You spell your last name at the clinic front desk. "V-A-R-G-A-S." The receptionist types Bargas, because the letter you said came out as be and the letter she needed was uve. You correct her, she fixes it, and everyone moves on.
Ten minutes later you tell her the appointment is at seven. That one comes out as seben, and nobody corrects it, because nobody notices. That second mistake is the one that will still be there in five years.
Three /v/ sounds going to a B. Only the first one is loud enough that someone stops you.
Below: why Spanish speakers say B instead of V, the rule that is firing inside your sentences, the one-second test that tells you which sound you just made, and the position where the error hides for years. (English speaker learning Spanish? You want the opposite lesson.)
Spanish spells two letters and says one sound
Vaca is a cow. Baca is the rack on the roof of a car. Say both out loud and you have said the same word twice.
On the phone, Spanish speakers say b de burro or v de vaca. In Venezuela, b de Bolívar and v de Venezuela: a country disambiguating a letter with its own name.
The philologist Francisco Javier Álvarez Comesaña, quoting the Real Academia Española, puts it more bluntly than most teaching materials do: "La pronunciación de la v como labiodental no ha existido nunca en español". The pronunciation of v as a labiodental has never existed in Spanish. Not "was lost." Never existed.
That is the Academy describing the standard language. Some varieties disagree. Yours may be one of them, and that is the last section of this post.
The merger is old enough that Latinists were joking about it five centuries ago. Beati hispani, quibus vivere bibere est: fortunate are the Hispani, for whom living is drinking. The pun is usually credited to the sixteenth-century humanist Julius Caesar Scaliger, though nobody is sure who said it first.
Spanish has one sound and two letters. English has two sounds and needs both.
Say "Barcelona." Now say "a Barcelona."
The first time, your lips closed completely and popped apart. The second time, after that little a, they came close and never quite touched. Same word, same speaker, two different consonants, four seconds apart. You have been doing that your whole life and you have never once noticed.
That is the rule. Spanish /b/ surfaces as a hard stop only after a pause or after a nasal consonant: Barcelona at the start of a sentence, un beso, hombre. Everywhere else it softens into an approximant that linguists write [β̞], and Duolingo describes in the clearest words anyone has managed: "the lips come close to touching but they don't quite touch".
When you say seben, you are not failing to produce an English /v/. You are successfully producing a Spanish rule, on schedule, in the exact environment it was built for. The rule does not know it has crossed into English. Nobody ever told it to stop. Korean and Tagalog speakers land on berry for very too, but by a different route: Tagalog never had /f/ or /v/ to begin with.
How to pronounce the V sound in English: teeth, not lips
Say vvvvvvv and hold it for three seconds. Now try to hold b. You can't. It happens once and it's over.
English /v/ is a voiced labiodental fricative, which unpacks into three plain instructions: the top teeth rest on the bottom lip (labiodental), air keeps escaping through the gap (fricative), and the voice is switched on (voiced). English /b/ is a voiced bilabial stop. Both lips close, pressure builds behind them, the closure breaks. That is one event, over in an instant.
Spanish uses lips for both, and no teeth anywhere. So the entire task is one move your language never asked for: get the top teeth onto the bottom lip and buzz. Pronunciation Studio's guide for Spanish speakers puts it the same way: English /v/ is "a voiced fricative using teeth and lip", and Spanish speakers "tend to replace it with a plosive /b/ or an approximant sound using both lips." Watch for the halfway failure. The teeth land, then slide back to the lips the moment the voicing starts, because /b/ is the voiced sound your mouth already trusts.
Three checks you can run alone, before you can even hear the difference:
- Hold it. If you can sustain it, it was V. If it popped and stopped, it was B.
- Mirror. Top teeth visibly touching the bottom lip means V. Lips pressing together means B.
- Throat. Finger on your voice box. Say fff, then vvv. Same mouth, and only the second one vibrates. That vibration is the only thing separating V from F.
The V at the start gets corrected. The one in the middle doesn't.
Very at the start of a sentence comes out as a hard, full [b]. It sounds like berry, which is a different English word, so the sentence breaks and someone asks you to repeat it. You get feedback. Over a few years, you fix it.
Seven, over, never, eleven, favorite: these come out as [β̞], the soft approximant, because that is what Spanish does between vowels. And [β̞] is close enough to /v/ that an English ear usually files it as a slightly odd V and keeps going. The sentence survives. Nobody stops you. Nothing gets fixed.
I am not claiming the middle of a word is physically harder, and I have not found a study measuring how English listeners hear [β̞] either. No study I could find measures Spanish speakers' English /v/ accuracy by word position at all. What is checkable is the feedback: one error is loud and draws correction, the other is quiet and doesn't, so they have very different life expectancies.
This is the accent feature that outlives fluency, alongside the phantom e in Espain and eschool. Ten years in the US, a job conducted entirely in English, and seben is still there, because that word was never corrected once. Word-initial practice does not reach it.
The drill list for this position is single words, said slowly, teeth landing every time:
seven, eleven, over, never, ever, every, river, driver, favorite, oven, given, moving, believe it, love it.
Both /v/ sounds sit between vowels, which is exactly where Spanish softens them automatically.
"It's a big job": the B that goes soft too
Say "Big job!", then "It's a big job." The second B, sitting after that a, quietly turned into [β̞] on you.
Every guide I read while writing this treats the problem as one-way, Pronunciation Studio included: add the missing V. That is half the fix. The Spanish rule reaches into English /b/ as well, and that is documented. In a review of Spanish-speaker pronunciation problems in Gist, Uribe Enciso and colleagues write that Spanish speakers "are prone to give English /b/, /d/ and /g/ their mother tongue values," and that "between vowels, these sounds are softer continuous sounds, not stops."
An English listener hears a slightly smeared B rather than a wrong word. Targets worth drilling with a hard closure: robber, rubber, cabin, habit, rabbit, maybe, neighbor, about it, a bad habit.
Try it in Conversa
Practice with AI characters who adapt to your level and give real-time feedback.
Try Conversa FreeMinimal pairs, sorted by where the sound sits
TV and TB. One is a television, the other is tuberculosis, and the only thing between them is whether your teeth touched your lip. Vee and bee both put the sound right at the front, so start there.
Word-initial, the easy position, where feedback already helps you:
van/ban, vote/boat, vest/best, vet/bet, vat/bat, very/berry, vent/bent, vile/bile, vowel/bowel, vigor/bigger.
Vase and base only work as a pair in American English; British vase is /vɑːz/.
Word-final, where Spanish gives you almost nothing to transfer:
curve/curb, carve/carb, serve/Serb, verve/verb, rove/robe, jive/jibe.
Between vowels, the position this whole post is about, and here is the awkward truth. Real intervocalic minimal pairs barely exist in English. There are a few (revel/rebel, ravel/rabble, saver/saber, lover/lubber) and then the well runs dry. Widen it to word-medial and marvel/marble and curving/curbing come back in, where the consonant before the /v/ is an r rather than a vowel. That scarcity is why nobody drills this position. When the pairs run out, drill single words and check the teeth every time.
The cognates where the habit is deepest
Conversación becomes conversation. Visitar becomes visit. Televisión becomes television. Your eye recognizes the word instantly, your mouth reaches for the version it already owns, and the Spanish consonant arrives attached to it. Cognates feel like free vocabulary, and for this sound they are the most expensive words on the page.
The sharper trap runs the other direction, where Spanish writes ⟨b⟩ and English wants ⟨v⟩. Recibir becomes receive, gobierno becomes government, gobernador becomes governor, probar becomes prove. These are harder than the conversación group, and not because the English spelling helps any less. Receive has a ⟨v⟩ sitting right there. The difference is what is already welded to the meaning: you have said recibir with both lips a few thousand times, so the word arrives with a bilabial attached and the ⟨v⟩ on the page has to argue with it. In conversation, the page and the mouth ask for the same thing.
When Spanish spells it with a b and English spells it with a v, the English word wants your teeth.
If you grew up in Chile or El Paso, your mouth may already do this
The Academy line near the top of this post says the labiodental v has never existed in Spanish. For a lot of readers, it exists in their own mouth.
Scott Sadowsky recorded speakers in Concepción, Chile, and found the labiodental [v] in 59.5% of all tokens of /b/. That is the majority realization of the phoneme, not an occasional variant. He also found that spelling had nothing to do with it, so this is not people reading a ⟨v⟩ and performing one.
Closer to this page's likely reader: in Ortega's master's thesis on Spanish in El Paso, Texas, every single participant produced [v] as a variant of Spanish /b/, and speakers who met English earlier produced more of it.
So the honest version is narrower. Standard Spanish has no phonemic /v/. If your variety hands you a labiodental [v] some of the time, that is documented, it is not an error, and it is a head start on the mechanics. The same variant is reported in parts of Spain, including Asturias, Catalonia and Andalusia. What none of those varieties does is make [b] and [v] mean two different words. That is the new job, and no amount of head start does it for you.
What to practice tomorrow
Two sentences, once a day, out loud, in front of a mirror.
For V: "Next vacation I'd love to visit the river," which is Pronunciation Studio's drill line. Vacation and visit start with the V, where you already have some control. Love ends on one and river buries one between vowels, which is where you don't.
For B: "Bob grabbed a rubber bag for a big job." Only the first B starts the utterance. Every one after it has a vowel in front, and every one of them will try to go soft.
Say them slowly enough that you can feel the teeth land, and stop caring about speed until the landing is automatic. This is a motor habit, and reading about it changes nothing on its own. Knowing the rule is running just tells you where to look. Someone will eventually correct your very. Nobody is going to correct your seven.