Sweden's project to introduce mandatory language requirements for foreign nationals seeking citizenship is facing a five-year timeline, with the Swedish Council for Higher Education proposing to push back the first reading and listening examinations until autumn 2029, followed by speaking and writing tests in 2030. Eighteen months after Stockholm University was commissioned to build the assessment framework, the newly appointed project manager has revealed that the task of crafting actual test items is starting more or less from scratch.
Lars Bokander took a one-year leave of absence from his position as an assistant professor at Jönköping University to assume full-time leadership of the development team on November 1st. He is tasked with overseeing the creation of four distinct examinations covering reading, writing, listening, and speaking. He tells The Local that while his academic background is in second language acquisition and test development, this is the first time he has been involved in a large-scale, high-stakes test development project (he has constructed two published vocabulary tests for research purposes).
Highlighting the broader context in Sweden, Bokander noted that the country lacks a deep tradition of high-stakes testing comparable to nations like the United Kingdom. Existing language assessments, such as national tests in compulsory and upper secondary schools, tests for Swedish for Immigrants (SFI), and the Tisus exam—which qualifies non-native speakers for university-level studies in Swedish and is also overseen by Stockholm University—carry significantly lower stakes because final grades do not depend exclusively on single test outcomes.
Since receiving the mandate from the Swedish Council for Higher Education in May 2025, Stockholm University spent its initial 18 months establishing an organizational structure, hiring and training core personnel, performing needs analyses, reviewing testing models used in other countries, and drafting preliminary item ideas. The first eight months of the current year were largely focused on staff recruitment, training, and securing initial participant pools for preliminary testing.
By the end of this year, Bokander and his team are scheduled to finalize the scope of what the examinations must measure, produce and evaluate prototype questions, establish preliminary test specifications, and continue recruiting test subjects. Bokander identified candidate recruitment as one of the project's primary hurdles, explaining that convincing hundreds or potentially more than a thousand individuals to participate in time-consuming pilot examinations requires substantial effort.
The extended timeline proposed in October by the Swedish Council for Higher Education means that more than five years will elapse between the initial May 2025 mandate and the rollout of the first exams. Bokander attributed the initial slow pace to institutional constraints, explaining that university personnel were already committed to teaching and research responsibilities when the government order arrived, making an immediate start difficult.
Addressing concerns over the duration, Bokander emphasized that ensuring adequate quality for a high-stakes assessment inherently requires extensive time. He noted that while a quick test could easily be written overnight, it would lack scientific credibility. To be effective, the assessment must fulfill standards of reliability, validity, and fairness, ensuring that candidates would achieve consistent scores on different days and that test results accurately reflect real-world language capabilities.
The design framework relies on the validity framework theory formulated by academic Michael Kane. Under this approach, the tests must accurately measure target proficiency levels—specifically level A2 for speaking and writing, and level B1 for reading and listening—while delivering beneficial outcomes for both test-takers and society. Bokander argued that the only way to satisfy these requirements is through repeated rounds of pilot testing across representative sample groups.
Stockholm University's project roadmap incorporates two separate pilot testing phases, planned for April–May 2027 and April–May 2028. Following the 2027 pilot, the university has allocated a full year to evaluate listening and reading data, construct new test items based on findings, and begin drafting and piloting the speaking and writing components. After the second pilot in 2028, five additional months will be dedicated to fine-tuning questions, bringing the materials to completion by October 2028. This overall timeframe aligns with estimates set out in the original 2021 Swedish government inquiry into citizenship testing systems.
The additional gap between finalizing the test structure in late 2028 and launching live exams in 2029 and 2030 stems from the decision to administer all tests digitally from the outset. Research indicates that transitioning paper assessments to digital platforms alters test characteristics, effectively creating a new test that requires separate validation. Building a digital-first model avoids duplicate development while accommodating listening components, which require seamless audio playback at examination sites.
The five-year development phase means that thousands of citizenship applicants will face delays, with those failing or delayed potentially sent back to the start of Sweden's four-year residency queue for citizenship. Wouldn't it be better to have a less valid test faster? "As a private person," he adds, "I can perfectly understand those arguments. Waiting five years sounds ridiculous. But in my job as a test developer, I have to follow best practice. We cannot just begin with a sloppy test, because we would be deciding who gets citizenship based on weak and unreliable data." While acknowledging that a five-year wait sounds unreasonable from an applicant's perspective, Bokander stressed that test developers must adhere to strict scientific standards. He noted that while the government could have purchased an off-the-shelf test of uncertain quality from a private provider, it chose to commission a dedicated framework to prevent citizenship decisions from relying on weak or unreliable data.
"The biggest challenge is finding enough people willing to do the piloting, because that takes a lot of time and you have to convince them that it's valuable for them somehow," Bokander says. "We're talking about hundreds or perhaps more than a thousand people."
