13th century), Belarus’ and Ukraine (15th century), since all supporters of the Ashkenazic-Khazar hypothesis (see e.g. Gumplowicz 1903, Koestler 1976 and Sobolev 1998, the latter with a rich literature) have failed to buttress their claims with convincing arguments. Arab support for the claim that the bulk of the contemporary Jews are of Khazar origin (see ‘Amāra 1967, Šākīr 1981, 1984) also does not endear the hypothesis to many readers. To be sure, the historically enormous Jewish populations in Poland and the East Slavic lands do not require a large Khazar input; theoretically the founding population could have been modest and of West European origin.

Future historians can still hope to shed light on the topic if they can determine if, where and when Khazar and non-Khazar Jews merged into a single community and what the significance of the use of the label “Ashkenazic” for some Caucasian Jewries is. For example, Nestor’s Russian Primary Chronicle distinguishes “Khazar” and “non-Khazar” Jews in Kiev by name but in the Vladimir-Suzdal’ Principality for the year 1175 only “Jews” are mentioned (see Paszkiewicz 1983, 93; Pinkus 1988, 4); in the Lezgian language of southeast Daghestan and northern Azerbajdzhan, Caucasian Jews are called “Ashkenazic” (Czortkower 1933, 148, fn 1).

However, in my view, the best evidence for the hypothesis that the Ashkenazic Jews are largely of (Slavic- and Turkic-)Khazar descent is linguistic and comes from the Yiddish language.

I believe that Yiddish is a mixed West-East Slavic language (and not a German dialect as is commonly believed—see below and my 2002). Of the two native Slavic substrata of Yiddish—Sorbian and pre-Ukrainian/pre-Belarusian (or in historical terms, “Kiev-Polessian”)—it is the latter imprint that unambiguously points to the existence of Slavic-speaking Jewries in parts of the former Khazar kingdom who eventually became speakers of Yiddish. Hence, one major venue for the birth of the Ashkenazic people would have to be in the contemporary Belarusian and Ukrainian lands, where an indigenous Slavic-speaking Jewry (as best established by the facts of Yiddish) could only be derived from the Turko-Iranian-speaking Khazars. Furthermore, Yiddish lexicon and grammar reveal links with Turkic and Iranian languages that have not been widely appreciated.

Linguistic arguments other than Yiddish have been presented in the past to determine the fate of the Khazar Jews after the 10th century but these are inconclusive. For example, a number of scholars over the last century or so have claimed to have found Khazar loans in coterritorial and contiguous languages (see Wexler 2002), but these terms, at best, can only reveal the historical areal of Khazar linguistic contacts, but not necessarily the fate of the Khazar Jews after the late 10th century, since the loans could predate the collapse of the Khazar Empire.

It is generally assumed that Khazar Jews, speaking Turkic, Iranian and other languages, could have become fluent in (pre-)Eastern Slavic as early as the 9th-10th centuries from contacts with the Slavs. Curiously, the earliest mention of the Slavic term for ‘German’ (or perhaps in these earliest attestations, in the meaning of ‘foreigner’), nmc /nemec/ (~ Uk німець),ą appears in a late 10th-century letter written in Hebrew by the Khazar king Josef to Ħasdaj ibn Šaprūţ, the representative of the Caliph of Córdoba (see Golb and Pritsak 1982, 75-121). This fact is important for two reasons: (i) It suggests that Hebrew ’aškenaz, the prevailing term for ‘German(s), German lands, Germany’ in Medieval Hebrew, was not yet in general circulation at that time (Biblical Hebrew ’aškenaz denoted an Iranian people). Note that Medieval Hebrew ’aškenaz, while now denoting Germany and German Jews, originally denoted also Khazars, Scythians and Sarmatians (see Krojs 1932, 1935, 387-389; Modelski 1910, 40, 78, 85, 92; Poznanski 1911, 76; Wexler 1987, 3, fn 9, 160, fn 49, 2002, chapter 4.7). (ii) It reveals contact with Slavic speakers, though can tell little about the status or origin of Slavic in the Khazar empire, since the few attested Khazar Slavicisms, including the above term, are found in a number of languages, see e.g. pre-migration Hungarian, Byzantine Greek and Arabic, and thus might reflect either a common corpus or knowledge of Slavic (which at this time was acquiring the status of an international lingua franca) well beyond the Khazar community. The Catalan Jewish traveler to Central Europe in the late 10th century, Ibrāhīm ibn Ja‘qūb, included the Khazars in his list of speakers of Slavic (see Peisker 1905, 113, fn 2, 133).

Since most linguists and non-linguists alike regard Yiddish as an outgrowth of High German from around the 10th century, I need first to motivate (albeit briefly) why I believe that Yiddish is in fact a mixed West-East Slavic language. Contrary to the popular view that Yiddish arose when (Judeo-?)French- and Italian-speaking Jews settled in the Rhineland and Bavaria in the 9th-10th centuries and eventually adopted/adapted local German dialects (see Vajnrajx 1973), I believe that Yiddish arose when Jews speaking Sorbian (a West Slavic language spoken in the mixed Germano-Slavic lands) first “relexified” their language to High German (and to a lesser extent, Hebrew and “Hebrew-like”, or Hebroid) “phonetic strings” approximately between the 9th-12th centuries (see details below). This means that Yiddish began as a West Slavic language with the unusual feature of possessing a predominantly German lexicon. Yiddish speakers who created an original literature in Hebrew, in the total absence of native speakers of Hebrew, had to use their native Slavic grammar with superimposed Classical Hebrew vocabulary to produce written “Hebrew”; this made the unspoken Hebrew (and most recently, the spoken Hebrew that came into being in the late 19th century) of the Yiddish-speaking Jews a bizarre Slavic language of the same magnitude as Yiddish itself (see Wexler 1990). By the 15th century at the very latest, East Slavic Jews in the Kiev-Polessian lands, comprising an unknown number of descendants of the Judaized Khazars, relexified their native Kiev-Polessian speech to the Yiddish that Ashkenazic immigrants brought to their area. Yiddish itself, readjusting to contemporary High German pronunciation norms, became now a lexifier source language for Kiev-Polessian. Since Khazar Jews are known to have migrated westward both before and after the collapse of the Khazar Empire in the late 10th century, it is conceivable that their descendants, if they were also speakers of Kiev-Polessian, and if they had contact with Yiddish speakers, could have already relexified to Yiddish prior to the Ashkenazic migration to Kiev-Polessie (the distribution of a Khazar tribal name in Central and Eastern Europe suggests where the Judaized Khazars might have settled: see Lewicki 1988 and discussion below). Important archaeological materials that lend support to the thesis that Khazar (and possibly Avar) Jews migrated westward into Europe are the stone fragments with Jewish motifs and some Hebrew writing from the 6th-8th century found in Čelarevo (Vojvodina, Serbia) and Khazar rings with Hebrew letters found in nearby Baranya district in southwestern Hungary (see Bunardžić 1978-79, 1980, 1985, Kiss 1970).

The strong likelihood that speakers of Kiev-Polessian relexified their language to Yiddish vocabulary by the 15th-16th centuries presupposes the existence of a significant Eastern Slavic-speaking Jewry. The latter, in turn, could only have its roots, for the most part, in the Judaized Turko-Iranian Khazar population that became Slavic-speaking after the collapse of the Khazar Empire in the late 10th century (and probably also before, limitedly). Hence, Yiddish offers the most reliable indication of the fate of the “lost” Khazar Jewry and compelling evidence for the claim that the contemporary Ashkenazic Jews are not in the main descendants of Palestinian Jews.

In some areas Yiddish I (relexified Upper Sorbian) and Yiddish II (relexified Kiev-Polessian) could have coexisted, while in others they could have assumed a complementary geography (e.g. Yiddish I might have prevailed in Poland and parts of the Ukraine while Yiddish II reigned supreme in the Belarusian and most Ukrainian areas), and, finally, in some areas, the two Slavic Yiddishes, drawn together by a similar relexified German lexicon and similar Slavic grammars, could have fused. The origins of contemporary Yiddish dialects should be sought in the mutual relationships between Yiddish I and II rather than in coterritorial German dialects. 

Now what is meant by relexification? Relexification means that a borrowing language adopts “empty phonetic strings” from a foreign language, while providing those strings with its own native syntactic and semantic functions. Hence, relexification is the mirror image of “calquing”, i.e. the formation of loan translations of non-native patterns of discourse using exclusively native morphemes (e.g. R выглядеть ‘appear’, with its prefix ‘out’ and the verb ‘see’ copies the component structure of German aussehen in the same meaning). In the case of Yiddish, Slavic-speaking Jews replaced their native lexicon with Germanisms, provided there were adequate overlaps between the German phonetic strings and the corresponding Slavic words. Where German lexicon underwent phonological or derivational processes (e.g. morphophonemic alternations) that violated Slavic norms, the Germanisms were usually blocked for use in Yiddish. These lexical gaps had to be filled by retaining some of the original Slavic lexicon or by using Hebrew vocabulary; where the latter was lacking, Yiddish speakers had to invent Hebroidisms (on the theoretical principles of relexification and a rich bibliography, see Horvath and Wexler 1997).

In addition to the fact that Yiddish can only make limited use of German resources, Yiddish often (under Slavic pressure) requires that accepted Germanisms be ascribed very idiosyncratic functions. Most native speakers and non-native observers have assumed that Yiddish was either a “deformation” or a “creative Jewish outgrowth” of High German, with attrition of Germanisms and acquisition of Slavicisms resulting from prolonged contact with the Slavic languages. Yet, the only major component of Yiddish that does not display significant innovative formal and/or semantic features is the Slavic component; this suggests that Yiddish was a Slavic language which uniquely exploited its two non-Slavic lexical components: Hebrew/Hebroid and German(oid).

Many studies of Yiddish morphosyntax and phonology have demonstrated the similarities between Yiddish and Slavic grammars (I exclude from this generalization most Old Yiddish texts, which are genetically Germanic and unrelated to contemporary Eastern Yiddish dialects; on “German” and “Slavic Yiddish”, see Wexler 1995). We have essentially two methodological options for establishing the claim of double relexification in Yiddish: (a) broaden the study of comparative Yiddish-German-Slavic grammar, and/or (b) show how it is possible to predict most of the Yiddish lexical corpus–in all of its componential variety. In Wexler 2002 I chose the second path, because corpus predictablity is a major diagnostic text for the relexification hypothesis (and invalidates the hypothesis of largescale Slavic influence on an allegedly Germanic Yiddish), and because the results are more striking and quicker to obtain than by a painstaking comparison of Yiddish, German and Slavic grammars.

The early German corpus of Yiddish can be predicted with considerable accuracy by comparing the lexicons and derivational machineries of German, Upper Sorbian and Kiev-Polessian. It is highly significant that we are able to anticipate which Germanisms are likely to be acquired through relexification and which are likely to be blocked from incorporation in Yiddish, and, hence replaced by non-German components. Blocked Gemanisms are replaced by original retained Slavicisms and Hebraisms; we cannot predict which Slavic and Hebrew components will be used in Yiddish, and but we can predict where in the lexicon Slavic and Hebrew (and especially Hebroid) components are apt to appear. The ability to make such predictions with an extremely high degree of accuracy is the most important diagnostic test for relexification. For example, related G Ahne ‘ancestor’ and Enkel ‘grandchild’ cannot both be expected in Yiddish since Slavic languages require different roots; not surprisingly, Yiddish has ejnikl ‘grandchild’ but ▲oves for ‘ancestors’. 

The high volume of blocked Germanisms in Yiddish, predictable only by the relexification hypothesis, explains why Yiddish has such an unusually large corpus of Hebraisms and Hebroidisms, far in excess of the Hebrew corpus of other Jewish languages. When the reduced extant Semitic Hebrew texts failed to supply replacements for blocked Germanisms, Yiddish speakers had to invent new Hebrew forms, as well as new meanings for old words. Germanisms used in Yiddish exclusively in violation of Slavic norms are few in number, and most of the examples can be identified as relatively recent post-relexification loans (after the 1600s and usually in consciously Germanized speech).

Relexification takes place exclusively when a speech community is seeking a new identity. Why, for example, might Sorbian-speaking Jews have relexified their language to High German vocabulary? A plausible answer is that numbers of pagan Sorbs identified themselves as Jews when they became household slaves of local Jewish owners or in order to avoid being sold into slavery by the Germans and Scandinavians. This is tantamount to saying that it was mainly Sorbian converts to Judaism who had the motivation to relexify, rather than the Jews themselves (see the evidence for conversion in Wexler 1991, 1993). This means that many so-called “Jewish” languages are better called “languages of converts to Judaism”. I assume that the Khazar ruling class that converted to Judaism in order to preserve neutrality vis-ŕ-vis the Byzantine Christians and the Arab Muslims in the Baghdad Caliphate also found relexification (this time to Yiddish and again to High German) attractive as a means of accentuating their newly acquired Jewish religious profile.

Most linguists and historians have assumed that the Ashkenazic Jewish immigrants arriving from the Germano-Slavic lands vastly outnumbered the indigenous Slavic-speaking Jews that they encountered in the Eastern Slavic lands (see most recently Geller 1994, 26). This assumption is based on the paucity of Yiddish-speaking observors who mentioned encountering monolingual Slavic-speaking Jewries before the 17th century (for details, see Wexler 1987). It is also likely that Yiddish speakers enjoyed a higher cultural and material level than the indigenous Slavic Jews, thus making Yiddish an attractive language for the latter. Yet, the revelation of Kiev-Polessian traits in the grammar of Old Yiddish (such as the dual number; see below) suggests that Slavic speaking Jewries were widespread at the time of the Ashkenazic eastward migrations. In order to affect such changes in the grammar of (Sorbian) Yiddish I, one would have to assume the numerical preponderance of the Slavic-speaking Jews over the Yiddish-speaking immigrants. I wonder if the Slavic-speaking descendants of the Khazars, who presumably shared a common culture with their non-Jewish neighbors, may have regarded Yiddish (and Ashkenazic culture in general?), in periods that prized a heightened ethno-religious identity, as a tool for accentuating the growing differences between themselves and the coterritorial Slavs who were gradually undergoing Christianization. The closeness of Western and Eastern Slavic languages at the time would also have facilitated the second relexification process.

The conjectured paucity of Yiddish speakers (regionally or overall) would have presented Kiev-Polessian Jews with the difficulty of acquiring Yiddish from native speakers, thus reinforcing the Kiev-Polessian “impact” on Yiddish. The presence of a prosperous German middle class in many urban centers of Eastern Europe (see Martel 1938, 195-196) could also have contributed to the attractiveness of Yiddish, which looked like a form of German. Polish sources from the 13th-14th centuries regarded the local Jews as speakers of German, i.e. as a religiously defined subgroup of the German colonists (Zientara 1974, 25-26). The identification of Yiddish speakers with the prestigious German language also could have swayed the Kiev-Polessian Jews to adopt Yiddish. A benefit of these Polish sources is that they provide a terminus post quem for the first relexification phase: the 13th century. The latter date is also confirmed by the fact that Sorbian Jews began to migrate to Poland at that time, and they are not described by contemporaries as being Slavic speakers. Given the practicality of German as a language of international trade, the burgeoning urban middle class and the legal system (see the application of the Magdeburg Law to Slavic cities), we must also reckon with the possibility that Yiddish could have spread from the German lands in the absence of significant numbers of native speakers.

There is no way to know in what precise areas Slavic-speaking Jews resided before the Mongolian/Tatar invasion in the early 13th century. But this invasion presumably propelled large numbers of Jews and other Slavs northward into the Belarusian lands (which had escaped the Mongols relatively unscathed) and westward into what was to become the Eastern Polish lands (in which a majority Ukrainian and Belarusian population resided); thus, the Tatar invasion must be seen as a sine qua non for the crystallization of Ashkenazic Jewry in the Kiev-Polessian lands and the second relexification phase. After the Tatar onslaught subsided, the Polish rulers invited Germans and Sorbian Jews inter alia to settle in the depopulated Eastern Slavic lands that they took over (see Martel 1938, 197). The Polish colonization of the Eastern Slavic lands, by encouraging the eastward migration of Sorbian Yiddish-speaking Jews (as well as the simultaneous westward migration of Slavic Jews?), would have also been a prerequisite for the second relexification phase. While Jews and Germans are frequently found side by side in some of the Polish lands, this is rarely the case in Belarus’ and Ukraine. Moreover, Polish Yiddish differs from that of the Eastern Slavic lands. These facts hint at a non-western origin for most of the Ashkenazic Jews. The use of a German lexicon by Jews should not deceive us into thinking that the speakers of Yiddish must also have come from the West (just as the use of English by Afro-Americans, either in European or African Creole form, does not mean the speakers descend from the British Isles!).

In addition, Khazar Jews joined Hungarian tribes in their migration into Central Europe before the 9th century. Lewicki 1988 has shown that place names based on Khazar tribal names (e.g. Kabar) appear in Hungary, northern Yugoslavia (Vojvodina), Poland, Austria, Slovakia, western Rumania, as well as in Ukraine and Southern Russia. Hence, the second relexification could actually have been initiated in Western or Central Europe, if contact was made there between Sorbian Jews speaking relexified Sorbian (= Yiddish I) and Khazar Jews speaking Eastern Slavic ([Judeo-]Kiev-Polessian). This raises the possibility that the time lag between the two relexifications might have been quite modest.

Yiddish appears to have valuable information on the western borders of the Kabar tribe. Consider the set of related German roots (of Middle High German origin): (a) begraben  ‘bury’, graben  ‘to dig’, Graben ‘ditch, trench’, Grab ‘grave’, Middle High G grabćre ‘gravedigger’/ (b) G Begräbnis ‘burial’, Totengräber ‘gravedigger’, Gräber ‘digger’/ (c) Grube ‘cavity, pit, mine’/ (d) Gruft ‘sepulchre, tomb, vault’/ (e) grübeln ‘ponder, muse, brood over’. Of this set Yiddish has acquired only the following morphemes: (a) bagrobn ‘bury’, grobn ‘to dig’, grobn ‘ditch’, (b) (ba)greber ‘gravedigger’/ (c) grub ‘pit; mine’. Yiddish lacks the German root for ‘grave’, using instead ▲kejver ‘grave; tomb’, ▲kvure ‘burial’, ▲kvores-man ~ ▲kvoresnik ‘gravedigger’, brengn cu kvore ‘bury (Jewish)’, while inventing karke ‘(Jewish) burial ground’ (< He qarqa‘ ‘land’). (b) G Totengräber ‘gravedigger’, though attested in Middle High German, is not ordinarily used in Yiddish, perhaps because of the ambiguity of G Gräber ‘graves’ ~ ‘gravedigger(s)’; Germanisms with multiple meanings that lack a Slavic parallel are often blocked in Yiddish. Slavic languages express ‘gravedigger’ by an agentive noun (rather than a compound noun), see e.g. Uk грабар, гробар, могильник (the failure of synonymous Middle High G grabćre, without Umlaut, to license relexification suggests the latter was unknown to Yiddish speakers).

German (d-e) are blocked in Yiddish since Upper Sorbian and Eastern Slavic lack a set of related roots that cover such a broad semantic terrain. See So (a) pohrjebać ‘bury’, hrjebać ‘to dig; scratch in the earth’, přirow, přěrow, hrebja ‘ditch, trench’, mohila, rownišćo ‘mound’, row ‘grave’/ (b) pohrjeb, chowanje ‘burial’, tótka, rowar, rowryjer ‘gravedigger’/ (c) jama, row ‘cavity, pit’/ (d) rownišćo, row ‘sepulchre, tomb, vault’/ (e) sej hłowu łamać ‘ponder’ (lit. ‘break one’s head’). See also Uk гребати ‘bury; be disgusted by; neglect; scorn’, гребти ‘to row; bury’.

It is surprising that any forms of (a) surface in Yiddish at all, in view of the formal similarities of the cognate Slavic forms (though they are not similar semantically). However, the existence of Yiddish Hebraisms for ‘bury’ and the existence of Y ■rov (< So row ‘ditch’) or ■riv (< Uk рів) suggest that even G (a) Graben ‘ditch, trench’ may have also been initially blocked in Yiddish. The association of (historically unrelated) G Graben and So row could have taken place while Common Slavic *g was either still a stop in Upper Sorbian or after it had become a fricative between the 12th and late 14th c (see Schaarschmidt 1998, 95-97 and Old So hrow!); h before a consonant tends not to be pronounced at all in contemporary Upper Sorbian.

Y ▲kejver ‘grave, tomb’ could have become popular due to its similarity to a partly Judaized Khazar tribal name, the Kabars, who participated with the Magyars in the settlement of Hungary. Old Polish Latin documents reveal the terms Kawyary (near Sandomierz 1387) and Kawyory (near Kraków, late 14th century) as names for Jewish cemeteries (for further Polish and Belarusian toponyms, see Wexler 1987, 212, fn 86). The most westerly attestation of the term appears to be in the second component of G Judenkiewer (Magdeburg, early 16th century), interpreted as Judenkiew with the German plural suffix -er (see Wexler 1987, 213 and fn 99). Magdeburg has one of the oldest Jewish communities in Slavic Germany, dating from the second half of the 10th century. I assume that the Polish Latin terms come from the name Kabar, and that when knowledge of the Judeo-Turkic ethnonym was lost among the Jews, the term could have easily been reinterpreted as He qever. For ‘cemetery’, Yiddish has a variety of Hebraisms and Aramaisms, e.g. besojlem, ▲bes-almen,bejsakvores ‘Jewish cemetery’ ~ ■molkes, etc. ‘non-Jewish cemetery’ < Slavic. German terms for ‘cemetery’ are not usually attested in Yiddish, see e.g. Friedhof, Kirchhof (> So kěrchow). 

All models which regard Yiddish as a form of German assume that Yiddish became progressively Slavicized through time. But the “Slavicization” hypothesis has two difficulties: (i) One has to prove that Yiddish speakers had an intimate knowledge of Slavic; mere contact with (Judeo-)Slavic speakers would not have been sufficient to account for the extensive “Slavicization” of Yiddish. It is imperative to collect evidence for widespread Yiddish-Slavic bilingualism from all historical places and periods. (ii) It is significant that there is little variation among Eastern European Yiddish dialects regarding the extent and even the very details of the Slavic impact (only a small recent corpus is of local origin: see Ljubarski 1927; Swoboda 1979-1980, 1990). How could Yiddish speakers, living across such a broad territorial expanse, agree on these details if the source of the Slavic influence in Yiddish is ascribed to bilingual interference in situ?

If Yiddish is a Slavic language rather than a Germanic language with a heavy Slavic imprint, then the characterization of the German component of Yiddish as native and the Slavic component as non-native is erroneous. The assumption that Yiddish is a Slavic language that underwent intensive Germanization of its lexicon rather than a Slavicized German dialect offers a smoother analysis of many hitherto vexing problems in Yiddish linguistics. Also it seems counterintuitive to claim that a language could be so radically influenced by interference from another language, as a Germanic Yiddish would have to be. Languages which have been in close contact with Slavic for as long, or almost as long, as Yiddish do not reveal the extreme “Slavicization” that allegedly characterizes Yiddish. Significantly, the Slavic and German components of Yiddish are largely in complementary distribution, with Slavic found almost exclusively in the grammar, phonology and phonotactics and German influence confined mainly to the lexicon. Impressionistically, about three quarters of the lexicon of contemporary Eastern European Yiddish are of German origin; the remainder consists roughly of Hebrew-Aramaic (c. 15%) and Slavic components (c. 10%) and a handful of unique Old Romance components that are unattested in German. One would be tempted to argue that the relatively impoverished German component of Yiddish, measured both by the quantity of its German roots in comparison with the putative German lexifier dialects, and the near absence of Yiddish synonyms of German origin, could easily be ascribed to attrition after many centuries in contact with Slavic languages. This is not convincing since (i) Yiddish never lost contact with German in Eastern Europe, and (ii) we can largely predict the original German lexical corpus of Yiddish and identify post-relexificational Germanisms (i.e. after the 16th century). Indeed, all relexified languages appear to have a smaller superstratal lexicon than the lexifier dialect itself; compare Haitian Creole with French. Finally, if Slavic were the cause of the attrition of the German component of Yiddish, we should expect a far larger Slavic, but certainly not a Hebrew, component.

The linguistic evidence for my two-tiered relexification hypothesis in the genesis and history of Yiddish can be summarized under five headings (here in brief; for further details, see my 2002):


(a) German morphophonemic alternations (derivational patterns) that do not enjoy Upper Sorbian and/or Kiev-Polessian parallels are blocked in Yiddish. German roots which match roots with similar form and meaning in Upper Sorbian and/or Kiev-Polessian (the words in question are mainly cognates) were usually blocked in Yiddish, since they were apparently perceived as Slavic elements. It is difficult to determine how much similarity in form and meaning is required to cause blockage of a Germanism. The following German words are available to Yiddish apparently since their Upper Sorbian (pseudo-)cognates are formally and/or semantically sufficiently different and the genetic relationship apparently was not immediately clear to naďve bilinguals, e.g. G heilen ‘to cure’ > Y hejln, G Heilung ‘cure’ > Y hejlung, a cognate of So cyły ‘all’ (curiously, G heil ‘whole, uninjured’ is not regularly used in Yiddish). Presumably, G Tausend ‘thousand’ is acceptable to Yiddish as tojznt since cognate  So tysac, Uk тисяча are too distant formally.

German roots are blocked in Yiddish if Upper Sorbian and/or Kiev-Polessian translation equivalents do not broadly overlap semantically. Future studies will need to determine whether the matching of German and Slavic lexicon by relexifiers (leading to the decision to relexify or block Germanisms) operated on the minimal domain of individual words or on the maximal domain of root sets. For example, Upper Sorbian has a set of derivatives that includes móc ‘power; be able; possess’, pomhać ‘to help’, pomoc ‘help’, móžno ‘possibly’. The likelihood of finding a German root with derivatives matching most or all of these meanings is small. The relexifiers would then have to choose between (i) linking all the Slavic and German forms as a set, and (ii) matching only parts of the two root sets, in accord with the principle of semantic similarity. It would appear that the second option is almost always elected. Thus, So móc ‘power’, móžnosć, móžnota ‘possibility’ are relexified to Y (cognate) maxt, miglexkejt, but So móc ‘be able’ > Y kenen, while So pomhać ‘to help’ and pomoc ‘help’ > Y helfn and hilf. If the first option had been selected, much more of the German lexicon would have been blocked, thus necessitating an even greater number of Hebraisms and substratal Slavicisms; in other words, relexification would have been in practice impossible. The problem is that if Hebraisms (replacing blocked Germanisms) were also acquired to express some of these meanings, it is not immediately clear if their presence is due to the failure to match all the forms of So móc with a single German root, or because of disparallels between parts of the two paradigms. For example, alongside kenen ‘be able’ (< German), Yiddish also uses zajn bekojex, zajn bixojles, (humorous) joxlen; for ‘perhaps, possible’, Yiddish has ▲efšer, tomer alongside miglex < German. Most likely, these Hebraisms were acquired because of the awkwardness of the un-Slavic alternation of cognate G kennen ‘know’/können ‘be able’ and the use of different Slavic roots for part of the paradigm, e.g. So snadź, snano ‘perhaps’.


(b) Yiddish frequently lacks the volume of German-origin synonyms that can be found in most German dialects.


(c) Yiddish uses the plural marker ■/-(e)n (< G -[e]n) with German nouns in violation of German norms and often in imitation of the “pseudo-dual” in Kiev-Polessian. This suggests that Yiddish once had a dual category. (See my 2002 for the argument that Y ■/-[e]n was chosen because of its similarity with the Common Slavic root infix -en- which became partly linked with the plural number, see Uk імя ‘name’, genitive імени ~ імя, plural іменa.)

The distribution of the plural marker (■)-(e)n in Yiddish differs radically from that of its German counterpart, in that it generally matches Ukrainian and Belarusian (and less frequently Russian) nouns which have a separate ending after the numerals 2-3-4 which consists of the plural ending + the singular stress position. This ending can be called the “new pseudo-dual” (since it differs from the Common Slavic dual endings and involves 3 and 4). The pseudo-dual is restricted to the nominative-accusative cases; in other cases the numerals govern the normal plural forms. Only Ukrainian and Belarusian nouns with mobile stress are potential candidates for the pseudo-dual; in contrast, Y (■)-(e)n is not restricted to nouns following a numeral. This would be consonant with the genuine dual which might have been obtained from Upper Sorbian in the first relexification phase. Since the Yiddish distributional facts now correlate well with the Ukrainian-Belarusian pseudo-dual, it might be more appropriate to speak of a pseudo-dual in Yiddish as well. Significantly, the pseudo-dual is not widely used in Ukrainian, Belarusian or Yiddish with nouns which denote a paired object (e.g. ‘ear’, ‘eye’, ‘scissors’). Furthermore, the pseudo-dual in Ukrainian and Belarusian (< Kiev-Polessian) can survive since it has formal marking; Yiddish lacks the ability to shift stress onto desinences (due to German influence). Since presumably the Yiddish pseudo-dual was originally not employed widely with nouns denoting paired objects, it not unexpectedly fell into disuse and (■)-(e)n became unambiguously a marker of plural number.

I suggest that the Kiev-Polessian pseudo-dual became part of Yiddish when Kiev-Polessian was relexified to Yiddish (and German) lexicon (presumably effacing the Upper Sorbian dual). Nowadays, (■)-(e)n denotes plural number exclusively. The rich literature on bilingual interference gives no examples of a dual category successfully borrowed by a language previously lacking the category. Hence, a dual number in Yiddish would be powerful evidence in support of the relexification hypothesis.

Contemporary Yiddish tends to apply the (■)-(e)n plural ending on phonological grounds (i.e. depending on the final consonant of the singular stem). This fact serves to obfuscate the historical functions of Y (■)-(e)n; still, a sizeable number of old Germanisms in Yiddish which arguably took the (■)-(e)n ending when it had a dual function can still be identified. I presume that originally nouns with (■)-(e)n in a dual function had another suffix to denote the plural.

The pseudo-dual is far more productive in contemporary Ukrainian than in Belarusian; the Belarusian dictionary compiled by Sudnik and Krywko (1999) has approximately 150 examples, mostly feminine nouns, while Ukrainian dictionaries cite over 600 nouns with the pseudo-dual, again, mainly with feminine nouns (see Pohribnyj 1964). Whereas Belarusian now has only about 15 masculine nouns, Ukrainian has over 200. Moreover, the distribution of the pseudo-dual in Ukrainian forms a better match with Y (■)-(e)n than Belarusian, thus providing precious linguistic support for the assumption that the bulk of the Kiev-Polessian Jews of mainly Khazar origin who first became Slavic speakers originally resided in the southern (i.e. pre-North Ukrainian) part of the territory. Yiddish data may also raise the possibility that East Slavic nouns now lacking the pseudo-dual once had it; e.g., Y trer(n) ‘tear (in the eye)’ could only have its (■)-n from Br сляза which has a pseudo-dual, since cognate Uk сльоза presently has stress shift in the root but no pseudo-dual. Either Y trer(n) was modeled on Br сляза, or Uk сльоза once had the pseudo-dual. Alternatively, the -n plural of synonymous G Träne (unknown in Yiddish) may be the basis for the Yiddish plural choice (especially after relexification). 

There is no way to posit an accurate chronology for the pseudo-dual in Ukrainian and Belarusian since stress is rarely marked in texts before the 16th century, but the existence of the phenomenon in both Ukrainian and Belarusian dialects suggests a Kiev-Polessian origin, i.e. predating the 14th-century (the date that Shevelov notes as the first attestation of the plural in Ukrainian after ‘2’: 1963, 234-236). Sorbian could not have developed a pseudo-dual since the stress became fixed on the initial root syllable between the 12th-14th centuries (Schaarschmidt 1998, 87-88) and the original dual category survived.

The East Slavic use of stress to distinguish the plural and pseudo-dual could, in theory, have been maintained in Yiddish after the second relexification phase. For example, some Yiddish nouns distinguish singular and plural stems by stress shift (together with a plural marker), e.g. (■)miníster (plural ministórn) ‘minister’ < G Miníster or Uk мінíстер (with fixed stress, ultimately < Latin), ▲tálmed (talmídim) ‘student’, ■pól’ak (pol’áken) ‘Pole’. In Y fúrman (furmánes) ‘driver’, the components are of German origin but (following So wóznik, Uk водій) lack the suppletion found in Middle High G vuorman (vuorliute). However, there are no examples of a non-singular noun in Yiddish with two stress patterns, say ■ministórn plural vs. *minístorn dual. The age of moveable stress in Yiddish nouns is unknown; it is widely believed that the Yiddish pronunciation of Hebrew loans probably arose in a language other than Yiddish. (I suspect that the tendency to move stress to plural endings in Modern Hebrew is either of East Slavic origin or Old Hebrew supported by parallelism with East Slavic.) 

The Yiddish dual must have become unproductive sometime after the second relexification phase in the 15th-16th centuries, to judge from the fact that Yiddish productively uses (■)-(e)n as a plural marker far in excess of G -(e)n (though modern German influence also contributed to the productivity of Y [■]-[e]n).

When a Yiddish noun with (■)-(e)n corresponds to a number of Ukrainian terms, only some allowing the pseudo-dual, we can reconstruct the likely Slavic input for relexification. For example, Uk lavka ‘shop’, but not kramnycja, takes the pseudo-dual. Hence, I assume that the Yiddish choice of (■)-n in gevelb(n ~ -er) ‘shop, store’ was motivated by Uk лавкa (vs. G Laden [Läden]). Consider also Uk війнa (with the pseudo-dual) ‘war’ ~ Y krig(n) vs. G Krieg(e). While synonymous Uk бій and боротьба both lack the pseudo-dual, the corresponding Y kamf ‘struggle, fight, combat’ pluralizes with (■)-n (vs. G Kampf[Kämpfe]). I suspect that Yiddish came to innovate in its distribution of (■)-(e)n in the dual function especially after the second relexification phase, when the Slavic substratum no longer operated so pervasively, and/or Jews gradually became monolingual in Yiddish. This seems borne out by the likelihood that Y kamf–with -f is a recent acquisition (vs. older Y kop with -p < G Kopf).

A number of German, Slavic and occasionally Hebrew nouns in Yiddish take two or three plurals, usually without semantic implications, see e.g. Y vogn(s ~ vegn ~ vegener) ‘cart’, kni(es ~ -en ~ zero plural) ‘knee’, noz (nez ~ ne[j]zer ~ rare -n) ‘nose’. Multiple pluralization in Yiddish could (i) result from the merger in the standard language of different dialectal choices, (ii) reflect differences and changes in productivity through time and space, (iii) result from a survival of the original plural and dual endings.

In Y ▲lošn ‘language’, double prefixation seems to entail a dual meaning; contrast the expected plural ▲lešojnes vs. lešojnesn ‘(Classical) Hebrew and Aramaic’ (Rejzen 1926, column 409). The “union” of the latter two languages is possible because they are closely related Semitic languages with common unspoken literary and liturgical functions in the Yiddish-speaking communities; as a result, they are also called by the common Y glottonym lošn kojdeš (lit. ‘holy language’). Double prefixation with Germanisms in Yiddish sometimes provides a means of distinguishing between dual number (with single prefixation) and plural number (double prefixation). The creation of the unambiguous plural form Y ojgenes ‘eyes’ (~ G Auge[n]) suggests an attempt to remove the ambiguity of ojgn ‘eyes’, which is conceptually dual or plural; ojgenes plural would leave ojgn (vs. G Augen plural) free to mark the dual unambiguously. 

Yiddish has some compound nouns of non-German origin with a dual meaning which are used in the literary language mainly as a plural and in colloquial Yiddish in the singular, e.g. Y (■)tate-mame ‘(two) parents’ (lit. ‘father’ + ‘mother’), ■zejde-bobe ‘(two) grandparents’ (lit. ‘grandfather’ + ‘grandmother’). The first example matches Polish ociec i mać, attested since the early 15th century or Uk батько-мати; the second example matches So dźěd a wowka ‘grandparents’.


(d) The gender assignment of many German and Hebrew nouns in Yiddish dialects follows the gender of the Kiev-Polessian rather than of German and Hebrew translation equivalents, see e.g. Y bet feminine, neuter ‘bed’ ~ Uk ліжкo neuter (all Slavic neuters > feminine in Yiddish) vs. G Bett neuter exclusively.

(e) An important piece of evidence for the hypothesis that Kiev-Polessian contributed some unrelexified lexicon comes from the geography of Eastern Slavicisms in Yiddish, many of which appear to come specifically from southern and western Belarusian and northern and western Ukrainian dialects–i.e. precisely from the area of the original Kiev-Polessian dialect up until its disintegration and realignments in c.1400, see e.g. Y ■bereze, ber’oze ‘birch’ ~ ubiquitous Br бярозa, while бярэзa is rare and limited to the southwest area; see also Uk березa (Dyjalektalahičnyj atlas belaruskaj movy 1963, map #29); Y ■bloxe ‘flea’ ~ Uk блоха ~ блиха (Atlas ukrajins’koji movy 1984, map #84). It is impossible to determine whether the number of unrelexified Kiev-Polessianisms in Yiddish was originally larger than it is today, or whether in the last six centuries since the disintegration of Kiev-Polessian, many Kiev-Polessianisms have been replaced by new Ukrainian and Belarusian localisms.

While there is broad agreement that Polish and the three Eastern Slavic languages are the major donors, there is no unanimity over the origins of many individual Yiddish Slavicisms. The relexification hypothesis offers a means of separating substratal from adstratal Slavicisms in Yiddish. More Eastern Slavicisms appear in Polish Yiddish than Polonisms in Eastern Slavic Yiddish. This fact was noted by Šulman (1939, 82), who cited a figure of about 10% Polonisms among the Slavicisms in Eastern Slavic Yiddish, as opposed to some 50% Eastern Slavicisms in Polish Yiddish, see e.g. PolY ■mučen ‘to torment’, ■p’ate ‘heel’, ■pi(š)čevke ‘trifle’ (< Uk мучити, пята, підшивка ‘lining’ vs. Pol pięta, męczyć [> Central PolY menčen], podszewka).

The relatively small number of Polonisms in Ukrainian-Belarusian Yiddish is remarkable for three reasons: (i) At the end of the 19th century the Jews constituted the dominant ethnic group in both small and large towns in the Vicebsk and Mahilëw gubernijas alongside Poles. (ii) According to the traditional theory, Jews allegedly migrated from Germany across Poland into Belarus’ and Ukraine. (iii) The meager impact of Polonisms on Eastern Slavic Yiddish stands in sharp contrast to the direction of diffusion of Slavicisms between Polish and the Eastern Slavic languages. The Polish impact on Belarusian and Ukrainian is much greater than the impact of the latter on Polish dialects, i.e. we see the opposite of the intra-Yiddish situation. No less than 14% of the Ukrainian vocabulary has been shown to be of Polish origin (Shevelov 1975, 452-453, fn 12). 

The intensive Polonization of Belarusian and Ukrainian means that some of the Polish impact on Eastern Slavic Yiddish may have come into the latter through the secondary intermediary of Belarusian and Ukrainian. This further reduces the direct impact of Polish (or Polish Yiddish) on Eastern Slavic Yiddish, since a certain amount of the Polonisms, from the point of view of the Yiddish target dialects, are essentially Ukrainianisms and Belarusianisms. For example, Y ■jatke ‘meat market’, though ultimately of Polish origin, could have been acquired directly from Pol jatka or from Uk яткa (where the Polonism first appeared in the 16th century). Another problem in identifying Polish influences in Yiddish is that Eastern Slavicisms can assume a Polish-looking form in Yiddish, e.g. Y ■blote ‘mud, filth, dirt’ could be < Uk болото ‘swamp, marsh; mud, dirt’ with the loss of the initial unstressed syllable, and not necessarily from Pol błóto.

Two other possible indications of unrelexified Judeo-Slavic corpus are when Yiddish Slavicisms either differ from coterritorial Slavic formally or are rare Slavicisms/Slavoidisms; both types of lexicon are critical for reconstructing the outlines of a Judaized Eastern Slavic. One example is Y ■pral’nik ‘laundry beetle’. Br пральнік, Uk пральник are found only at scattered points in south central Belarus’ (on the Ukrainian border), at some points in northwest Ukraine next to the Belarusian frontier and in the areas between Rivne and Luc’k and between Rivne and Novhorod-Volyns’kyj (see Wexler 1987, 95, 186-188; forms such as Br [a]пранік, Br, Uk прач, etc. are the preferred terms). See also Pol dialectal praln’ik ‘laundry beetle’ in the Sejn district, perhaps < Belarusian. The scattered nature of the few surviving locales with Br пральнік and Uk пральник suggests a once popular form in Kiev-Polessian. 

Finally, there are concepts which are expressed in Yiddish by Slavic elements often without German or Hebrew synonyms; these terms appear to be substratal elements that withstood relexification in the first and second relexification phases, probably in order to denote culture-specific terms. These Slavicisms fall into a number of unrelated semantic domains, see e.g. Jewish religion and culture (this is a very significant fact, see e.g. Y trejbern ‘porge the meat of ritually forbidden parts’ ~ Uk теребити ‘to peel, shell; eat greedily; clear a field’), flora and fauna. Many of the Eastern Slavic or Upper Sorbian terms in Yiddish denoting flora and fauna are also widely used in early Slavic toponyms, e.g. Slavic bagno ‘swamp’ is much rarer in Slavic placenames than Slavic blato, etc. (and aside from Ukrainian and Belarusian, limited to West Slavic languages); the latter is used in Yiddish but not the former. The retention of some Slavic tree terms, see e.g. Y ■dub ‘oak’, ■bereze ‘birch’ (and variants), may reflect the revered status of these trees in pre-Christian Slavic society (see Gimbutas 1967, 744-745). The blockage of relexification to Germanisms is also likely in the semantic domain of topography. Many Slavic roots which are still abundantly attested in German place names of Sorbian and Polabian origin are preserved in Yiddish (where they are presumably of Sorbian origin), e.g. Y ■ričke ‘stream, creek’ < So rěčka, Uk річка.

Sorbian fruit terms are often kept in Yiddish. For example, today Yiddish lacks G Beere ‘berry’, using uniquely ■jagde of Slavic origin; see also Y ■malene ‘raspberry’ < So malena, Uk малинa vs. G Himbeere (~ Maline < Slavic). Also some Slavic food terms appear to have been retained in Yiddish, see e.g. blince ‘pancake’. The Southeastern German dialects have borrowed the term as well. Yiddish may have succeeded in retaining the Upper Sorbianism perhaps because the German baked goods were prepared in a different way. Kieser (1972, 164) notes that Plinsen was mainly made out of buckwheat flour and milk. Владимирская (1982, 75) discusses the different meanings of R блин(ц)ы in dialects of the Balakleev region of the Xarkiv oblast’.

When Yiddish has concepts that have a parallel in Slavic languages but not in German, I would expect to find such concepts denoted in Yiddish by unrelexified Slavisms or newer Hebraisms/Hebroidisms. An example is Y mexutn ‘father-in-law’. If Yiddish speakers fail to use the Slavic equivalent (e.g. Uk сват), this could reflect a change in its original meaning, either among Yiddish and/or Slavic speakers, thus necessitating the use of a new word in Yiddish. A new disparallelism, either during or after the second relexification phase, could have necessitated the use of a distinctive Hebroidism, like Y mexutn (see Wexler 1993, 174, 1997). Interestingly, in some East Slavic languages, the term denoting ‘matchmaker’ (either male or female) is also expressed by the same root, see e.g. R сват, свахa. But in Yiddish, ‘matchmaker’ is expressed solely by Hebroid šatxn. This is reminiscent of the Ukrainian dialectal practice of distinguishing сват, свахa ‘parent-in-law’ from старостa ‘matchmaker’ (< старий ‘old’). The use of a special term ‘parent-in-law’ is typical of Slavic, Turkic, Mongolian but not German.


While Eastern Yiddish dialects reveal Kiev-Polessian grammatical features that could not have entered the former via bilingual interference, there are almost no Altaic grammatical or phonological features in Yiddish (with the possible exception of the periphrastic construction for Hebrew verbal elements shown below). This suggests that the descendants of the Khazar Jews acquired Yiddish when they were already predominantly Slavic-speaking. The geography of these lexical terms and the grammatical construction within Yiddish (within Germany as far west as the original limes sorabicus dividing the Holy Roman Empire from mixed Germano-Slavic areas in the 9th century) and their presence in other languages support the claim of a Khazar migration to Western Europe (i.e. western German lands) before the collapse of the Khazar Empire at the hands of Kievan Rus’ in the late 10th century. An open question is how long the Kiev-Polessian Jews retained fluency in a Turkic language. Several years ago I raised the possibility of a Turkic substratum that might have manifested itself in the stereotyped Ukrainian speech of Jews in Ukrainian plays (so-called “intermediaries”) of the 17th-18th centuries. The chief feature of the Jewish Ukrainian speech was sibilant confusion, which could be ascribed to a Turkic language, though other explanations for the stereotyped speech come to mind (see details in Wexler 1987, 192ff and 1994).

Yiddish has three types of evidence for specific links with Turko-Iranian-speaking (Jewish?) communities:


(a) Yiddish ordinarily uses a periphrastic conjugation to integrate Hebrew verbal material.

Most Hebrew verbal material reaches Yiddish in the form of the masculine singular participle, which becomes indeclinable in Yiddish and must be conjugated periphrastically by means of one of two German auxiliary verbs. Occasionally, Eastern Yiddish conjugates Hebrew finite verbal material in a standard, non-periphrastic conjugation; dual integration of a single Hebrew stem is rare, see e.g. batkenen ‘inspect (slaughtered animals for impurities)’ ~ periphrastic ▲bojdek zajn ‘examine, scrutinize, inspect’ < Hebrew bādqū ‘they inspected’ and bōdeq ‘he inspects, inspecting’ + G sein ‘be’, respectively. The periphrastic conjugation is intended almost exclusively for Hebraisms in Slavic Yiddish; it is exceedingly rare in Western (German) Yiddish and is totally unknown in German Rotwelsch (slang), where the non-periphrastic conjugation prevails. The periphrastic conjugation for Hebraisms is available to other Jewish languages, such as, e.g. Karaite (also used for Arabisms), 17th-century Judeo-Eastern Slavic and Balkan Judeo-Spanish. Significantly, the periphrastic conjugation is also extremely productive in Iranian, Turkic languages and Dungan (the Chinese spoken by Muslims) for the integration of Arabic verbal material. The geography of the periphrastic conjugation prompts me to suggest that the Yiddish construction might have its roots in a Judeo-Turkic language, such as Khazar. In that case, the periphrastic conjugation most likely would have reached Yiddish only in the second relexification phase. I also suspect that the actual distribution of the periphrastic conjugation in Yiddish reflects the use of parallel periphrastic and/or compound constructions in the Kiev-Polessian substratum, and hence can block German translation equivalents (see details in my 2002).

The fact that German may also use prefixed verbs or verbal complements does not seem to affect the distribution of the Yiddish periphrastic conjugation. For example, Y ▲bojdek zajn ‘examine, scrutinize, inspect’ is matched by German terms which were blocked in Yiddish for one reason or another, see e.g. G be(auf)sichtigen, beobachten, beschauen, (er)forschen, (nach)prüfen, untersuchen, etc. The non-periphrastic Y batkenen expresses a ‘religious inspection’ (vs. Old Hebrew ‘inspect’), for which there is no immediate German counterpart. The Ukrainian translation equivalents are always prefixed verbs or verbs which tend to have an adverbial complement, see e.g. Uk допитувати, наглядати, оглядати, перевіряти, (старанно) досліджувaти, вислухувати. In Yiddish, the notion ‘to slander, calumniate’ requires a prefixed Germanism, a simplex with a complement or a periphrastically integrated Hebraism, such as Y baredn < redn ‘to talk’ (< G bereden, also ‘talk over, discuss, persuade’, meanings unattested in Yiddish), ▲redn rexiles ojf (lit. ‘speak gossip about’), ▲mojce-šemra zajn ojf (lit. ‘take out a bad name on’), redn rišes (lit. ‘speak evil’; the second term also means ‘anti-Semite’), maxn a bilbl (lit. ‘make confusion, libel’), ▲malšn zajn ojf (< lit. ‘inform against’). The corresponding Ukrainian terms also tend to be periphrastic, see e.g. Uk зводити наклеп, порочити репутацію ‘to slander, calumniate’ (lit. ‘bring slander, insult a reputation’, respectively) ~ non-periphrastic but prefixed доносити, обвинувачувати, засуджувати ‘denounce’.


(b) Yiddish has occasional Hebraisms with an atypical Ashkenazic phonetic form. Y ▲ta’arebret ‘bier’ < Old He ţāhārāh should be either tohoro in Ashkenazic “whole” Hebrew (i.e. monolingual Hebrew texts read by Yiddish speakers) or “merged” (i.e. colloquial) Y to’ore. The qamac and pataħ diacritics in Old Hebrew > Yiddish as /o ~ u/ in open and /a/ in closed syllables. Other exceptions are found in Slavic Hebraisms, see Br кагал ‘Jewish community (organization)’ (vs. Y kōl, kūl). The deviant reading of the Hebrew diacritics could theoretically be a relic of the reading norms of Khazar Jews inherited by Ashkenazic Jews. Similarly, unexpected forms of Hebraisms in Western Yiddish (created to the west of the Elbe river) could be vestiges of Judeo-French or Judeo-Italian reading norms (see Jacobs et al. 1994, 396-397; Wexler 1988, 96-116).


(c) Yiddish has a few Iranian words which are also attested in Eastern Slavic, but not always in the same form and meaning. This suggests that Yiddish could have acquired the terms through its own contacts with the relevant source languages and not indirectly through coterritorial Slavic intermediaries. Two variants of a single Iranianism are (i) Br, UkY šabaš ‘tip given to a musician at a wedding by guests who participate in the dancing’ and (ii) pan-Y šibeš ‘trifle; small coin’ (the latter is spelled as if it were a Hebrew term but has not acquired the meaning of He šibūš ‘complication’, which could theoretically also become *šibeš in Yiddish). The first variant, usually in a truncated form (but often with a meaning closer to that of the second variant), is found in non-Jewish languages extending from Russian to Dutch slang (see Wexler 1987, 64-69, 218 and 1993, 108-110). The meaning of the first variant, restricted to Yiddish, is identical to that of the etymon of both variants—Persian šābāš.


In 1993 I regarded the Ashkenazic Jews as a “Slavo-Turkic people in search of a Jewish identity”; that may have been an appropriate description for the first relexification phase. For the second relexification phase, first proposed in 2000, it is more accurate to speak of a “Turko-Slavic people”. Hopefully, geneticists will soon have something to say about the changing mix of Slavic and Turkic ethnicities in the ethnogenesis of the Ashkenazic Jews; for the time being, though, Yiddish linguistics is the sole proof of a Khazar component in the ethnogenesis of the Ashkenazic people.



ąThe following symbols of component origin are used with examples:  ▲ = Hebrew origin, = Hebroidism; ■ = Slavic origin, = Slavoidism.



‘Amāra M. Isrā’īl. Hal hija sāmijja? — Cairo, 1967;

Bunardžić R. Izveštaj sa zaštitnog arxeološkog iskopavanja rano-srednjovekovne nekropole lokalitetu “Ciglana” kod Čelareva// Graħia za proučavanje spomenika kulture Vojvodine. — Novi Sad, 1978-1979. — V. 8-9 — P. 33-67;

—. Izložba menore iz Čelareva. — Beograd, 1980;

—. Čelarevo. Risultati delle ricerche nelle necropoli dell’alto medioevo. — Rome/Novi Sad, 1985;

Czortkower S. Podłoże rasowe żydów kaukaskich// Archiwum Towarzystwa Naukowego we Lwowie. — 1933. — part 3, V. 6, book 8. — P. 145-176;

Geller E. Jidysz: Język Żydów polskich. — Warsaw, 1994;

Gimbutas M. Ancient Slavic religion. A synopsis// To honor Roman Jakobson. — The Hague-Paris, 1967. — V. 1. — P. 738-759;

Golb N. and O.Pritsak. Khazarian Hebrew documents of the tenth century. —  Ithaca/New York/London, 1982;

Gumplowicz M.E. Początki religii żydowskiej w Polsce. — Warsaw, 1903;

Horvath J. and P.Wexler (eds.) Relexification in creole and non-creole languages. (With special reference to Modern Hebrew, Haitian Creole, Romani and Rumanian.) — Wiesbaden, 1997;

Jacobs N.G., E.F.Pince and J. van der Auwera. Yiddish// E. König and Johann van der Auwera (eds.) The Germanic languages. — London, 1994. — P. 388-419;

Kieser O. Diatopik eines Wortschatzes nach Sachgruppen. Dialektuntersuchung des Kreises Liebenwerda. — Giessen, 1972;

Kiss A. 11th century Khazar rings from Hungary with Hebrew letters and signs// Acta archaeologica hungarica. — 1970. V. 22. — P. 341-348;

Koestler A. The thirteenth tribe. The Khazar Empire and its heritage. — London, 1976;

Krojs Š. Hašemot aškenaz usfarad// Tarbic. — 1932. — V. 3(4). — P. 423-435;

—[Krauss S.] Die hebräischen Benennungen der modernen Völker// S.W.Baron and A.Marx (eds.) Jewish studies in memory of George A. Kohut. — New York, 1935. P. 379-412;

Lewicki T. Kabarowie (Kawarowie) na Rusi, na Węgrzech i w Polsce we wczesnym Średniowieczu// G.Labuda and S.Tabaczyński (eds.) Studia nad etnogenezą Słowian i kulturą Europy wczesno-średniowiecznej. — Wrocław/Warsaw/  Kraków/Gdańsk/Łódź, 1988. — V. 2. — P. 77-87;

Ljubarski K.F. Ukrainiše verter in der jidišer šprax (hojptzaxlex in doremdikn dialekt)// Записки Одеського Інституту Народньої Освіти. — 1927. — V. 1. — P. 161-130 (reverse pagination);

Martel A. La langue polonaise dans les pays ruthčnes. — Lille, 1938;

Modelski T.E. Król “Gebalim” w lisćie Chasdaja. — Lvov, 1910;

Paszkiewicz H. The rise of Moscow’s power. — Boulder/New York, 1983;

Peisker J. Die älteren Beziehungen der Slawen zu Turkotataren und Germanen und ihre sozialgeschichtliche Bedeutung. — Berlin/Stuttgart/Leipzig, 1905;

Pereswetoff-Morath A. A grin without a cat. Jews and Christians in medieval Russia.  — Lund, 2002. — V. 1-2;

Pinkus B. The Jews of the Soviet Union. The history of a national minority. — Cambridge, 1988;

Poznanski S. [Review of Modelski]// Zeitschrift für hebräische Bibliographie. — 1911. — V. 15. — P. 74-76;

Pritsak O. The origin of Rus’. Old Scandinavian sources other than the Sagas. — Cambridge, MA., 1981. — V. 1;

Rejzen Z. Di deklinacje fun dem šem-ecem in jidiš// Filologiše šriftn fun YIVO. — 1926 — V. 1. Columns 387-412;

Šākīr ‘A. ’ar-R. Dawlat al-xazar al-džadīda aw isrā’īl. — Beirut, 1981;

—. Al-mamālīk aş-şahājina wa-l-maşīr al-‘arabī. — Cairo, 1984;

Schaarschmidt G. A historical phonology of the Upper and Lower Sorbian languages. — Heidelberg, 1998; 

 Shevelov G.Y. The syntax of modern literary Ukrainian. — The Hague, 1963;

—. On lexical Polonisms in literary Ukrainian// V.Erlich, R.Jakobson, C.Miłosz, R.Picchio, A.M.Schenker and E.Stankiewicz (eds.) For Wiktor Weintraub. — The Hague/Paris, 1975. — P. 449-463;

Šulman M. Slavizmen in der leksik fun jidiš// Afn špraxfront. — 1939. — 3rd series. —  V. 3. — P. 71-109;

Swoboda V. Ukrainian in the Slavic element of Yiddish vocabulary// Harvard Ukrainian studies. — 1979-1980. — V. 3-4. — P. 818-825;

—. Ukrainianisms in J.M.Lifšic’s Judeš-rusišer verter-bix// P.Wexler (ed.) Studies in Yiddish linguistics. —  Tübingen, 1990. — P. 101-109;

Vajnrajx M. [Weinreich]. Gešixte fun der jidišer šprax. — New York, 1973. — V. 1-4.

Wexler P. Explorations in Judeo-Slavic linguistics. — Leiden, 1987;

—. Three heirs to a Judeo-Latin legacy: Judeo-Ibero-Romance, Yiddish and Rotwelsch. — Wiesbaden, 1988;

—. The schizoid nature of Modern Hebrew: A Slavic language in search of a Semitic past. — Wiesbaden, 1990;

—. Yiddish–the fifteenth Slavic language// International journal of the sociology of language. — 1991. — V. 91. — P. 1-150, 215-225;

—. The Ashkenazic Jews. A Slavo-Turkic people in search of a Jewish identity. — Columbus, 1993;

—. Judeo-Slavic frontispieces of late 18th-19th-century books and the authentication of  “stereotyped” Judeo-Slavic speech// Welt der Slaven. — 1994. — V. 39(2). — P. 201-230;

—. Languages in contact: The use of Rotwelsch and the two Yiddishes// R.P-c.Hsia and H.Lehmann (eds.) In and out of the ghetto: Jewish-gentile relations in late and medieval and early modern Germany. — Washington/New York, 1995. — P. 109- 124;

—. The relexification hypothesis in Romani// J.Horvath and P.Wexler (eds.) Relexification in Creole and non-creole languages. (With special reference to Modern Hebrew, Haitian Creole, Romani and Rumanian.) — Wiesbaden, 1997. — P. 100-161; 

—. Што могуць ідыш, сучасны іўрыт, эсперанто, а таксама сербалужычане, палешукі і хазары сказаць пра паходжанне габрэяў-ашкеназі?// Arche. — Mінск 2000. — V. 3. — P. 117-126;

.  La relexification en Yiddish: (Les Juifs, les Sorabes, les Khazars et la région deKyjiv-Polissja.)// Canadian Slavonic papers/Revue canadienne des slavistes. 2001. V. 43(4). P. 533-556;

—. A German-Slavic bilingual unknowingly knows Yiddish and Modern Hebrew quite well: Why and what?// Welt der Slaven. — 2002. — V. 47(1). — P. 57-84;

—. Two-tiered relexification in Yiddish (Jews, Sorbs, Khazars and the Kiev-Polessian dialect). — Berlin/New York, 2002;

Zientara B. Foreigners in Poland in the 10th-15th centuries: Their role in the opinion of Polish medieval community [sic!]// Acta poloniae historica. — 1973. — V. 29. — P. 5-28.

Атлас української мови. — Київ, 1984 — V. 1;

Владимирская Е.А. Лексическая система южнорусского говора в условиях русско-украинского билингвизма// Русские говоры на Украине. — Киев, 1982. — С. 58-85;

Дыялекталагічны атлас беларускай мовы. — Мінск, 1963;

Погрібний M.I. Словник наголосів української літературної мови. — Київ, 1964;

Соболев Д. Возврашчение в Хазарию// Двадцать два. — 1998. — Т. 108. — С. 162-192;

Суднік М.Р. and M.Н.Крыўко (eds.) Тлумачальны слоўнік беларускай літаратурнай мовы. — Мінск, 1999.