Language barriers in remote work:
the problem that went global.

Remote work promised to make geography irrelevant. And it did — for geography. Language is a different problem. For the hundreds of millions of people working daily in a language that is not their own, remote work didn’t remove the barrier. It amplified it, relocated it into every chat thread and video call, and made it harder to see.

Before remote work, language friction in the workplace was at least partially visible. An office with a significant non-native-speaker population developed informal accommodations over time — a colleague who helped bridge misunderstandings, meetings with more patience for repetition, the side conversations in common languages that happened in the kitchen or the hallway. These compensating mechanisms were imperfect and insufficient, but they existed.

Remote work eliminated them. What remains is predominantly two channels: asynchronous text communication, where the written word must carry all the nuance that gesture, tone, and physical presence once helped convey; and video calls, where the cognitive demands of real-time second-language processing are unchanged but the human context that made those demands manageable has been stripped away. The language barrier in remote work is the same barrier it has always been, in a context where the compensating mechanisms have been removed.

The async channel and its tradeoffs

Text-based asynchronous communication is, in one sense, more hospitable to non-native speakers than in-person or real-time settings. Writing allows time for composition, revision, and care. A non-native speaker can formulate ideas in their first language, translate deliberately, review before sending, and produce something that reflects their actual thinking rather than the compressed, simplified version that real-time spoken English demands.

In practice, though, async communication introduces its own compounding of language friction rather than resolving it.

The first issue is tone. Written English is highly susceptible to tone misreading, and non-native speakers — who may have learned English primarily in formal or academic contexts — often produce text that is technically correct but tonally ambiguous to native speakers. A message that is intended as a direct, efficient response can read as curt or dismissive. A message intended as collaborative and tentative can read as passive-aggressive or evasive. These misreadings accumulate over weeks and months of daily async communication, shaping perceptions of colleagues who have no way of knowing the misreading is occurring.

The second issue is the resolution cycle. In-person or real-time settings allow misunderstandings to be corrected within seconds — a clarifying question, a facial expression that signals confusion, a colleague who steps in to rephrase. In async communication across time zones, the same misunderstanding may cost 24 hours or more to surface, and longer still to resolve. For multilingual teams where language compounds the ambiguity, this cycle is longer, more frequent, and more damaging to project velocity.

The video call problem

If async text is the medium where language friction is diffuse and slow-burning, video calls are where it is acute and consequential. Video calls have become the primary venue for the decisions that matter most — strategy discussions, project reviews, conflict resolution, hiring decisions, performance conversations. They are, in the remote work context, what meetings are in the office context: the moment where organizational influence is distributed and where the gap between what you can think and what you can communicate in real time determines your standing.

For non-native speakers, the video call replicates all of the dynamics that make in-person meetings difficult, with additional disadvantages. The compressed bandwidth of video — the slight audio delay, the pixelated facial expressions, the inability to read a room — requires even more linguistic precision to compensate for the lost social context. The meeting pace is set by participants for whom language is not an additional cognitive load. The cultural norms of participation — when to interject, when to hold back, how to express disagreement politely in this particular team’s culture — are learned through close social observation that is harder to perform remotely and across cultural lines simultaneously.

The result is a dynamic that most remote-first teams have noticed but rarely named: non-native speakers tend to contribute less in video calls than their native-speaker counterparts, and their contributions tend to be shorter, less elaborate, and less likely to involve direct disagreement or challenge — precisely the forms of contribution that are most valuable in high-stakes discussions.

“When I am in a video call and someone says something I think is wrong, I need maybe 15 seconds to formulate the objection clearly in English. In that 15 seconds, the conversation has moved on. By the time I would speak, it feels too late.” — A software engineer at a distributed tech company, reflecting on English-only video meetings.

The async-first movement and its limits

The most thoughtful remote-first companies have responded to the video call problem by embracing async-first culture: defaulting to written communication, minimizing synchronous meetings, and using long-form writing to make decisions that would otherwise happen in calls. This is a genuine improvement for non-native speakers, and the documented productivity benefits of async-first culture apply with particular force in multilingual teams where meeting overhead includes language overhead.

But async-first culture is not async-only culture, and for good reasons. There are decisions and discussions for which the speed and social bandwidth of real-time conversation is genuinely necessary — rapid problem-solving, emotionally sensitive conversations, situations requiring quick alignment across many stakeholders. These conversations will happen in video calls regardless of async norms, and the dynamics described above will apply when they do.

Async-first culture also creates its own language dynamics that require attention. Long-form written communication rewards a specific kind of English fluency — the ability to write clearly, persuasively, and at length in a formal register. This is a different skill from spoken fluency, and a different skill from native-speaker fluency. Non-native speakers who communicate well in speech may struggle with the written-communication expectations of async-first teams; non-native speakers who communicate well in writing may be disadvantaged in the synchronous moments that async culture cannot eliminate.

The timezone layer

Language and timezone interact in ways that are easy to underestimate. In a globally distributed team, the participants available for a given synchronous meeting are already constrained by timezone overlap. When language is added as an additional constraint, the team is further divided: participants for whom the meeting language is a first language have a structural advantage in the highest-overlap windows, when the most important meetings are scheduled. Non-native speakers who are already at the edge of their timezone overlap — joining a call at 7am or 9pm to accommodate teams in other regions — carry the additional burden of second-language processing on top of the cognitive disadvantage of unusual working hours.

This layering is invisible in most remote team data. Timezone fairness is discussed. Language fairness is not. The result is that the populations most likely to be at the edge of timezone coverage — team members in non-North-American and non-Western-European time zones — are also frequently the populations for whom English is a second language. The compounding of these disadvantages is systematic, but because neither is tracked or named in most organizations, neither is addressed.

What actually helps

The most effective interventions operate at the infrastructure level rather than requiring individuals to change their behavior. Expecting non-native speakers to simply “work on their English” or “be more assertive in meetings” is not an intervention — it is a polite redirection of the structural problem toward the people it is harming.

Infrastructure-level interventions include:

• Real-time voice translation for live meetings, which removes the processing overhead of second-language real-time speech for participants who are not operating in their first language. This is the most direct intervention for the video call problem.
• Multilingual transcription of recorded meetings, which allows non-native speakers to review and fully absorb meeting content in their first language rather than relying entirely on second-language comprehension during the call.
• Explicit async-for-complexity norms, which route the communication types where non-native speakers have the largest advantage — careful, revised writing — toward the decision types where the quality of thinking matters most.
• Team norms that explicitly welcome code-switching — acknowledging that the impulse to think in one’s first language while communicating in a second is normal, manageable, and not a sign of deficiency. This does not solve the problem but reduces the invisible social pressure that compounds language friction into disengagement.

The common thread across these interventions is that they treat language friction as an infrastructure problem rather than an individual problem. This reframe matters because the individual-problem framing — which is the default in most organizations — directs investment toward language training, which takes years to produce meaningful improvement, rather than toward the communication infrastructure that could produce improvement immediately.

Remote work changed where knowledge workers work. It did not change the underlying reality that real-time spoken English is the default medium for high-stakes professional communication in most globally distributed teams, and that the cognitive overhead of operating in a second language concentrates precisely in the moments that matter most. The solution is not to make more people fluent in English faster. It is to build communication infrastructure that works across languages rather than defaulting to one.

Related reading: The Language Tax Your Workplace Is Paying · Babel for Remote Teams · Babel for Digital Nomads