A Multilingual DOI-Based Model to Increase the Visibility of Latin American Research, a pilot observational study

This study combines a narrative literature review with a descriptive observational analysis of platform-level DOI resolution data. No experimental intervention was conducted; all data are derived from existing platform records.

We reviewed the bibliometric and scientometric literature on the relationship between publication language and research impact metrics, including citations, views, and database visibility. The literature synthesis draws primarily from a structured evidence report generated via semantic search across the Elicit platform, encompassing over 138 million academic papers indexed in Semantic Scholar and OpenAlex. Studies were included if they examined multilingual scientific journals or compared outcomes across different language publishing approaches using quantitative methods (bibliometric analysis, citation analysis, or usage metrics). Conference abstracts, editorials without data, and studies focused exclusively on translation quality rather than impact metrics were excluded. Ten empirical sources meeting inclusion criteria were identified and form the basis of the literature synthesis presented in the Introduction and Discussion sections.

ScienceBank (sciencebank.com) is an open-access journal launched in 2025. As of the time of analysis, the platform had published 21 articles: seven available in English and Spanish, one available in English, Spanish, and French, one available in English only, and the remainder in English only. Each language version carries an independent DOI registered with Crossref under the prefix 10.61340.

DOI resolution data were obtained from Crossref resolution reporting infrastructure. A DOI resolution is defined as a successful HTTP redirect from a DOI link to its target article landing page, and serves as a proxy for article access and discoverability. Monthly resolution counts were extracted for the period January 2025 through July 2026 (19 months). Per-article resolution data were available for selected high-performing articles identified by the platform. Given the platform's early stage of development, citation data were not available for analysis; ScienceBank articles had not yet accumulated citations in major indexing databases at the time of this study. DOI resolutions therefore represent the primary quantitative outcome measure, with the explicit acknowledgment that resolutions reflect access attempts and discoverability rather than scholarly citation impact. Resolution-success figures for February–July 2026 were not yet available from Crossref reporting infrastructure at the time of writing; resolution-attempt counts are reported for this period, consistent with the near-total attempt-to-success conversion (>97%) observed in every prior month of the study. Data analysis

DOI resolution data were analyzed descriptively. Monthly trends were examined to characterize platform growth trajectory over the observation period. For bilingual articles, resolution counts for each language version were compared to assess differential engagement across language audiences. Articles with notable resolution counts were examined individually to contextualize their content and likely audience.

Platform growth: DOI resolution trajectory (January 2025–July 2026) ScienceBank recorded 48 successful DOI resolutions in January 2025, its first operational month. Over the subsequent 19 months, total monthly resolution activity increased approximately 13-fold, reaching 647 resolution attempts by July 2026 (Table 1). The growth trajectory was non-linear across two distinct phases. The first phase (2025) was characterized by an early peak in July 2025 (410 resolutions), preceded by high activity in May 2025 (395 resolutions), followed by a mid-year trough in September–October 2025 (133 and 187 resolutions respectively) and a recovery from November 2025 onward. A second, larger growth phase began in February 2026 (526 resolutions) and continued through July 2026 (647 resolutions), surpassing the original 2025 peak and establishing a new platform high (Figure 1). Resolution successes closely tracked resolution attempts throughout the period for which both figures are available, indicating reliable DOI infrastructure.

Figure 1. Monthly Crossref DOI resolution data for ScienceBank (10.61340) from January 2025 through July 2026. High-performing article: Sustained engagement with regionally

The most accessed article on the platform across both analyzed periods (May 2025 and February 2026) was DOI 10.61340/RCCOOP, an English-language case study examining the frequency of candidate protective genetic variants against Alzheimer's, specifically REELIN-COLBOS (RELN H3447R) and APOE Christchurch (APOE R136S), in a cohort of 135 Colombian oldest-old individuals from the Resist AD Project. This article recorded 81 resolutions in May 2025 and 102 resolutions in February 2026, appearing among the top 10 most accessed items on both occasions. The sustained engagement across two distinct time periods, separated by nine months, indicates durable rather than transient interest in the article.

The topical profile of this article is analytically informative. It addresses a globally prominent scientific question (genetic resilience against Alzheimer's disease) through a regionally specific lens (Colombian populations with the PSEN1 E280A mutation, one of the largest familial AD cohorts in the world). This combination of universal scientific relevance coupled with regional data specificity appears to generate simultaneous interest from both international research networks and regional medical and scientific communities. Its English-language publication maximizes discoverability in international databases, while its Colombian focus sustains relevance within Latin American neuroscience networks.

Bilingual article: Near-parity of engagement across language versions Article 10.61340/OOPCPS (English version) and 10.61340/OOPCPS01 (Spanish version) represent the same work published with independent DOIs in both languages. Since publication in November 2025, both versions ranked among the ten most accessed items on the platform, receiving 23 and 21 resolutions respectively (Table 2). The difference of two resolutions between versions (a 9.5% relative difference) demonstrates near-equivalent engagement across language audiences, providing direct evidence that the independent-DOI multilingual model enables measurable parallel discovery by distinct linguistic communities.

The bibliometric evidence reviewed here converges on a finding that is both consistent and structurally important: English-language publications receive substantially more citations than non-English publications in major international databases. Studies across multiple disciplines and geographic contexts report a citation advantage of approximately 2-fold for English articles (Di Bitetti & Ferreras, 2017; Lira et al., 2013). In Brazilian ophthalmology journals, for example, English articles received a mean of 1.51 citations compared to 0.57 for Portuguese articles, with English articles being twice as likely to receive any citation at all (Lira et al., 2013).

However, interpreting this advantage as evidence of superior quality or impact would be a category error. The citation advantage of English is a structural artifact of database indexing, not a reflection of research quality. Web of Science and Scopus, the dominant citation indexes used in research evaluation worldwide, have historically favored English-language journals, creating a self-reinforcing cycle: English articles are more discoverable because they are better indexed, they accumulate more citations because they are more discoverable, and they appear more valuable because they have more citations (Moskaleva & Akoev, 2019). The role of non-English publications is systematically underestimated when assessment relies solely on these mainstream indexes, particularly in the social sciences and humanities (Moskaleva & Akoev, 2019).

The equity implications are significant. Non-native English speakers face higher participation barriers in the current publishing system, including cognitive burden, editing costs, and the structural disadvantage of publishing in a second language (Di Bitetti & Ferreras, 2017; González-Dambrauskas et al., 2024). Latin American researchers, in particular, face a compound disadvantage: language barriers compound funding inequalities that are already substantial (Turba et al., 2026; Lourenco et al., 2023). Multilingual publishing models that reduce the cost of participation for non-English-speaking researchers, and that create accessible pathways for regional scientific communities to engage with research findings, address an equity need that citation maximization strategies cannot. Context-dependence of language effects on visibility

The DOI resolution data from ScienceBank's first nineteen months illustrate a nuanced pattern that aligns with the bibliometric literature's central finding: language effects on research visibility are highly context-dependent. The platform's top-performing article (RCCOOP) is English-language, and its sustained high resolution counts across two distinct time periods suggest that English publication was instrumental in achieving broad discoverability. At the same time, this article's persistent appeal (returning to the top ten nine months after its first peak) reflects the regional specificity of its content rather than language alone: research on genetic protection against Alzheimer's disease in Colombian populations is directly relevant to Latin American neuroscience networks regardless of language.

This observation instantiates a pattern identified in the bibliometric literature: the citation advantage of English is mediated by audience size and database coverage rather than content quality. As Santos et al. (2021) demonstrated within the SciELO Citation Index, native-language publications maintain comparable or superior citation impact within their linguistic communities. Brazilian articles in Portuguese achieved 48.7% citation impact within the same language context, essentially matching the 48.5% impact of Brazilian English articles on Portuguese publications. This suggests that for research with inherently local or regional relevance, public health interventions, studies of locally prevalent diseases, and culturally specific phenomena, publishing in the national language may optimize impact on the most relevant stakeholders even when it reduces international citation counts. The bilingual article's near-parity in resolution counts (23 English vs. 21 Spanish) provides direct observational evidence for the parallel audience model. This finding supports the argument advanced by Kirillova et al. (2019): journals publishing parallel versions in both national languages and English can achieve citation levels comparable to major translated journals, particularly when technological infrastructure allows citations across versions to be tracked and merged. The independent-DOI architecture implemented by ScienceBank makes this tracking technically feasible at the article level for the first time in a platform designed specifically for Latin American research.

A recurring theme in the bibliometric literature, reinforced by the ScienceBank data, is that infrastructure, not reader preferences, is the primary determinant of language effects on citation impact. The finding that open access had a higher relative impact on views and citations for native-language publications than for English publications (Moskaleva & Akoev, 2019) is particularly instructive: when accessibility barriers are removed, non-English research finds its audience effectively. This implies that the systematic undercitation of non-English articles reflects discoverability constraints rather than indifference to non-English research. ScienceBank's growth from 48 to 647 monthly DOI resolution attempts over its first nineteen months, despite being an early-stage platform without major database indexing, is consistent with this interpretation. Rather than plateauing after its initial 2025 peak, the platform entered a second, larger acceleration phase beginning in February 2026, suggesting that early growth was not a one-time novelty effect but the leading edge of continuing demand. This trajectory reinforces the argument that genuine demand for accessible, regionally relevant scientific content is not fully met by existing publishing infrastructure, and that this demand has continued to build well beyond the platform's first year. The mid-year dip observed in September–October 2025 likely reflects seasonal patterns in academic activity and the platform's still-developing content volume at that time rather than a structural limit on engagement.

The relationship between journal ranking and language use identified in veterinary sciences, where quartile 1 journals published 99.2% of content in English versus 27.4% in quartile 4 journals (Krauskopf et al., 2017), illustrates the perverse incentive structure that multilingual journals face. Journals seeking higher impact factors are structurally incentivized to abandon multilingual publishing, creating a paradox in which the journals most visible to the international community are the least accessible to non-English-speaking communities. Platforms like ScienceBank, which are designed from inception as multilingual and open-access, are insulated from this incentive by their stated mission, but they face the corresponding challenge of achieving indexing in major databases to enhance the discoverability of their content.

Several limitations constrain the interpretation of the ScienceBank data. First, the absence of citation data, a consequence of the platform's early stage, limits the analysis to access-side metrics. DOI resolutions capture discoverability and access attempts but do not indicate whether articles were read, cited, or acted upon. Second, the platform's current article volume (21 publications) is insufficient for statistical analysis of language effects; the per-article findings presented here are descriptive and illustrative rather than inferential. Third, the absence of geographic data on resolution origin prevents direct assessment of whether Spanish-language resolutions originate primarily from Spanish-speaking regions, which would be the strongest evidence for the parallel audience hypothesis.

Fourth, ScienceBank is currently not indexed in Web of Science or Scopus, meaning that the resolution data capture organic discovery (direct links, social sharing, DOI-based access) rather than database-mediated discovery. Fifth, resolution-success figures for February–July 2026 were not yet available from Crossref reporting infrastructure at the time of writing, so the most recent six months of data are reported as attempts only; given that successes tracked attempts within 1–3% in every prior month, this is unlikely to materially change the growth trajectory reported here, but it should be confirmed once final figures are released. Future analyses with larger article volumes, geographic resolution data, and citation metrics will be required to test the hypotheses suggested by the current descriptive findings. Implications for multilingual publishing practice

The evidence synthesized here supports a framework in which language strategy should be matched to audience and purpose rather than optimized uniformly for citation maximization. For research with primary relevance to international scientific communities, English-language publishing with rigorous open-access provisions remains the most efficient strategy for achieving citation impact. For research with primary relevance to regional communities particularly in public health, clinical medicine, ecology, and social sciences, native-language publishing within regional indexing systems may optimize impact on the most relevant stakeholders. For research with both global relevance and regional significance, parallel multilingual publication with independent DOIs can capture benefits from both audiences simultaneously.

Realizing the equity potential of multilingual publishing at scale requires three complementary infrastructure investments. First, major citation databases should expand indexing of non-English and multilingual journals, particularly those serving the Global South. Second, funding agencies and research evaluation systems should develop frameworks that recognize local and regional citation impact alongside international citation metrics, preventing the systematic devaluation of research that serves non-English-speaking communities. Third, financial support for translation and parallel publication, identified as a significant resource constraint (Kirillova et al., 2019), should be incorporated into research funding structures, particularly for journals serving low-resource scientific communities. Conclusions

Multilingual publishing with independent DOIs for each language version, as implemented by ScienceBank, has the potential to reduce language-related inequities in scientific communication while improving the discoverability of research from Spanish-speaking regions. The first nineteen months of ScienceBank's operation demonstrate sustained and, in recent months, accelerating growth in platform engagement, and provide preliminary evidence that Spanish and English versions of the same article can achieve near-equivalent access levels when each version is independently discoverable via its own DOI.

The broader bibliometric literature confirms that multilingual publishing does not universally drive higher citation counts compared to English-only publishing; rather, parallel versions capture distinct benefits from international databases (where English dominates) and regional networks (where native languages maintain strong citation impact). The effectiveness of any given language strategy is therefore highly dependent on target audiences and indexing infrastructure.

These findings reinforce the argument that scientific equity and citation maximization are distinct (though not incompatible) goals. The current publishing system rewards English-language publication regardless of audience appropriateness, creating systematic disadvantages for researchers who serve non-English communities or study locally relevant phenomena. Addressing this inequity requires not only platforms willing to publish multilingually, but infrastructure investments in multilingual indexing, evaluation reforms that recognize local impact, and resource support for parallel publishing. ScienceBank's continued growth through its first nineteen months, including a renewed acceleration well beyond its initial launch period, represents one concrete and technically tractable step in this direction. Disclaimers

The author serves as the Head of Publishing at ScienceBank. To support bibliographic queries, Elicit AI was utilized. This editorial is a preprint and has not undergone peer-review, post-publication review is encouraged.