The translation tax
Stand on a construction site and watch an engineer take field notes. If English is their first language, they talk fast. Details come in full sentences. "Rebar spacing at the east curtain wall footing looks tight, maybe 150 instead of 200, check the drawing before concrete placement tomorrow morning." The observation is dense because the engineer is not thinking about language. They are thinking about the structure.
Now watch an engineer whose first language is Farsi, Mandarin, or Hindi do the same thing. The observation quality does not drop. The engineering judgment is identical. But the transcription step changes. The engineer sees the same tight rebar spacing, makes the same professional assessment, and then pauses. They compose an English sentence in their head before speaking or typing. The sentence comes out shorter, more generic, less specific about what they actually observed.
This is not a fluency problem. Many internationally-trained engineers are fully fluent in professional English. The issue is cognitive load. Composing prose in a second language while simultaneously evaluating structural work takes more effort than doing either task alone. The engineer compensates by simplifying. Shorter notes. Fewer qualifiers. Less context about what prompted the observation.
The result is a thinner capture record, which produces a thinner draft, which requires more revision before the report is ready for seal. The time the engineer saved by writing a brief note gets spent later, adding back the context that was lost in translation.
The scale of this in Ontario
The Professional Engineers Ontario (PEO) licensing data tells a clear story. Approximately one-third of Ontario's more than 73,000 licensed engineers were educated outside Canada. Since 2005, PEO has licensed more International Engineering Graduates than graduates of Canadian-accredited engineering programs in most years. The Greater Toronto Area, where the majority of Ontario's structural and building-science firms operate, has one of the most linguistically diverse professional workforces in the world.
The common first languages in GTA engineering practices include Farsi, Mandarin, Hindi, Arabic, Urdu, Punjabi, Tagalog, Portuguese, Russian, and Ukrainian. A twelve-engineer structural firm in the GTA is likely to have engineers whose first language is not English on staff. A thirty-engineer firm almost certainly does.
These are not junior staff. Many are experienced P.Eng holders who have been practising for a decade or more. They review construction, exercise professional judgment, and sign sealed documents. They do all of this competently in English. But their field notes, the raw material that becomes the sealed report, could be richer, faster, and more detailed if the capture step did not require simultaneous translation.
What gets lost
Consider a specific scenario. An engineer is reviewing a concrete placement on a high-rise podium. They observe that the form ties at the south elevation are spaced closer than the approved shop drawings indicate. In their first language, they would note the exact location, the approximate spacing they observed, the drawing reference they are comparing against, the weather conditions affecting the pour, and a preliminary assessment of whether this is a conformance issue or a field adjustment within tolerance.
In English as a second language, under time pressure on site, the same observation often becomes: "Form tie spacing south elevation appears non-conforming. Verify against approved drawings." The engineering judgment is preserved. The factual accuracy is preserved. But the contextual richness, the material that makes a sealed field review report defensible and specific, is thinner.
When this note reaches the drafting step, the engineer or the drafting tool must expand it. The original observation had more detail than what was captured. That detail existed in the engineer's head, in their first language, at the time of the site visit. By the time drafting happens, some of it has faded. The result is a report that reflects what was written down, not everything that was observed.
What Fermito does about it
Fermito accepts voice and text input in 15+ languages. The supported list includes the languages most commonly spoken in Ontario engineering practice: Farsi, Mandarin, Hindi, Arabic, Urdu, Punjabi, Tagalog, Portuguese, Russian, Ukrainian, French, Spanish, Italian, and Tamil, along with auto-detection for less common languages.
An engineer standing on a construction site can dictate their field observations in Farsi. They can type notes in Mandarin. They can mix languages within a single visit record if that is how they naturally work. The capture device does not care. The transcript records what was said or typed.
Translation happens at generation time, not at capture time. When Fermito assembles the draft, it reads the full visit record, including any non-English content, and produces the report in the firm's signing language (English, for Ontario firms). The translation is not a separate step. It is part of the same generation pass that applies the firm's template, pulls from the standard phrase library, and structures the observations into a reviewable document.
This means the engineer does not plan around language. They do not think "I need to compose this in English because the tool requires English input." They think about the structure they are observing. They capture what they see in whatever language lets them capture the most. The language question is resolved downstream, automatically, without the engineer making a workflow decision about it.
What this looks like in practice
Here is the difference in a real workflow:
Without multilingual capture. Engineer arrives on site. Observes concrete placement. Mentally translates observations into English. Types abbreviated English notes. Returns to office. Opens Word template. Expands abbreviated notes into full report prose, trying to recall details that were lost in the abbreviated capture. Reviews and revises. Signs and seals.
With multilingual capture. Engineer arrives on site. Observes concrete placement. Dictates observations in their first language, at full speed, with full detail. Returns to office. Fermito generates a draft in the firm's house format, in English, with the full observational detail preserved through translation. Engineer reviews the draft, which is richer because the capture was richer. Revises as needed. Signs and seals.
The engineering judgment is identical in both cases. The site visit is identical. The review and seal are identical. The difference is in the capture-to-draft pipeline, and specifically in whether that pipeline forces the engineer to translate their own observations before the tool can process them.
What this does not change
Multilingual capture does not change the language of the sealed document. Ontario engineering firms sign and issue reports in English. The OBC is published in English. PEO operates in English. CSA standards are published in English (and French). The sealed report that leaves the firm is an English-language document, regardless of what language the field notes were captured in.
Multilingual capture does not reduce the review obligation. The reviewing engineer, typically the principal or a senior P.Eng, still reads every word of the draft before signing. If anything, multilingual capture raises the review bar slightly because the reviewing engineer is reading translated observations rather than the original language. The attestation step, where the engineer explicitly takes ownership of the document, is unchanged.
Multilingual capture does not automate judgment. The engineer on site is still the person who decides what to observe, what to document, and what to flag. The language of their notes does not change the substance of their professional opinion. It changes the throughput of their capture, which changes the richness of the draft, which changes the amount of revision needed before seal.
The productivity case
The drafting time reduction from multilingual capture is not separate from the broader drafting time reduction that Fermito provides. It is additive. A firm that uses Fermito to cut drafting time by 60 to 70% will see an additional benefit for engineers whose first language is not English, because those engineers will produce richer capture records, which will produce more complete first drafts, which will require less revision.
The incremental benefit is difficult to isolate from the baseline. But the direction is clear: an engineer who captures at full speed in their first language produces more detailed raw material than an engineer who self-translates on site. More detailed raw material produces a better first draft. A better first draft means less time in review and revision.
For a firm with several internationally-trained engineers on staff, each producing 4 to 8 field review reports per week, even a modest reduction in per-report revision time compounds across the team. The language barrier in the capture step is not a dramatic obstacle. It is a persistent friction that slightly thins every report and slightly extends every revision cycle. Removing it does not transform the workflow. It sharpens it.
A capability, not a workaround
It is worth noting what this feature is not. It is not a workaround for engineers who struggle with English. It is not an accommodation. It is a recognition that field observation quality is highest when the engineer is not simultaneously translating, and that a sealed report is strongest when the capture record is richest.
Ontario's engineering workforce is multilingual by composition and increasingly so. PEO's licensing data shows that trend. A drafting tool that requires English input is forcing a third of its potential users to work through an unnecessary translation layer at the point in the workflow where observational detail matters most.
Fermito removes that layer. The engineer observes. The engineer captures in whatever language serves the observation best. The draft arrives in English, in the firm's format, ready for review and seal. The language decision happens once, at generation time, instead of every time the engineer opens their mouth or picks up their phone on site.