GoTranscript
>
All Services
>

En/blog/code Switching In Interviews Transcribe Translate Accurately

Blog chevron right How-to Guides

Code-Switching in Interviews: How to Transcribe and Translate Accurately

Michael Gallagher
Michael Gallagher
Posted in Zoom Apr 23 · 26 Apr, 2026
Code-Switching in Interviews: How to Transcribe and Translate Accurately

To transcribe and translate code-switching accurately, you need to mark every language shift, keep the speaker’s exact meaning, and avoid “smoothing” mixed speech into one language. A good workflow uses consistent labels, clear formatting, and quality checks for high-stakes lines where responsibility or intent could change.

This guide shows practical conventions you can use in interviews, with formatting examples, translation notes, and QA steps that help you preserve meaning.

Primary keyword: code-switching in interviews

Key takeaways

  • Mark every language shift in the transcript so readers can see when and how the speaker switches.
  • Keep the speaker’s intent and tone; don’t “blend” languages in a way that changes meaning.
  • Use a repeatable format: speaker labels, language tags, and bracketed notes for uncertain words.
  • For translation, decide whether to keep some original words (with notes) or fully translate, and stay consistent.
  • QA the moments that carry risk: accusations, consent, instructions, numbers, dates, and legal or medical terms.

What code-switching is (and why it matters in interviews)

Code-switching happens when a speaker shifts between languages or dialects in the same conversation, sentence, or even phrase. In interviews, people may switch to quote someone, show emotion, use the “right” term, avoid a topic, or signal identity.

If you remove or blur these shifts, you can change meaning, intent, or power dynamics. That matters in journalism, research, HR, legal settings, customer interviews, and oral histories.

Common code-switching patterns you’ll see

  • Inter-sentential: whole sentences switch languages.
  • Intra-sentential: switching happens inside a sentence.
  • Tag switching: short inserts like “you know,” “¿verdad?,” “wala,” “na,” etc.
  • Loanwords vs switches: widely adopted words may not signal a true “switch,” but context decides.

When code-switching creates higher risk

  • Responsibility: “He said” vs “I said,” or quoting vs asserting.
  • Consent: agreement, refusal, conditions, or boundaries.
  • Claims and accusations: allegations, denials, or threats.
  • Numbers: dates, amounts, quantities, dosages, addresses.
  • Instructions: steps, requirements, safety warnings.

Transcription rules for code-switching: capture the shift, keep the intent

The goal of transcription is not to “clean up” speech. Your job is to represent what was said in a readable way without hiding language changes that carry meaning.

Set a written style guide before you start so every transcriber and reviewer follows the same rules.

Rule 1: Use clear speaker labels and keep them stable

  • Label speakers consistently (e.g., INTERVIEWER, PARTICIPANT, or names/roles).
  • Don’t change a speaker label because the speaker switched languages.

Rule 2: Mark language shifts where they occur

Readers should never have to guess where one language ends and another begins. Use one of these consistent methods.

  • Inline language tags: add short tags before the switched segment.
  • Bracketed language labels: add labels like [Spanish] or [Arabic].
  • Italicize one language: only if your client accepts it and the file format supports it.

Example (inline tags):

PARTICIPANT: I told him it’s not okay. [ES] No es justo, ¿sabes?

Example (bracketed labels):

PARTICIPANT: I told him it’s not okay. [Spanish] No es justo, ¿sabes?

Rule 3: Don’t blend languages to “smooth” readability

A common mistake is rewriting code-switched speech as all-English (or all of one language) in the transcript. That hides the switch and can remove emphasis or cultural meaning.

  • Write what the speaker said in the language they used.
  • Only normalize minor filler sounds if your verbatim level allows it.
  • If the speaker mixes a phrase, keep it mixed and mark the boundaries.

Example (keep the mix):

PARTICIPANT: I was like, [ES] “no puedo,” and then I just left.

Rule 4: Preserve quoted speech and stance

Code-switching often appears in quotes. Make quotes obvious so readers know whether the speaker is reporting, imitating, or asserting.

  • Use quotation marks for short direct quotes when clarity improves.
  • Use a note when the speaker shifts voice or quotes without clear boundaries.

Example (quote plus stance):

PARTICIPANT: He told me, [FR] “Tu dois venir maintenant,” like it was an order.

Rule 5: Handle partial words, overlaps, and unclear audio transparently

Unclear segments inside a switch can change meaning more than usual. Use a consistent convention.

  • [inaudible 00:12:43] for missing audio with timestamps.
  • [unintelligible] when you hear speech but can’t decode it.
  • [word?] for best-guess with uncertainty marked.
  • Overlaps: use [overlapping] or separate lines if your format supports it.

Example (uncertainty in switched segment):

PARTICIPANT: And then he said [ES][algo?] no sirve” and I felt embarrassed.

Translation approach: decide what to keep, what to translate, and how to show both

Translation adds a second layer of choices. You must decide how much of the original language to retain in the translated version while still making the text easy to read.

Choose one of these approaches based on your audience, purpose, and deliverables.

Option A: Translate everything into the target language (with notes for key terms)

  • Best when the reader needs a smooth target-language document.
  • Add translator notes when a word has no clean equivalent or carries cultural meaning.

Example (full translation + note):

PARTICIPANT (English translation): I told him it wasn’t fair, you know? [Translator note: Speaker switched into Spanish for emphasis.]

Option B: Keep key original words or phrases and translate the rest

  • Best when certain terms matter (titles, insults, kinship terms, place-specific phrases).
  • Use a clear convention so the reader understands what stayed in the original language.

Example (retain term + gloss):

PARTICIPANT (English translation): I was like, “no puedo” (I can’t), and then I just left.

Option C: Two-column or dual-line format (source + translation)

  • Best for legal review, academic research, or sensitive interviews.
  • It makes switches visible and easier to audit.

Example (dual-line):

PARTICIPANT (Source): I was like, [ES] “no puedo,” and then I just left.

PARTICIPANT (Translation): I was like, “I can’t,” and then I just left.

What not to do in translation

  • Don’t erase uncertainty: if the source is unclear, don’t “fix” it in translation.
  • Don’t over-explain in the body: keep explanations in translator notes, not rewritten dialogue.
  • Don’t change register: slang, formality, and politeness often carry intent.
  • Don’t invent expansions: translate what is there, not what you think they meant.

Formatting conventions you can copy: tags, notes, and glosses

A consistent format prevents confusion for editors, attorneys, researchers, and anyone else who reads the file later. Pick a convention and apply it the same way throughout the project.

Language tags: pick a style and stick to it

  • Short tags: [ES], [FR], [AR] placed right before the switched phrase.
  • Full tags: [Spanish], [French] if your audience may not know abbreviations.

Translator notes: use a clear label

Use one label across the entire document, such as [Translator note:] or [TN:]. Keep notes short and factual.

  • Explain untranslatable wordplay or cultural references.
  • Flag when the speaker code-switches for sarcasm, quoting, or emphasis if it affects interpretation.
  • Clarify pronoun ambiguity only if the source language requires it and the audio supports it.

Example (TN for ambiguity):

PARTICIPANT (English translation): He told them to leave. [TN: In the source language, the pronoun could refer to “him” or “them”; audio context suggests “them.”]

Glosses: keep them tight

  • Use parentheses for a short gloss after a retained phrase.
  • Use a footnote/endnote only if your client requires it.

Example (gloss):

PARTICIPANT: She called me [HI] “pagal” (crazy), and everyone laughed.

Names, titles, and honorifics

  • Keep proper names as spoken, unless the client provides a preferred spelling.
  • Keep honorifics if they signal relationship or status (e.g., “Tía,” “Aunt,” “Señor”).
  • Add a note only when the meaning matters and the audience will not know it.

Quality assurance (QA) checklist for high-stakes code-switched segments

Code-switched lines often carry the highest risk of misinterpretation because small choices can shift responsibility, intent, or timeline. QA should focus on those “meaning hinges,” not just spelling.

Step-by-step QA workflow

  • 1) Re-listen to every switch: play the 5–10 seconds before and after each language change.
  • 2) Confirm boundaries: check that the language tags start and stop in the right place.
  • 3) Verify stance: confirm whether the speaker is quoting, paraphrasing, or asserting.
  • 4) Audit pronouns and subjects: ensure “I/you/he/she/they” matches the audio and source grammar.
  • 5) Double-check numbers: repeat-listen dates, times, amounts, addresses, and units.
  • 6) Scan for false friends: confirm words that look similar across languages but differ in meaning.
  • 7) Review notes: make sure translator notes describe, not persuade.

High-risk “responsibility” phrases to QA carefully

  • “He said I should…” vs “I said I should…”
  • “I didn’t” vs “I did,” especially when a negation is short or slurred.
  • “We” vs “they,” which can change who acted.
  • Conditionals: “if,” “unless,” “only if,” “as long as.”

When to escalate to a second linguist

  • The segment includes allegations, consent, or instructions and the audio is unclear.
  • The speaker uses region-specific slang you cannot confidently interpret.
  • The translation would require heavy guessing to sound fluent.
  • Two different readings are plausible and would change meaning.

Pitfalls to avoid (with quick fixes)

Most errors in code-switched transcripts come from trying to make the text “nicer” to read. You can keep readability without hiding the switch.

Pitfall 1: Treating code-switching like a mistake

  • Problem: you delete non-target-language words or replace them with “proper” equivalents.
  • Fix: keep the original words and mark the shift, then add a brief gloss if needed.

Pitfall 2: Over-translating emotion or tone

  • Problem: you add intensifiers (“really,” “extremely”) that weren’t present.
  • Fix: mirror the level of intensity; use a translator note if the switch signals emphasis.

Pitfall 3: Collapsing dialect into “standard” language

  • Problem: you replace dialect grammar with standardized grammar, changing voice.
  • Fix: follow the project’s verbatim level and preserve the speaker’s style where it matters.

Pitfall 4: Inconsistent conventions across files

  • Problem: one file uses [ES], another uses italics, another uses nothing.
  • Fix: add a one-page style sheet to the project folder and enforce it in QA.

Practical decision guide: choosing the right deliverable for your audience

Different audiences need different levels of detail. Decide on deliverables before transcription begins, especially if multiple stakeholders will review the interview.

If your audience is general readers

  • Use a target-language translation that reads smoothly.
  • Keep only the most important original terms with short glosses.
  • Use translator notes sparingly.

If your audience is researchers or reviewers

  • Use a dual-line (source + translation) format.
  • Tag language shifts clearly and consistently.
  • Keep a glossary of repeated code-switched terms.

If the interview is sensitive (legal, HR, medical, compliance)

  • Prefer source + translation and keep timestamps.
  • Escalate ambiguous segments for second review.
  • Maintain an audit trail of changes during proofreading.

Accessibility note (when interviews become media)

If the interview will be published as audio or video, you may need captions or subtitles that handle multilingual speech. In the U.S., the FCC provides rules for closed captioning of video programming, which can apply depending on distribution and context.

See the FCC closed captioning consumer guide for an overview of how captioning is regulated in that context.

If you are captioning multilingual content, consider whether you need separate subtitle tracks for each language and how you will indicate non-English speech on screen. For help with formatted deliverables, see GoTranscript’s closed captioning services and subtitling services.

Common questions

Should I translate code-switched words in the transcript itself?

If you deliver a transcript in the original languages, keep the words as spoken and add a brief gloss or translator note when the meaning matters. If you deliver a translated transcript, translate the meaning but consider keeping key original terms with a gloss when they carry cultural or legal weight.

What’s the best way to mark language shifts?

Use bracketed labels like [Spanish] or short tags like [ES] right before the switched phrase. Choose one method and apply it consistently across all files.

How do I handle slang or regional terms I can’t translate cleanly?

Keep the original term and add a short translator note explaining the closest meaning in context. If the segment affects responsibility or consent, consider a second linguist review.

What if the speaker mixes languages inside the same word or uses borrowed words?

Transcribe what you hear and mark the switch where it actually happens. If a borrowed word is fully common in the surrounding language and does not signal a meaningful switch, you can leave it untagged, but stay consistent within the project.

Can I “clean up” grammar for readability?

Only if the project calls for clean-read transcription and you can do it without changing meaning or voice. Avoid standardizing dialect or rewriting code-switches into a single language.

How do I QA code-switching when I don’t speak one of the languages well?

You can still QA structure: confirm where the switch happens, keep timestamps, and avoid guessing. For meaning-critical segments, you should bring in a qualified bilingual reviewer.

Do I need captions or subtitles as well as a transcript?

If the interview will be published as video, captions or subtitles help viewers follow code-switched speech. Transcripts, captions, and subtitles serve different purposes, so plan deliverables early.

If you want a workflow that keeps every language shift clear while staying readable, GoTranscript can help with transcripts, translations, and multilingual deliverables. You can start with professional transcription services and request formatting conventions that match your interview needs.