Two Peoples, One Land
Disputes

The pre-modern population of Karabakh

Whose people were there first, and what was the demographic balance before Russian conquest?

History all versions

The pre-modern Karabakh dispute turns on scale. Armenian historiography foregrounds the Armenian-majority highland core, the medieval melikates, monasteries and ecclesiastical institutions of Artsakh. Azerbaijani historiography foregrounds the wider Karabakh khanate, where Russian-era surveys record a Muslim majority and where Turkic-speaking Muslim political authority was real before Russian annexation. The strongest academic reading keeps both truths in frame: highland Karabakh had deep and continuous Armenian institutional life, while lowland Karabakh and the khanate as a whole were demographically and politically mixed, with a Muslim majority in the broader unit. The dispute becomes deceptive when either side changes the scale without saying so. Mountain Karabakh, Karabakh khanate, later NKAO and modern Azerbaijani administrative districts are not interchangeable maps.

Perspectives (3)

How to read this section
Each position is laid out in its own voice first, the way its proponents argue it. Where that argument relies on omitted facts, logical fallacies, or recognised state-propaganda techniques, those are noted in a separate Critique block under the position. The intent is not to suppress any view, but to show what each side asserts and where its case is weaker than the assertion makes it sound.
state-armenia
Armenian state position: an unbroken Armenian land

The Armenian position emphasises Artsakh as one of the historic Armenian provinces, with medieval churches, manuscript culture, noble melik houses and continuous Armenian ecclesiastical jurisdiction. This view is strongest when speaking about the mountainous core and weaker when it treats all of the wider Karabakh khanate as if it had the same demographic profile. Its historical claim is cultural-continuity based: Armenian life in the highlands was not a nineteenth-century invention, and the Armenian built landscape is not plausibly reducible to later migration.

The full position internal divisions, supporting actors, reception, daily reality — click to collapse

Internal divisions

The position is not unified across Armenian institutions. The strongest version comes from the diasporic ARF Dashnaktsutyun tradition, which integrates Artsakh continuity into a wider Hai Dat programme that includes Sèvres claims and Genocide recognition. The post-Soviet Yerevan establishment treats the Karabakh question more pragmatically: ter petrosyan argued in "War or Peace?" for phased compromise that would not require maximalist claims of perpetual Armenian ownership. The Republican Party under kocharyan and sargsyan (both Karabakh-born) ran the maximalist line through the 2010s. The 2018 Velvet Revolution reversed this; pashinyan in October 2022 recognised Azerbaijan's territorial integrity including Karabakh, splitting the establishment from the Karabakh Committee tradition that founded post-Soviet Armenia.

How prominent figures argue this

kocharyan, who served as president of the Republic of Artsakh before the Republic of Armenia, frames Karabakh as a continuous Armenian polity from the medieval melikates onward. The diaspora frame is more historicist: it stresses the Dadivank, Gandzasar and Amaras monastic complexes, the manuscript culture documented by Hewsen, and the inscriptional record. pashinyan does not contest the historical claim but treats it as analytically separable from contemporary territorial sovereignty editorial.

Carriers

The argument is carried by the Artsakh government in exile (formally dissolved 1 January 2024 but with a continuing diaspora presence), ARF organisations in paris, washington and Beirut, and Armenian Apostolic Church institutions in etchmiadzin. State media in Yerevan now treats the historical claim with deliberate distance.

Reception

Public sentiment in Armenia remained committed to Karabakh through 2020. After the 44-day war and the 2023 exodus, polling moved sharply toward acceptance that Karabakh is lost as a political project even where the historical claim retains affective force. The roughly 100,000 displaced Karabakh Armenians integrated into Armenia constitute a constituency that contests Pashinyan's position. sourced opinion

Daily reality

The Armenian state since 2023 has avoided official territorial maps that include Karabakh. Schoolbooks have been revised. Place-names of formerly Armenian-held towns no longer appear on official meteorological reports. The diaspora continues commemorations: 26 February for Sumgait, 9 May for the 1992 capture of Shusha, 19–20 September for the 2023 operation.

Statistics

Pre-1988 NKAO Armenian share: ~76% in 1979, ~77% in the 1989 census. Post-2023 Armenian residents in Karabakh: 14, per UNHCR registration. Refugees registered into Armenia September–October 2023: 100,617.

Critique

The weakness of maximal Armenian framing is scale drift. Evidence for Armenian continuity in the mountains does not prove Armenian demographic predominance in every lowland district of the khanate.

Sources: Robert H. Hewsen, 2001 · Razmik Panossian, 2006
state-azerbaijan
Azerbaijani state position: heritage as "Caucasian Albanian"

The Azerbaijani state position stresses the Muslim-majority Karabakh khanate and the political authority of Turkic Muslim khans before Russian annexation. It also often reattributes Armenian churches and inscriptions to Caucasian Albanian heritage, presenting Armenians as later arrivals whose claims were amplified by Russian resettlement. The demographic point about the khanate as a whole has real evidentiary basis in Russian surveys and court histories.

The full position internal divisions, supporting actors, reception, daily reality — click to collapse

Internal divisions

The Azerbaijani position is institutionally homogeneous in public but has internal variation that rarely surfaces. Independent Azerbaijani historians (Altay Goyushov, the Eldar Zeynalov circle) accept much of the Bournoutian reconstruction in private discussions but cannot publish openly under YAP state-media conditions. The official line was consolidated under heydar aliyev from 1993, building on Soviet Azerbaijani historiography of the 1950s–70s pioneered by Ziya Bunyadov, who developed the Caucasian Albanian reattribution thesis. ilham aliyev amplified the line through statements at heritage-restoration ceremonies in shusha post-2020.

How prominent figures argue this

ilham aliyev in his 2021 Shusha "cultural capital" addresses claimed that the medieval Armenian inscriptions of Karabakh "must be objectively studied" and that "much of what is presented as Armenian heritage is in fact the heritage of Caucasian Albania". The state-funded Institute of History under the Azerbaijani National Academy of Sciences carries the academic version. Diaspora amplification comes through Turkish presidential addresses, including ilham aliyev joint appearances with Erdoğan invoking the "one nation, two states" frame.

Carriers

The thesis is carried by the National Academy of Sciences, the Baku state museums (the Heydar Aliyev Centre, the National Museum of History), Azerbaijani schoolbooks (mandatory from grade 7), the Azerbaijani Diplomatic Academy, and pro-state media (Trend, Azertac). The post-2020 reconstruction in Agdam, Fuzuli and Shusha features new museums staffed with curators trained in this reading.

Reception

Domestic reception is high. Independent polling is suppressed but state-media penetration is near-total. Internal academic dissent exists in private but is professionally costly. Diaspora reception varies: Turkic-world publics accept the frame; Iranian-Azerbaijani communities are more sceptical given their access to Persian-language sources that complicate the Bunyadov thesis. sourced opinion

Daily reality

The "Albanianisation" thesis has direct material consequences. Dadivank Monastery under post-2020 Azerbaijani control is signposted as a "Caucasian Albanian" site rather than an Armenian one. Gandzasar receives similar treatment. The Tsitsernavank basilica is being restored under Albanological narration. The Julfa khachkar cemetery in nakhichevan was destroyed between 1998 and 2005 under the same intellectual umbrella, documented by Caucasus Heritage Watch and the AAAS satellite study.

Statistics

Karabakh khanate ~1823: ~22% Armenian, ~78% Muslim per the 1823 Russian survey. Highland Karabakh (the "five melikates"): ~85% Armenian. Heritage destruction in nakhichevan 1980–2005: ~89 churches and ~5,840 khachkars per Ayvazyan's inventory cross-referenced with current satellite imagery.

Tensions and recent shifts

The post-2020 retake of Karabakh allowed the doctrine to move from rhetoric to administration. The full destruction or restoration-under-altered-attribution of Armenian heritage is now the test of whether Albanological reattribution is academic theory or operational policy. editorial

Critique

The weakness is overextension. A Muslim-majority khanate does not erase Armenian-majority highland districts, Armenian inscriptions, ecclesiastical continuity or the medieval Armenian melikates.

Sources: Thomas de Waal, 2003
academic-consensus
Academic consensus: both peoples present, with regional variation

The academic consensus is mixed rather than symmetrical. George Bournoutian and Thomas de Waal both show that the wider Karabakh khanate was majority Muslim in early Russian records, while the mountainous Armenian core had continuous Armenian settlement and institutions. The best formulation is therefore regional: lowland and urban-political Karabakh were strongly Muslim/Turkic-Persianate; highland Karabakh was overwhelmingly Armenian in many districts. Later nationalism compresses these layers into a single ownership claim.

The full position internal divisions, supporting actors, reception, daily reality — click to collapse

Internal divisions

The "consensus" is a synthesis across several scholarly traditions. Bournoutian writes from the Iranian-American historiographical tradition (translation, Persian sources, Russian khanate-era surveys). de Waal writes from the journalistic-political tradition (interviews, archives, post-1988 fieldwork). Hewsen is more Armenian-leaning but methodologically rigorous. Saparov reconstructs Soviet decision-making. They differ on emphasis, particularly on how to weigh the seventeenth-century Shah Abbas relocations and the 1828 Turkmenchay migration, but converge on the regional-stratification picture.

How prominent figures argue this

The position is academic rather than political, so its "figures" are scholars. Bournoutian's translations of Mirza Jamal Javanshir's Tarikh-e Qarabagh established that the khanate's Persian-language court historians themselves recorded Armenian melikate autonomy, undermining the strongest forms of both maximalist readings. Hewsen's Atlas maps district-by-district demography. Outside the Caucasus field, comparativist genocide and ethnic-conflict scholars (Stuart Kaufman, Donald Bloxham) draw on this base for their own work.

Carriers

Academic carriers: Princeton University Press, NYU Press, Oxford UP, Columbia UP, Berghahn, Routledge. Conferences: ASN (Association for the Study of Nationalities), AAS, AAASS. Online platforms: Eurasianet, OC-Media, Civilnet (the latter two with NGO funding). State institutions in both armenia and azerbaijan cite the academic record selectively sourced opinion.

Reception

The academic consensus is the framing most quoted by International Crisis Group, HRW, and the European Parliament in their reports. It is rarely the dominant frame in either Armenian or Azerbaijani domestic media; both states prefer their own historiographies. Western diplomatic actors lean on it because it permits even-handed framing. Russian and Turkish state media largely ignore it.

Daily reality

The academic consensus shapes how international NGOs phrase their interventions, how the OSCE Minsk Group formulated its principles, and how the ICJ frames historical context in proceedings. Outside elite policy and academic spaces, the consensus has limited cultural reach in the region. Both publics receive it through partisan reframings. editorial

Statistics where relevant

The 1823 Russian survey of Karabakh khanate gives ~22% Armenian / ~78% Muslim overall. The five highland melikates record ~85% Armenian. The 1832 imperial survey of the Erivan khanate (post-Turkmenchay) shows ~50% Armenian after migration, up from ~20–25% pre-conquest. The 1897 census gave Erivan governorate ~53% Armenian. NKAO 1923 founding: ~94% Armenian. NKAO 1989: ~77% Armenian.

Tensions and recent shifts

Post-2020, academic debate has moved toward documenting the destruction of the historical record itself: heritage erasure, archival inaccessibility, the closure of pre-2020 monitoring. The consensus's weakness is that it cannot enforce itself; both state historiographies operate inside their educational systems regardless of what scholars find. editorial

Critique

The historical record supports presence and rootedness for both communities, but not the same kind of claim in every subregion.

Sources: George A. Bournoutian (trans.), 1994 · Thomas de Waal, 2003 · Robert H. Hewsen, 2001