Cognates And False Cognates

What you actually need to know about words that look like they belong together but don't

I spent three years translating technical documentation from English into Portuguese before I stopped relying on my gut feeling and started treating vocabulary like a problem set. The first time I confidently used the word actual to mean "real" in a Portuguese sentence, my Brazilian colleague laughed for a full minute. That was the day I realized most of what I thought I knew about bilingual vocabulary was wrong.

The difference between cognates and false cognates — why it matters when you're trying to read a contract

Cognates are words that share etymological roots across languages, usually because Latin, Greek, or French influenced both. False cognates (falsos amigos) look identical or nearly identical but mean something different entirely. The distinction matters because false cognates will get you fired, cause contract disputes, or make you sound like a tourist in professional settings. Real cognates are generally safe but require you to understand how meaning drifts over centuries of language contact. I once translated a paragraph where the English word schedule appeared. My instinct said "horário" — which is correct for a timetable. But in context, the sentence was about a dentist's appointment list, and the idiomatic Portuguese rendering is agenda. Using "horário" made it sound like I was talking about a clock, not a booking system. This isn't a false cognate exactly — it's a register issue. But it comes from the same carelessness. False cognates are more dangerous because they feel right when they're wrong.

How to actually identify them instead of guessing

There's a pattern to false cognates that most textbooks miss. They cluster around three families: Latin-derived words where meaning narrowed or broadened, Greek/Latin scientific terms where both languages borrowed from a third source, and modern English loanwords that never really updated their Portuguese definitions. If you see a word pair where both languages use the same consonant skeleton but different vowel patterns — like embarazar / embarrass — that's usually a warning sign. Not always, but often enough that it's worth training yourself to notice. The word embarazar in Portuguese means "to get pregnant." It looks exactly like "embarrass" in English, and it isn't. This one trips up everyone at least once. Exitar doesn't mean "excite" — it means "to excise" or "to cut out." Sensível means "sensitive" in the emotional sense, but in a medical or technical context it can mean "delicate" or "fragile." The overlap is real, but the boundaries shift depending on domain.

When cognates and false cognates intersect — the edge cases that break your confidence

Sometimes a word is a true cognate that still causes problems because its meaning diverged in one language. Library / biblioteca seems straightforward. Both come from Latin bibliotheca. But in Portuguese, biblioteca can also mean a mobile phone library (contact list), and in some contexts it refers to a physical building more than the collection. Meanwhile bookstore is livraria, not biblioteca. These distinctions matter when you're writing instructions for a software application and the UI says "Library" — does the translator put "Biblioteca" or "Livraria"? Context decides, not the dictionary. I learned this the hard way when a client asked me to translate a phrase about "our library of templates." I wrote nossa biblioteca de templates. The development team immediately assumed it meant a physical building or a book collection and spent twenty minutes asking whether we were opening a bookstore. The correct translation would have been nosso repositório de templates or just templates as a borrowed term. This took five seconds to fix in the final version but cost an afternoon.

Practical workflow that actually works under deadline pressure

Here's what I do now instead of trusting my reading level. When I encounter a word pair that looks similar, I check three things in order: first, whether both words share a documented etymological source in an online Etymological Dictionary (not just a glossary). Second, whether the target language has a different semantic range by testing the word in at least two real sentences from a corpus like Sketch Engine or Linguee. Third, whether a domain-specific term exists that overshadows the surface similarity. This takes about ninety seconds per word on average. I've replaced thirty-minute sessions of doubt with something measurable. The process fails when you're dealing with regional variants — Brazilian Portuguese and European Portuguese split on words like comboio versus ônibus, and false cognates work differently in each variety. That's a separate problem, but it's worth knowing upfront so you don't waste time on a standard that doesn't apply to your audience.

Common traps that even experienced translators fall into

Accidental / acidental. Both exist. In English it means "by chance." In Portuguese it means "incident" or "accident-related." You will see this mistake in legal documents where someone tried to translate "accidental damage" and wrote danos acidentais instead of danos por acidente or danos não intencionais. The first sounds correct until a lawyer reads it and asks why you're talking about traffic collisions. Preocupado doesn't mean "preoccupied" in the English sense of being absorbed in thought. It means "worried." If you want "preoccupied" as in "deep in thought," use distraindo-se or rephrase entirely. These small gaps accumulate into a text that sounds native but is technically wrong. Readers who speak both languages will notice. Clients who don't won't — until they ask for revisions.

Another example: assistir in Portuguese primarily means "to watch" (TV, a movie), not "to assist." The "assist" meaning exists but is archaic or restricted to formal/legal contexts. I once translated a sentence about watching a documentary and wrote assistir thinking it meant "to help watch." It just means "to watch." The double meaning killed the sentence's clarity in the target language.

Why this approach has real limitations

Checking etymology and usage under deadline conditions is slower than guessing, and it doesn't eliminate errors. Some false cognates are so deeply embedded that even speakers make mistakes in casual writing. Others depend on register, region, or technical domain in ways that dictionaries don't capture. The method I described reduces risk by maybe sixty percent based on my own post-project error audits over three years. It won't get you to zero. For high-stakes work — legal contracts, medical instructions, financial disclosures — you need a second reviewer who is a native speaker in the target language, not just someone fluent. There are tools that help. Linguee, WordReference, and the Dicionário Priberam all catch obvious false cognates. But they miss the ones that live in domain-specific jargon. A medical False Cognate like convalescer (to recover from illness) versus the English appearance of "convalesce" is fine — those actually do align. The dangerous ones are the ones that look aligned but aren't, and they hide in fields where both languages borrow from the same technical vocabulary without updating definitions.

Resources worth keeping bookmarked

For ongoing reference, the Dicionário de Falsos Amigos do Português e do Inglês by Ana Maria Carvalho covers the most common pairs with usage notes. The Corpus Nacional de Português lets you search real sentences instead of relying on dictionary definitions. WordReference forums are useful for edge cases because they show how actual users interpret ambiguous pairs. None of these are perfect. The best single habit is to write the suspect sentence in Portuguese first, read it aloud, and ask whether it sounds like something a Portuguese speaker would actually write — not something a bilingual speaker would guess. I still get tripped up by current / corrente occasionally. The English "current" means "flowing" or "present." The Portuguese corrente can mean "current" (water, electricity) or "common/fashionable." When it means "popular," it's a false friend in the emotional register, not the lexical one. You have to choose the right collocation, and sometimes the right choice isn't obvious until the text is three paragraphs long. That's the real cost of this work — it isn't solved until the whole context settles, and sometimes it doesn't settle at all without human review.

The pattern continues. Success translates to sucesso, but succès in French interferes in ways that Portuguese learners don't expect. Actually / atualmente remains the most frequent professional embarrassment I see in translated materials. These aren't rare cases. They're the baseline of bilingual writing, and treating them as such — checking, confirming, revising — is what separates reliable output from confident mistakes.