The Origins of Yiddish—A Response to Philologos, Part Fir

Sep 5, 2014 by

[Many thanks to Merlin Dorfman for bringing the Philologos’ essay to my attention and to Martin W. Lewis for his helpful comments and corrections to this post! I am also grateful to Olaf Koeneman, George Walkden, Stephane Goyette, and David Gil for their valuable comments, suggestions and criticisms.]


Thus far, this series of posts on the origins of Yiddish (see here, here, and here) has built up an argument against Paul Wexler’s radical view of Yiddish as a “re-lexified Slavic language”. While Slavic languages have had a monumental influence on Yiddish (more on which below), its vocabulary and even grammar are solidly Germanic. Nor is Wexler’s claim (cited in Noonan 2010: 56) that “Western and Eastern Yiddish are genetically unrelated, the former being a Germanic language” and the latter being a Slavic one, substantiated by historical data. These two claims—the Slavic origin of (Eastern) Yiddish and its lack of genetic relationship to Western Yiddish—are contradicted very strongly by Beatrice Santorini’s (1989) careful analysis of historical changes in word order, particularly in embedded clauses. As explained in more detail in my earlier post, the change from the auxiliary-final to verb-second (V2) word order in embedded clauses occurred in two stages, the first of which affected both Western and Eastern Yiddish. As the two varieties share a common innovation, the basic principles of historical linguistics dictate that they must be closely related. Furthermore, the second stage of this change, which affected only the Slavic-influenced Eastern Yiddish at the time when Ashkenazi Jews must have come into intense contact with Slavic-speaking peoples, is best understood as a “tweaking” of Germanic syntax by Slavic speakers rather than an adoption of Germanic grammatical elements into an otherwise solidly Slavic syntactic system.

Moreover, the major arguments that Wexler has put forward to support his Sorbian scenario have not been left unchallenged. Consider, for example, the argument based on the final devoicing—or rather lack thereof—in Yiddish. Philologos summarizes it thus:

“Already in early medieval German, before Yiddish split off from it as alleged by both the “standard” and “Regensburg” theories, a process took place whereby final voiced consonants like “b” and hard “g” were devoiced; hence, German weg, “way,” became pronounced as “vek,” and leib, “body,” as “leip.” In Yiddish, on the other hand, these words are pronounced “veg” and “leib” to this day. Since, Wexler reasoned, all the Slavic languages of Eastern Europe except Sorbian also have final devoicing, why would German-speaking Jews moving into Slavic-speaking areas have re-voiced such sounds? The answer he gives is that this never happened. Rather, as native Sorbian speakers who abandoned it for German, the first speakers of Yiddish continued to voice final consonants as they always had done.”


However, Wexler’s opponents Alice Faber and Robert King counter that this lack of final devoicing (i.e. pronouncing final consonants as voiced) is compatible with their Bavarian scenario. In particular, King (1980) argues that the lack of final devoicing is a consequence of the apocope rule that deleted word-final vowels (cf. Middle High German bluome ‘flower’ and Yiddish blum). Crucially, both the apocope rule and the lack of final devoicing are found in both Yiddish and Bavarian dialects of German.







Interestingly, several dialectal features have been identified as being shared by Yiddish and German dialects spoken in the same area as Sorbian (the so-called “East Central German dialects”, especially the Upper Saxon and Lower Silesian dialects). For example, Yiddish and East Central German dialects share the non‑affricate pronunciation corresponding to pf in many other German dialects, including Bavarian (and Standard German). Thus the pronunciation of the Yiddish epl ‘apple’, ferd ‘horse’, and kop ‘head’—compared to the Standard German Apfel, Pferd, and Kopf—places it on the same side of the so-called Speyerer Line (see Wikipedia map on the left) as the East Central German dialects. Note, however, that this particular feature may be used as an argument in support of Max Weinreich’s Rhineland scenario, as the German dialects spoken there are also on the same non-affricate side of this isogloss.

While a number of shared features have been identified between Yiddish and East Central German dialects (Jacobs 2005: 16-17 lists three such features), a greater number of commonalities emerge from a comparison of Yiddish with the Bavarian dialects of German (eight shared features in Jacobs’ list). Besides the abovementioned apocope rule and the lack of final devoicing, Yiddish and Bavarian dialects share “early and complete unrounding of front rounded vowels” (Jacobs 2005: 16), as in Yiddish ibər ‘over’ (cf. Middle High German über). Yet the same unrounding is also found in the Upper Saxon German dialect, which points to East Central German dialectal area as the Yiddish homeland. Moreover, commonalities such as these need not be shared innovations and may rather be changes that happened independently in different dialects of West Germanic. It is known that the same unrounding of front rounded vowels occurred in the ancestor of (Old) English, resulting in forms such as /fe:t/ (rather than /fø:t/), which eventually transformed into Modern English feet in the process known as the Great Vowel Shift. As the unrounded vowel is attested already in Old English, it is most likely that the unrounding affected its continental West Germanic ancestor. But that continental ancestor came from an area far from Bavaria, in fact from the Low German zone. Thus, it is clear that the unrounding must have happened at least twice in the history of West Germanic. What then prevents us from postulating independent unrounding for Yiddish and Bavarian?

Similarly, another major argument for the Bavarian scenario based on the diminutive -l is not unproblematic. Philologos writes:

“the ubiquitous Yiddish diminutive “l” -ending in words like meydl, a girl (from moyd, maid or young woman); yingl, a boy (from yung, a youth), and bisl, a little bit (from bis, a bite), characterizes Bavarian-Austrian dialects but not those of western Germany, where we find Mädchen, Jungchen and bisschen. It is hard to square things like this with the standard, Weinreichian view.”

Yet, it is not as hard as Philologos makes it sound: as we have seen with the agentive suffix –nik and the diminutive suffixes ‑tshik and ‑ke, borrowed by Yiddish from Russian, it is not unheard of for languages to borrow derivational morphemes. The presence of the diminutive –l in Yiddish shows a connection with Bavarian but not necessarily that of common descent. Nor does the fact that both Yiddish and Bavarian allow “double diminutives” (cf. Yiddish štot ‘city’, štetl ‘market town’, and štetələ ‘little market town’ [endearing]; Bavarian stuob ‘room’, stiable ‘small room’, stiabarl ‘very small room’) make a conclusive argument. In fact, many languages with rich systems of diminutive morphemes allow stacking them a certain way. Consider, for instance, Italian: pezzo means ‘piece’, pezzetto is ‘small piece’, and pezzettino is ‘(very) small piece’; likewise, boccia means ‘flask, decanter’, boccetta is ‘small bottle or flask’, and boccettina is ‘tiny bottle, vial’. Slavic languages allow such diminutive stacking as well (as mentioned by Jacob 2005: 17, fn. 11): as in Russian gorod ‘city’, gorodok ‘town’, and gorodoček ‘little town’ [endearing].

While the affinity of Yiddish and eastern German dialects appears more pronounced, Yiddish does have some Loter elements, albeit not numerous, as was recognized by Max Weinreich himself (see Jacobs 2005: 13-14). The only dialects that can safely be excluded from consideration are the Low German dialects of northern Germany and the Low Countries because (the German component of) Yiddish exhibits the Second Sound Shift, shared with High German but not with Low German. Compare the /ts/ in Standard German zwei and Standard Yiddish tsvey with /t/ in Dutch twee (and English two).

To sum up, from the early days of Yiddish study, scholars disagreed about whether it is an East Central or a Bavarian dialect that should be analyzed as the base for Yiddish: Gerzon (1902) took the former position, while Mieses (1924) followed the latter view. Yet, as was recognized by Noyekh Prilutski more than a century ago, “no Yiddish dialect corresponds to any single German dialect” (Prilutsky 1917, cited from Jacobs 2005: 16, highlight mine). Moreover, such fusion by a mixed Jewish language of features present in geographically non-overlapping non-Jewish dialects is not a peculiarity of Yiddish. For example, Judeo-Italian combines two features found in complementary sets of Italian dialects. The first feature is the lack of the masculine/feminine distinction in the plural and using the same definite article li for both masculine and feminine nouns (e.g. li donni ‘the ladies’ instead of le donne). This feature is attested only in the dialects of Italy’s deep south (e.g. in Calabria). Yet in Judeo-Italian it co-exists with the 7-vowel system (i.e., /a/, /ɛ/, /e/, /i/, /ç/, /o/, /u/), found in many non-Jewish dialects of central Italy (and in Standard Italian), but not in the south. Crucially, the combination of li donni and a 7-vowel system is not found in any non-Jewish Italian dialect. This fusion of dialectal features from different regions testifies to internal migrations and contact among Jewish communities all over Italy.


Going back to Yiddish, it appears to be a combination of Judeo-German dialects from the three areas proposed as its homeland (see the map on the left): Rhineland, Bavaria, and East-Central Germany. Therefore, as they say, “more research is needed” on the exact location of the Yiddish homeland. But it may still turn out to be an unanswerable question, as Yiddish may well have been born out of the fusion of several Judeo-German dialects. In at least one sense, such view of Yiddish as a mixed language rather than as a descendent of a particular historical dialect of German is part and parcel of Max Weinreich’s intellectual contribution rather than a contradiction to it. He saw Yiddish as a shmeltsshrakh (“fusion language”) par excellence, born out of a melding of the four main “components”: Germanic, Slavic, Hebrew-Aramaic, and Loez (Judeo-Romance). Consider, for instance, Weinreich’s famous example: noxn benčn hot der zejdə gəkojft a sejfər ‘after the blessing grandfather bought a holy book’, which contains a Loez-origin root benč- (cf. Latin benedicere ‘bless’) to which a Germanic suffix –n is attached, as well as Slavic-origin zejdə (cf. Russian ded ‘grandfather’) and Hebrew-origin sejfər (meaning ‘book, scroll’ in Hebrew but restricted to ‘Jewish holy book’ in Yiddish).

But it must be stressed that although lexical elements (roots, as well as bound morphemes) in Yiddish come from different components, its grammar is undeniably Germanic in its core. As discussed in the previous two posts (here and here), the Yiddish word order is Germanic, particularly its Verb-Second requirement, illustrated by the placement of the auxiliary hot ‘has’ right after noxn benčn ‘after the blessing’ and not after the subject der zejdə ‘the grandfather’. Another Germanic trait in Yiddish is its system of indefinite and definite articles, with definite articles expressing case. For instance, the definite article in der zejdə ‘the grandfather’ expresses nominative case (as opposed to accusative, dative, or genitive form dem zejdə). Slavic languages that Yiddish was in contact with—Polish, Ukrainian, Belarusian, and Russian—famously lack articles and express case by suffixes on the nouns themselves (e.g. Russian ded ‘grandfather.NOM’ vs. deda ‘grandfather.ACC’, etc.). Hebrew, in contrast, has a definite marker ha- but lacks indefinite articles; nonetheless, the Hebrew-derived sejfər ‘holy book’ in Yiddish appears with a Germanic indefinite article a.

The influence of other “component” languages, particularly Slavic, on the grammar of Yiddish remains an under-researched topic and a fruitful avenue for future study. As discussed in the previous post, the emergence of the embedded Verb-Second in the 1600s in Eastern Yiddish has been regarded by some as a result of the influence of Slavic languages (see Santorini 1989). Yet the question remains as to precisely what constructions in which languages served as the impetus for the embedded V2 in Yiddish. Beatrice Santorini points at the topicalization in Polish, while I point in the previous post to the so-called Generalized Inversion constructions (which I illustrate with Russian) as the model for the innovation that produced the embedded V2 in Yiddish.* Note, however, that both Santorini and I took a shortcut that is potentially misleading: we both looked at constructions in modern Slavic languages, whereas it is the older forms of these languages that should be examined to find the “culprit” for the embedded V2 in Yiddish. After all, Slavic languages have been changing since 1600 just as Yiddish has.

Another problem with most work on linguistic contact with Yiddish is that some languages in the region are almost entirely overlooked. Among such under-studied connections are those with Ukrainian and Lithuanian, both important languages in the heavily-Jewish area of Eastern Europe.** A couple of illustrations will suffice to show why much more attention needs to be paid to these languages and their contact with Yiddish before a full picture could emerge.

To start with Ukrainian, in addition to Generalized Inversion constructions parallel to those found in Russian (discussed in the previous post), Ukrainian has another structure with the appearance of (embedded) V2, known as the no/to-passive. In such sentences, an element that is not a nominative subject appears in front of an auxiliary that in turn occupies the second position. As can be seen from the following example (from Lavine 2004), such structures can be embedded in Ukrainian. (As before, the verb in the V2 position is in boldface, the element in the first position is underlined, and the embedded clause is in brackets. The negation marker is a clitic in Ukrainian, always immediately preceding the verb, hence it “doesn’t count” for purposes of establishing the second position.)***

(1) Modern Ukrainian

Ja spodivajusja, ščo [cej žart ne bude vykorystano “Pravdoj Ukrajiny”].

I hope that this joke.ACC not AUX.FUT used Pravda.INST Ukraine

‘I hope that this joke won’t be used by Ukrainian Pravda.’

However, Lavine (2004) shows that Old Ukrainian of the 17th century—when it might have served as the catalyst for the relevant change in Yiddish—was different in that the relevant construction lacked the auxiliary; moreover, the participle typically appeared at the end (unlike in modern Polish, where similar sentences likewise lack an auxiliary).

(2) Old Ukrainian (17th century)

Pavlusja u Varšavi stjato.

Pavlus’.ACC in Warsaw beheaded

‘Pavlus was beheaded in Warsaw.’

As for Lithuanian, the accepted wisdom is that its influences on Yiddish was merely lexical rather than grammatical (see Lemkhen 1995, Jacobs 2005: 22). Yet previous studies typically have looked for traces of exact copying of Lithuanian structures rather than of a more subtle “tweaking” of the sort I have discussed in the previous post. Modern Lithuanian has a construction, which James Lavine calls “Inferential Evidential” and which fits the Generalized Inversion profile: the verb (which shows no usual agreement with the subject) appears in the second position and the element that precedes it is not a nominative subject (see Lavine 1999, 2010 for details).****

(3) Modern Lithuanian

Ingos nuraminta vaikas.

Inga.GEN calm-down child.NOM

‘Inga must have calmed the child down.’

It remains to be seen, of course, whether this construction existed in its modern form in the 17th-century Lithuanian, where it could have affected (Eastern) Yiddish.

Given those productive avenues of research, I must agree with Philologos that “the questions of where and when Yiddish originated” are far from being “open-and-shut”. Further discoveries await!



*For linguists: the difference between Santorini’s proposal and mine reduces to whether the relevant construction involves A’-movement, in the case of topicalization, or A-movement, in the case of Generalized Inversion.

**Yet another language that must be included in the mix is Belarusian; a forthcoming post will discuss its history and current status.

***The placement of the auxiliary in the V2 position is confirmed by examples such as the following, where the auxiliary precedes a verb-phrase-modifying adverb pidpil′no ‘secretly’ (from Lavine & Freidin 2002: 278).

(i) Inozemcja bulo pidpil′no posadženo do v′jaznyci.

foreigner.ACC was secretly placed to prison

‘A foreigner was secretly put into prison.’

****For linguists: Lavine (1999, 2010) shows that the subject occupies an A-position (particularly, Spec-TP) and that the nominative element may not precede the ma-participle (examples (12) and (16c), respectively, in Lavine 1999). Furthermore, he argues that the ma-participle is a combination of the verbal root with a bound auxiliary –ma; in syntax, it is merely the auxiliary that is found in the V2 position, whereas it combines with the verbal root post-syntactically.




Gerzon, Jacob (1902) Die jüdisch-deutsche Sprache: Eine grammatisch-lexkalische Untersuchung ihres deutschen Grundbestandes. University of Heidelberg doctoral dissertation. Frankfurt am Main: J. Kauffmann.

Jacobs, Neil G. (2005) Yiddish. A Linguistic Introduction. Cambridge: Cambridge University Press.

King, Robert (1980) Final devoicing in Yiddish. In: Marvin I. Herzog, Barbara Kirshenblatt-Gimblett, and Dan Miron (eds.) The field of Yiddish. Studies of language, folklore, and literature. Fourth collection. Philadelphia, PA: Institute for the Study of Human Issues. Pp. 371-430.

Lavine, James E. (1999) Subject Properties and Ergativity in North Russian and Lithuanian. In Katarzyna Dziwirek, Herbert Coats, and Cynthia M. Vakareliyska (eds.) Annual Workshop on Formal Approaches to Slavic Linguistics. Ann Arbor, MI: Michigan Slavic Publications. Pp. 307-328.

Lavine, James E. (2004) The morphosyntax of Polish and Ukrainian –no/-to. Journal of Slavic Linguistics 12(1).

Lavine, James E. (2010) Mood and a Transitivity Restriction in Lithuanian: The Case of the Inferential Evidential. Baltic Linguistics 1: 115–142.

Lavine, James E. and Robert Freidin (2002) The Subject of Defective T(ense) in Slavic. Journal of Slavic Linguistics 10(1-2): 253-289.

Lemkhen, Khatskl (1995) Di hashpoe fun litvish oyfn yidishn dialekt in Lite. Oksforder Yidish 3: 6-130.

Mieses, Matthias (1924) Die juddische Sprache. Eine historische Grammatik des Idioms der intergralen Juden Ost- und Mitteleuropas. Berlin and Vienna: Benjamin Harz.

Noonan, Michael (2010) Genetic Classification and Language Contact. In: Raymond Hickey (ed.) The Handbook of Language Contact. Oxford: Wiley-Blackwell. Pp. 48-66.

Prilutsky, Noyekh (1917) Der yidisher konsonantizm 1 [= Yidishe dialektologishe forshungen 1.] Warsaw.

Santorini, Beatrice (1989) The Generalization of the Verb-Second Constraint in the History of Yiddish. Ph.D. dissertation, University of Pennsylvania.


Related Posts

Subscribe For Updates

We would love to have you back on Languages Of The World in the future. If you would like to receive updates of our newest posts, feel free to do so using any of your favorite methods below: