Skip to content
CertTranslate
Non-Text ElementsHaitiHaitian Creole, French (code-switching), English (code-switching) → English

Labeling Speakers in a Foreign WhatsApp Chat

A client submitted a year of WhatsApp screenshots in Haitian Creole as USCIS evidence.

The chats had no speaker labels in the text — only bubble color and alignment indicated who sent each line.

About this case

Document
Personal correspondence — WhatsApp chat screenshotsKonvèsasyon chak jou (Haitian Creole)
Origin
Haiti
Languages
Haitian Creole, French (code-switching), English (code-switching) → English
Intended use
USCIS (I-130 / I-485, bona fide marriage evidence)
The short answer

How do you label speakers in a WhatsApp chat translation that has no names in the text?

A client submitted a year of WhatsApp chat screenshots in Haitian Creole, with French and English code-switching, as bona fide marriage evidence. The original messages carried no speaker labels in the text. We attributed each line to either '[You]' or the saved contact name '[Hubby♥]' — both labels the app itself prints in its reply-quote boxes. No gender or identity was inferred by the translator. The certified translation was delivered for the client's USCIS filing.

The documents

What We Received

A client submitted three batches of WhatsApp chat screenshots, photographed onto single pages by month. The chats were in Haitian Creole, with frequent French and English code-switching. The contact in every chat was the saved name of the client's spouse, including a heart emoji. The translation was needed as personal-correspondence evidence for an I-130 / I-485 filing.

The packet was a certified text-message translation spanning more than a year of daily messages. Each line had to be attributed to its sender — but the source carried no name on each line.

The challenge

Why This Required Special Handling

WhatsApp screenshots have no speaker labels in the text — the only built-in signals are visual (bubble color and alignment). A flat transcription drops both, and the reader cannot tell who said what.

Adding '[Him]' or '[Her]' labels breaks a core rule of certified translation: the translated text must reproduce, not infer. Gender is not in the source, and a label that requires a guess can silently flip when applied across hundreds of bubbles in a multi-batch order.

Our approach

How We Handled It

We labeled every line with only the markers WhatsApp itself prints in the source. Right-aligned outgoing bubbles became '[You]' — the exact label the app uses in its reply-quote boxes for the phone owner's messages.

Left-aligned incoming bubbles took the contact name from the chat header, including the heart emoji. Both labels are visible inside the source, so a second translator can verify any attribution against the same screenshot.

We added one more rendering rule. WhatsApp's reply feature became '↳ Replying to [Speaker]: <quoted text>' indented before each reply. System rows — voice calls, missed calls, voice messages — were marked '[System]' or '[Voice message, M:SS]' and italicized. Embedded photos, forwarded files, and Status reshares were described in brackets at the speaker's position.

Two supporting standing notes accompanied the certification block on every batch. One documented the language mix — Haitian Creole bodies with French and English insertions preserved verbatim. The other documented bracket conventions for non-text elements so an adjudicator could distinguish voice messages, system rows, and shared media from typed text at a glance.

The outcome

The Outcome

Eight deliverables were rendered under the same order number, each using the same convention. The packet was delivered to the client for inclusion in their USCIS marriage-based filing.

We have since used this convention on other foreign-language chat orders covering I-130 evidence and I-751 evidence, with no RFE related to speaker attribution. The pattern works on any messaging app whose UI prints both the contact name in the header and a phone-owner label in reply-quote boxes.

Key takeaway

What This Means for You

If you submit chat screenshots as evidence, the translation should not invent speaker labels. Use markers visible in the source — the contact name and the app's own 'You:' label. Any reader can then verify each attribution by re-reading the screenshot.

Have a similar situation?

$24.95 per physical page · no word cap

Start my order
Behind the case

Authorship, sources & history

Case author

Marie-Claire Joseph

Haitian Creole & French Certified Translator

Update log

  1. Updated

    Re-checked the cited sources and replaced a USCIS link that does not open with 8 CFR 103.2(b)(3) on eCFR, the USCIS regulation on translations of foreign-language documents.

  2. Published

    This case study was first published.

Sources & references

  1. 8 CFR 103.2(b)(3) — Translations

    Electronic Code of Federal Regulations (eCFR) · Verified

  2. I-130, Petition for Alien Relative

    USCIS · Verified

Continue reading

All translation cases

All identifying information has been removed from document images. Case details are shared with client permission. No personal data is stored or displayed on this page.

Back to top