Skip to content
JewishHubAzerbaijan

Research

The Baku Model

An examination of what published evidence a conversation about interconfessional coexistence in Azerbaijan can rest on — and where that evidence does not exist.

The Baku Model: what the published record can and cannot support about interconfessional coexistence

JewishHub Azerbaijan · Updated: 19 September 2026

Abstract

Azerbaijan is often cited as a case of untroubled coexistence between a Muslim majority and a Jewish community that has lived there for centuries. This paper neither verifies nor refutes that claim: it establishes what published evidence the claim can rest on at all. Three series are examined — the two-rung social-distance measure of the Caucasus Barometer (CRRC) for Azerbaijan, 2009–2013; the Jewish population of Baku across the censuses of 1897–1989; and published estimates of the number of Juhuri speakers. The conclusion is cautious and has three parts. First, on this evidence willingness to do business with Jews rose while willingness to intermarry stayed in the low single digits: the two measures diverge, and “tolerance” without naming the rung is too broad a word to describe anything. Second, the series stops in 2013 and there is nothing to continue it with. Third, for the language there is no such series at all: published estimates differ by more than threefold, and census data on mother tongue are absent from the available sources entirely. The paper's principal result is a map of what is not known.

Methodology

Design

The work is a secondary analysis: it contains no survey, interview or field observation of its own, and not one figure was produced by this project. The choice is deliberate. The project is a digital archive, not a research institute; entering the sociological field without a sampling frame, ethical review and trained interviewers would mean producing data that cannot be trusted. Instead, published series are assembled and each is described in detail: what it measures and what it does not.

Source selection

A series was included if three conditions held: it has a known publisher, every figure carries a base year, and the primary source is reachable at a stable URL. What the publisher says about its own method was recorded separately. This is why the paper contains no indicator drawn from tourism or overview material, even though such material is generous with percentages.

Handling disagreement

Disagreement between sources is not resolved by picking the more authoritative one. Both figures stay in the dataset and both are plotted; the caption says who publishes what, and for which year. This applies also where a page contradicts itself: the box figure and the body figure appear side by side. The reader should see the dispute, not our preferred outcome of it.

Limits of inference

An attitude survey measures what people tell an interviewer, not what they do. A census counts those who declared themselves so, not those who are. An estimate of speaker numbers without a stated method is an expert judgement, not a measurement. None of these series licenses a claim about causes: the rise in willingness to do business between 2011 and 2012 may reflect a change in views or a change in the sample. The words “shows that” do not appear in this paper.

Social distance: two rungs

CRRC's Caucasus Barometer asked two questions in Azerbaijan that form rungs of a social-distance scale. The wording is quoted verbatim from the codebook.

  • BUSINJEW

    “Would you approve or disapprove of people of your ethnicity doing business with Jews?”

  • MARWJEW

    “Would you approve or disapprove of women of your ethnicity marrying Jews?”

Fig. 1. Share approving, Azerbaijan, 2009–2013
Doing businessMarriage

Source: CRRC, Caucasus Barometer, variables BUSINJEW and MARWJEW. Percentages as published; they may not sum to 100 because of rounding. The questions were not asked in 2008, and the survey was not fielded in Azerbaijan after 2013.

Limits of these data

  • There is no friendship rung in the questionnaire. A classic Bogardus scale has seven rungs; there are two here, so calling this a Bogardus scale is only admissible with that caveat — these are two separate measures of social distance, not a continuous scale.
  • The marriage question is asked only about women of the respondent's own ethnicity. The dataset contains no symmetrical question about men, so the figure reflects attitudes to women's marriage, not to marriage in general.
  • The series stops in 2013. These questions were not asked in 2008, and after 2013 CRRC stopped fielding the Caucasus Barometer in Azerbaijan. There is nothing to continue the line with.
  • The 2011→2012 jump on the business rung (39 % to 59 %) is large and the source offers no explanation for it. It may be a real change; it may be a change in sampling or question order. Without CRRC's methodological reports for those waves, nothing can be asserted.

What the data show

The two rungs diverge

Across five waves the share approving business dealings with Jews rose from 33 % to 60 %, while the share approving marriage stayed between 3 % and 7 % and ended lower than it began. The distance between the rungs is not an incidental detail but the main thing visible in these data: consent to deal and consent to marry move independently. The summarising word “tolerance” conceals exactly that difference.

The community was counted twice over

The 1989 census returned 22,836 Jews and 8,438 Tats for Baku — in practice Mountain Jews, recorded in a separate column. Anyone building a series from these censuses must decide whether to add them, and that decision moves the picture by a quarter. Here the figures are added and the split is shown in the breakdown, because the split is itself a historical fact and not an error term.

About the language we do not know the main thing

How many people speak Juhuri today is unknown. Published estimates range from 30,000 to over a hundred thousand; some aggregate country figures with base years from 1989 to 2010, and community size is confused with speaker numbers. Something else, however, is well documented: two changes of script in nine years, the absence of schools teaching in the language, and a single place, Qırmızı Qəsəbə, where it is still passed to children. The mechanism is known; the magnitude is not.

The Jewish population of Baku, by census

The census series 1897–1989 and one estimate for 1994. The figures appear as the Short Jewish Encyclopedia gives them.

Fig. 3. Jews of Baku by census, 1897–1994
CensusEstimate

Source: the Short Jewish Encyclopedia. For 1989 two census columns — “Jews” and “Tats” — are added together; that is exactly how Soviet statistics split one community. The 1994 figure is a Jewish Agency estimate, not a census, and is drawn hatched.

Emigration to Israel, people per year

1989
385
1990
7,371
1991
5,283
1992
2,520
1993
3,162

Between 1989 and 2003, 29,565 people emigrated to Israel.

Juhuri: what we do not know

Published estimates of speaker numbers differ by more than threefold. What follows is not a trend but a spread: every point is labelled with its publisher and base year.

Fig. 2. Published estimates: how many people speak Juhuri
SpeakersCommunity size

Each row is a separate publication. Speaker estimates and community size count different things and are kept apart. Two figures contradict themselves within a single page; both are retained.

Limits of these data

  • None of these figures comes from a census. No consulted source gives Soviet census data on the mother tongue of Tats and Mountain Jews — the one series from which a real trend could be built.
  • The spread reaches more than threefold between reputable publishers, and two figures contradict themselves within a single page. This is a chart of the state of knowledge, not of the number of speakers.
  • These are quantities of different kinds. Speaker estimates and community size count different things; they cannot be mixed in one series, so the chart keeps them apart.

What is reliably documented about the language

  1. 1908–1909
    The first books in Juhuri

    The first two books appear: *Mətləb siynigo* (a translation of Yosef Sapir's Russian pamphlet *Zionism*, 1903) and the prayer book *Qol tefila* with a parallel Juhuri translation by Asaf Pinhasov (1884–1920).

  2. 1915–1916
    The first newspaper — in Baku

    Several issues of the first Juhuri newspaper appear in Baku. From 1927 the centre of publishing in the language moves to Dagestan, and the Derbent dialect rather than the Quba one becomes the basis of the literary norm.

  3. 1929
    The switch to the Latin alphabet

    The decision is taken to move the script from the Hebrew alphabet to the Latin one. Until then Juhuri was written in semi-cursive Hebrew letters.

  4. 1938
    Cyrillic replaces Latin

    Cyrillic replaces Latin — the second change of script in nine years. In the same year the constitution of the Dagestan ASSR declared Tat (in practice Juhuri) one of the republic's ten official languages.

  5. 1950-е
    The shift to Russian

    On M. Zand's account, already in the 1950s many Mountain Jews were fluent in both Juhuri and Russian, or knew the community language only as a second one. There were no Juhuri-medium schools and nothing to make up the gap; the next generation produced a great many semi-speakers — who understand but barely speak.

  6. 2010
    The UNESCO Atlas — and a dispute over wording

    The Atlas of the World's Languages in Danger includes Juhuri. Exactly how is reported differently: Wikipedia writes “definitely endangered”, the Jewish Language Project simply “endangered”, adding its own assessment that for most locations the position lies between definitely and severely. The exception is Qırmızı Qəsəbə near Quba, which the same page rates “vulnerable” because the language is still passed from parents to children there.

What needs checking next

The list is made of specific documents and databases this work could not reach. It is the section's principal result.

  1. Soviet census data for 1926–1989 on the mother tongue of Tats and Mountain Jews in Azerbaijan. This is the only series from which a real generational trend could be built; it exists in archives and is absent from every secondary source consulted.

  2. CRRC's methodological reports for the 2011 and 2012 waves — without them the jump on the business rung cannot be interpreted.

  3. The record for Baku's Great Choral Synagogue in the database of the Center for Jewish Art at the Hebrew University of Jerusalem: it holds catalogue fields present in no source available to us and should settle the dispute over the construction date. It is unreachable from our environment.

  4. The ethnographic work of I. Sh. Anisimov and of Liya Mikdash-Shamailov on the daily life of the Caucasian Mountain Jews — the whole cuisine section, which currently rests on cookery sites, should be checked against them.

How to cite this work

The four formats are generated from one description, so they cannot drift apart.

JewishHub Azerbaijan. (2026). The Baku Model: what the published record can and cannot support about interconfessional coexistence. JewishHub Azerbaijan. https://jewaz.org/en/research/

Data sources