Quantitative similarities between the banjo and a diverse collection of West African lutes

The paper explores the African roots of the American banjo and seeks to clarify which West African lute traditions are most similar to the early gourd banjo. While the banjo’s Black origins and transmission via the Caribbean are well established, identifying specific African progenitors is complicated by centuries of cultural change and the creolized nature of the instrument. The banjo is a plucked spike lute with distinctive features (short drone string, membrane over a pot or gourd, full-spike neck) and early documentary evidence from the 17th–18th centuries confirms its early form. West African lutes share many traits (plucked playing, drone strings, skin membrane resonators), but also differ (e.g., unfretted round necks with hide tuning knots vs. early banjo’s flat neck with friction pegs; differing neck attachment methods). The author emphasizes that there is no single “banjo ancestor,” but identifying “banjo cousins” can illuminate the cultural and social contexts of origin. Prior scholarship highlights Senegambia—especially non-griot Bakic-speaking communities like the Jola ekonting—as a promising source of instrumental and stylistic parallels, while also noting possible influences from Niger Basin cultures. The study’s purpose is to quantitatively assess structural similarities between the gourd banjo and a broad set of West African lutes to evaluate these hypotheses.

The background synthesizes organological and ethnomusicological scholarship on West African lutes and the banjo’s development. Key works include Charry’s classification of West African plucked lutes and distinctions between spike vs. handle lutes, and ethnographic/organological observations connecting banjo features to Senegambian and Niger Basin traditions. The literature details early Western Hemisphere banjo depictions (Hans Sloane’s Jamaican “strum strump,” the Old Plantation watercolor, Haitian examples) and the banjo’s diffusion in North America, including cultural exchanges and the minstrel stage’s role. Comparative discussions address playing styles (notably down-stroking techniques in Senegambia similar to early banjo styles), social organization (griot vs. non-griot contexts), and the proposed “banjo genome” (Pestcoe & Adams) emphasizing informative character states for genealogical inference. The review positions Senegambia—particularly Bakic-speaking groups like the Jola ekonting—alongside Niger Basin lutes (e.g., Hausa) as likely influential, and frames quantitative cluster analysis as a needed enhancement over purely typological enumeration.

Data source and sampling: The study coded 62 instrument taxa (the gourd banjo plus 61 West African lutes) based on the comprehensive list and descriptions compiled by Pestcoe and Adams (2018c), itself built on Charry (1996) and extensive literature/personal communications. Instruments using modern or non-17th-century materials (e.g., tin cans) or pan-West African modern instruments were excluded. Instruments were grouped by the language families of the associated cultures (following Güldemann, 2018). Coding scheme: Six morphological/sociocultural characters were coded, allowing multistate entries when applicable: (1) player type (griot vs. non-griot), (2) bridge type (bipedal, block, cylindrical, fan), (3) resonator material/form (calabash, gourd, wooden), (4) neck type (full-spike vs. semi-spike), (5) neck-to-resonator attachment (over rim vs. penetrating body), and (6) number of strings (1–5). The gourd banjo was coded as non-griot player, bipedal bridge, gourd resonator, full-spike neck, neck penetrating the body, and four strings, reflecting early Caribbean and North American evidence. Analytical procedures: Character data were converted to Nexus format in R (ape) and a morphological Euclidean distance matrix was computed (Claddis). Two similarity approaches were implemented using phangorn: (a) UPGMA clustering to produce a dendrogram of nested similarities; (b) NeighborNet to construct a split network capturing conflicting signals and potential horizontal cultural transfer. Phenetic similarity on the NeighborNet was measured as minimum path distances along weighted edges derived by non-negative least squares.

• UPGMA dendrogram: The gourd banjo clusters as sister to six Senegambian lutes associated with Bakic language-speaking peoples—the bunchundo, busunde, ekonting, kisinta, kusunde, and nopata—which themselves form a zero-distance cluster. Shared traits with the Bakic cluster include non-griot player, bipedal bridge, gourd body, and full-spike neck; the banjo differs in string number and neck attachment method. The next-closest group includes instruments from diverse language groups (e.g., Omotic/Chadic speakers including Hausa, Kilba, Kotoko; Saharan Toubou; Adamawa Bana; Atlantic Fulbe). - NeighborNet network: The network indicates substantial reticulation, consistent with horizontal cultural transfer or unresolved dichotomies. Visually, the banjo groups near the same Bakic lutes as in UPGMA, but it is separated by a slightly smaller minimum phenetic distance from three Niger Basin lutes played by Omotic language speakers: Hausa gurmi (0.5353), Kotoko gulum (0.5759), and Kilba gullum (0.5964). The Bakic cluster’s minimum distances are slightly larger (0.6077). Additional proximities include Fulani-Wolof molo1 (0.6774); broader sets span multiple groups with greater distances. - Quantitative distances (Table 2, minimum NeighborNet path to banjo): gurmi 0.535263; gulum 0.575917; gullum 0.596424; Bakic set (bunchundo, busunde, ekonting, kisinta, kusunde, nopata) 0.607682; molo1 0.677421; further instruments show distances ≥0.8677 and many ≥0.9759.

The analyses support a strong resemblance between the gourd banjo and Bakic Senegambian folk lutes, particularly the Jola ekonting, aligning with previously noted similarities in construction and down-stroking technique. Simultaneously, the NeighborNet reveals equal or slightly closer phenetic proximity to certain Niger Basin lutes (Hausa, Kilba, Kotoko), highlighting heterogeneous cultural inputs in banjo formation. The reticulated network structure is consistent with expectations for material culture evolution in geographically proximate, interacting societies and argues against a simple bifurcating ancestry among lutes. Social implications emerge from the Bakic connection: Bakic societies are traditionally more egalitarian and lack a hereditary griot class, suggesting that their informal, accessible lute-playing contexts may have been especially compatible with the disrupted social environment of enslaved communities in the Caribbean. While Niger Basin traditions often exist within more stratified settings, Hausa lute practice can also be associated with less formal contexts, potentially informing aspects of banjo technique and construction (notably the neck penetrating the body). Overall, the findings refine the narrative from a search for a single ancestor to a mosaic of influences, with Bakic Senegambia and eastern Sahel/Niger Basin both salient.

This study quantitatively demonstrates that the early gourd banjo shares its closest cluster affinities with Bakic-speaking Senegambian folk lutes (e.g., Jola ekonting), while also exhibiting very close phenetic distances to certain Niger Basin lutes (Hausa, Kilba, Kotoko). These results reinforce the view of the banjo as a creole instrument shaped by multiple West African traditions and suggest that the egalitarian, non-griot musical cultures of Bakic societies may have been particularly impactful in the Caribbean context. The work contributes a replicable, data-driven framework (UPGMA and NeighborNet over a defined character matrix) for assessing instrument similarities. Future research should incorporate richer data on tuning systems, performance practice, and cultural meanings; expand instrument coverage as additional traditions are documented; and continue using network-based approaches to model horizontal cultural transmission across West African musical cultures.

• Data dependence on secondary compilations (Pestcoe & Adams) and literature-derived codings, which may omit undocumented instruments or under-report character variability. - Limited character set focused on structural features; excludes detailed performance practice, tuning systems, and social usage nuances that might alter similarity relationships. - Potential sampling bias due to exclusion of modern-material instruments and pan-regional modern forms; historical changes over three centuries complicate direct comparisons. - Network reticulation and conflicting splits indicate uncertainty in exact relationships; similarities do not equate to direct ancestry. - Language-group assignments approximate historical affinities but may not fully capture cultural exchange patterns relevant to instruments.