According to a 2017 study published in the Journal of International Business Studies, managers evaluate non-native speakers as having 18% lower leadership potential than native speakers, even when their objective technical performance is identical.

The review room was twelve feet by twelve feet. It had beige walls and a grey carpet with a pattern of dark blue speckles. In the center of the room sat a rectangular table made of medium-density fibreboard with a maple-grain laminate. A black mesh ergonomic chair was positioned on each side of the table. A plastic water pitcher sat in the corner, perspiration forming a ring on the surface of the sideboard.

The Geometry of the Review

Mark sat in the chair facing the door. He held a blue plastic pen and a three-page document printed on 20-pound white paper. The document was titled “Annual Performance Review – Fiscal Year 2023.” Elena sat across from him. She was a senior software engineer who had joined the company forty-two months prior. She had moved from São Paulo to the Chicago office six months after her start date.

Mark looked at the third section of the document. This section was titled “Core Competency: Communication and Influence.” The section contained five sub-categories, each with a rating scale from one to five. Mark had circled a two for the sub-category titled “Articulation of Complex Concepts in Group Settings.” In the comments box below, he had written: “Elena often hesitates during high-pressure architecture reviews. She needs to demonstrate more confidence and clarity when defending her technical decisions to the stakeholders.”

Mark did not know that Elena was mentally navigating a grammatical structure that required three different verb tenses while he spoke. He watched her eyes. They moved slightly to the left as she waited for him to finish his sentence. To Mark, this appeared as a lack of conviction.

The evaluation process for a modern corporation follows a specific sequence of data entry. First, the Human Resources department creates a template within the Human Resources Information System, or HRIS. The template includes twenty-four distinct behavioral markers. These markers are grouped into three “buckets”: Technical Skill, Culture Fit, and Communication. Each bucket is assigned a weight. At this particular company, Technical Skill was weighted at 40%, while Communication and Culture Fit were each weighted at 30%.

When the manager opens the review portal, the software presents a series of radio buttons. The manager clicks the button that most closely aligns with their observation. The software then calculates a weighted average. If the average falls below 3.5, the system automatically flags the employee for a “Performance Improvement Plan” or a “Developmental Focus Area.” The software contains no field to input the primary language of the employee or the cognitive load required to translate technical jargon in real time.

I adjusted my chair and looked at the clock on the wall. It was a circular analog clock with a black rim. I made sure to keep my eyes on the screen and move my mouse in a deliberate fashion as the department head walked through the hallway. I opened a spreadsheet and began entering numbers into column C.

The Emoji Specialist and the Latency Gap

Marie B.-L. worked in the office three doors down from the review room. She was an emoji localization specialist. Her desk was covered in small sticky notes. Each note contained a single symbol and a list of countries. She once spent four hours explaining how the “folded hands” emoji was interpreted as a “high five” in certain regions and a “prayer” in others. She kept a printed chart of the 3,633 emojis currently recognized by the Unicode Consortium.

Marie often talked about the “latency of meaning.” She said that the time it takes for a thought to leave one brain and enter another is not a constant value. It changes based on the medium. She noted that a person might be the most intelligent person in a room of five people, but if those five people are using a medium the person has only mastered at 70%, the person will be recorded as being 30% less intelligent.

In the review room, Mark asked Elena a question about the database migration. Elena looked at the whiteboard. The whiteboard was 48 inches wide and 36 inches tall. It had several dried stains from green and red markers that had not been properly erased. She began to speak. She stopped after the fourth word. She searched for the term “concurrency.” She found it after a 1.4-second delay. During that 1.4-second delay, Mark looked down at his paper and tapped his pen against the laminate table. He made a small mark in the margin of the review form.

The ink from the pen was black. It was a ballpoint pen with a rubber grip. The mark he made was a small dash. To Mark, the dash represented “uncertainty.” In the final report, this dash would be synthesized into a sentence about “leadership presence.”

Cognitive Load and Linguistic Survival

A live conversation typically moves at a rate of 130 to 150 words per minute. In a technical environment, this speed can increase as participants use shorthand and acronyms. When a person is operating in their second or third language, the brain must perform a three-step operation: decode the incoming audio, map the meaning to a conceptual framework, and then encode a response back into the target language. This process requires a significant portion of the brain’s prefrontal cortex. This is the same area used for complex problem-solving and executive function.

When the prefrontal cortex is occupied with the mechanics of language, it has less capacity for the “performance” of confidence. The voice may drop in volume. The posture may become more rigid. The person may use simpler vocabulary to ensure accuracy. These are all linguistic survival strategies. However, in the context of a corporate performance review, these strategies are often coded as “low engagement” or “lack of strategic thinking.”

The firm used a software called Transync AI for its international calls. The tool was designed to reduce the word error rate in live speech to under 5%. It provided bilingual subtitles on the screen during the meeting. When Elena used the tool during calls with the Tokyo team, the subtitles appeared in sub-0.5 seconds. The Tokyo engineers could see her words in Japanese and English simultaneously. In those meetings, the Tokyo team rated Elena’s “clarity” as a five out of five. Mark was not present for those meetings. He only saw her in the Chicago office, where there were no subtitles.

The review form stayed on the table. It had a coffee stain on the bottom left corner of the first page. Mark moved to the “Goals for Next Quarter” section. He wrote: “Improve verbal communication speed. Practice presenting to the team.” He did not write anything about the technical architecture Elena had designed for the migration, which had saved the company $41,200 in monthly hosting costs.

I watched a pigeon land on the windowsill outside. It stayed there for twenty seconds and then flew toward the parking lot. The parking lot was 60% full. Most of the cars were silver or white. I typed the number 1,482 into the spreadsheet.

Marie B.-L. came into the breakroom. She was carrying a mug with a picture of a cat on it. She looked at the review room through the glass partition. She said that in the 1920s, intelligence tests were often given to immigrants in a language they did not speak fluently. The results were then used to justify restricted immigration quotas. She said the practice had not ended; it had just moved into a different set of forms.

“The practice had not ended; it had just moved into a different set of forms.”

– Marie B.-L.

Standardization as a Junk Drawer

The “Communication” box on the form was a junk drawer. It held the sound of an accent. It held the silence between sentences. It held the cultural difference in how one approaches a superior. Because the form was standardized, it could not distinguish between a person who did not understand the technology and a person who did not have the specific English word for the technology at that exact moment.

Mark finished the review. He stood up and smoothed his shirt. Elena stood up as well. She shook his hand. The handshake lasted approximately 1.5 seconds. Mark went back to his desk and logged into the HRIS. He clicked “Submit.” The data was now part of the permanent record. It would be used to determine Elena’s salary increase for the following year.

The system recorded a 2.8% raise. If the communication score had been a four, the raise would have been 4.5%. The difference was $1,940. This figure was not based on the code she wrote or the systems she maintained. It was based on the 1.4-second delay she took to find the word “concurrency” while sitting in a room with beige walls and a grey carpet.

I closed the spreadsheet. The department head had gone to lunch. I looked at the three-page document Elena was carrying as she walked back to her desk. She placed it in a blue folder and put the folder in the bottom drawer of her desk. She then opened her IDE and began to write a script. Her fingers moved across the keyboard at 85 words per minute. There was no hesitation. The code was clean. The logic was sound. On the screen, she was perfectly articulate.