What is code-switching, and why do keyboards fail at it?

Code-switching is the practice of moving between two or more languages within a single conversation — often within a single sentence. If you say “te mando el report mañana, don’t worry”, you just code-switched twice. It’s not broken language or laziness; it’s a fluent, rule-governed way that hundreds of millions of bilingual people communicate every day — Spanglish, Hinglish, Portuñol, Taglish, and many more.

Code-switching is normal — and structured

Linguists have studied code-switching for decades, and one finding is consistent: it follows grammar. Speakers don’t switch randomly; they switch at points that are syntactically valid in both languages, and they do it to pick the word that fits best — the English technical term, the Spanish phrase that carries the right warmth. It’s a feature of bilingual fluency, not a bug.

People code-switch for lots of reasons: a concept is easier in one language, a term of art only exists in the other, or the mix itself signals identity and closeness. The point is that for a huge number of people, the mix is the natural mode — not a fallback.

So why does every keyboard assume one language?

Because the tools were built on a monolingual assumption. Look at the stack you type through:

• Autocorrect picks a single active language and “fixes” every word that isn’t in it — so your English words get mangled inside a Spanish sentence, or vice versa.
• Predictive text is trained per language and can’t predict across a switch.
• Dictation runs one language model, so it guesses a language for every sound and drops the words it can’t place.
• Translation needs a clean source language — but a mix has no single source.

The result: the way you most naturally think is the one input method your devices handle worst. So bilingual people do unpaid work all day — translating in their heads, picking one language and suppressing the other, editing transcripts — just to produce text in the language the situation demands.

The hidden tax on code-switchers

Every mixed thought you have to “clean up” before typing is a small tax: slower messages, more edits, and a nagging sense that the tools weren’t made for you. Multiply that across every email, ticket, and chat in a workday and it’s real time — and real cognitive load — spent bridging the gap between how you speak and what you’re allowed to write.

What fixing it actually requires

A tool that respects code-switching can’t just be “multilingual autocorrect.” It has to do two things in sequence:

• Understand the blend. Read intent across both languages and the switches, rather than forcing everything into one.
• Compose for the moment. Use the context around the cursor and the register the situation demands to produce finished text — not a transcript of the mix.

That’s the design behind Brocaly: speak the way you actually think — code-switched, half-formed, thinking out loud — and get clean, register-appropriate text right at your cursor, in any app. Code-switching isn’t the problem to be corrected; it’s the input to be understood.

Keep reading: Spanglish to English in one step, or how voice-to-text, dictation, and translation differ.