Grounded in second-language acquisition research
Stop pretending.
Start speaking.
Parla drops you into a living world — cafés, markets, train stations — with a live AI tutor in your ear. Real conversations, real situations, and a curriculum built on how people actually acquire languages. Not streak psychology.
20+ languages · Speak out loud from day one · 7-day free trial
Buongiorno! Cosa le porto?
Good morning! What can I get you?
Un cappuccino e… come si dice «croissant»?
Si dice «cornetto»! Allora — un cappuccino e un cornetto?
It's called a «cornetto»! So — a cappuccino and a cornetto?
The illusion of proficiency
A 700-day streak that can't order a coffee
Gamified apps are brilliant at one thing: keeping you tapping. It feels like progress — but the learning science is blunt: performance during smooth, easy practice is a poor predictor of what you can actually do. Speaking is a skill, and the only way to build it is to speak.
The streak-app way
- Tap words from a word bank
- Points and gems for showing up
- One course, the same for everyone
- Sentences about apples and bananas
The Parla way
- Say full sentences out loud, from day one
- Progress measured by what you can actually produce
- A learner model that tracks every word you know — and how well
- Ordering dinner, asking directions, making friends
Cultural immersion, on demand
A world to walk into
Immersion is the gold standard for language learning — but most of us can't move to Rome for a year. Parla brings the immersion to you: a living environment where you order, haggle, ask, and answer your way through everyday life, in the language you're learning, from wherever you are.
This is situated, task-based learning — the same principle behind why living abroad works. Every scene is retrieval practice in disguise.
Every scene is a task
Buy dinner at the market. Ask for directions to the station. Handle the pharmacy. Tasks have real outcomes — and finishing them takes real language, not multiple choice.
Your tutor sees what you see
The tutor knows the scene you’re standing in, the words you’ve mastered, and what’s due for review — so every conversation is pitched exactly at your level.
Characters, not chatbots
A barista in Rome, a shopkeeper in Oaxaca. Personas with personality make you negotiate meaning the way real conversation demands.
The world feeds your reviews
Everything you say and hear flows back into your learner model. The word you fumbled at the café shows up again — right before you’d forget it.
What you get
Everything is engineered for one outcome: you, speaking
A live tutor that talks back
Full-duplex voice conversation in real time. Interrupt naturally, get corrected gently in the moment — like a patient tutor who never gets tired.
Chunks, not word lists
Fluent speakers build sentences from ready-made pieces — «come si dice…?», «me pone…?». Parla teaches these formulaic sequences as units, so speech comes out whole.
Reviews timed to your memory
FSRS spaced repetition models when you’ll forget each word and resurfaces it just before you do — tracking whether you can recognize it or actually say it.
Input pitched at your level
The tutor keeps what you hear ~95–98% within words you know, so you understand almost everything while always stretching just past your edge.
Practice anywhere — even a loud café
Noisy-environment mode switches to push-to-talk, so you control the mic. No pressure, no misfires, no waiting for silence to practice.
20+ languages, one tutor
Spanish, Italian, Japanese, Mandarin, French, German, Korean, Arabic and more — with new languages added over time.
The science
Grounded in 40 years of research on how adults acquire languages
We didn't design Parla around engagement metrics and then look for evidence. We started from the second-language acquisition literature and built the product it prescribes.
You learn to speak by speaking
Output hypothesis — Swain (1985)
Input alone plateaus: immersion students who only listen and read understand well but under-produce. Speaking forces your brain to actually process grammar, not just skim meaning.
Conversation is the engine
Interaction hypothesis — Long (1996)
Negotiating meaning in real back-and-forth — clarifying, repairing, being understood — drives acquisition, with the biggest gains for learners outside immersion.
Understand almost everything, stretch a little
Comprehensible input — Krashen (1985); Nation (2006)
Learning happens fastest when input sits just above your level. Parla gates tutor speech to your known vocabulary and meters in the new.
Recall beats rereading, spacing beats cramming
Cepeda et al. (2006); Roediger & Karpicke (2006)
Retrieval practice at expanding intervals is the most robust finding in memory research. It’s the backbone of Parla’s review scheduling.
Easy practice feels good and teaches less
Desirable difficulties — Soderstrom & Bjork (2015)
In-session performance and long-term learning come apart. Tapping matching pairs feels smooth precisely because it demands so little.
Skills are specific
Transfer-appropriate processing — DeKeyser (2007)
Practice improves what you practice. Speaking practice builds speaking. Tapping builds tapping.
Our promise: engagement never comes at the cost of learning. If a feature makes the app stickier but you worse at speaking, it doesn't ship.
How it works
Three steps, no gems
Tell us why you’re learning
Travel, family, work, love — your goals shape the scenes you’ll practice and the chunks you’ll learn first.
Step into a scene and speak
Order at the café. Find the platform. Your live tutor listens, replies, and nudges you when you slip — out loud, from the first session.
We remember, so you don’t have to
Every word and chunk you meet is scheduled for review at the moment you’d otherwise forget it. Show up and speak; the system does the bookkeeping.
Languages
Where will you belong?
Tap a language to hear a real conversation.
Pricing
Simple pricing. No ads, no gems, no guilt.
Lessons, chat, and vocabulary review are free. Go Pro to unlock unlimited live speaking practice — with a 7-day free trial.
Early Access
Limited availability. Unlimited access to all languages. $14 now and forever + 7 day free trial. Over 50% off.
$14/month
Basic
Speak with your AI language learning partner in your native language and one target language of choice.
$20/month
Polyglot
Unlimited access to all 15+ AI language learning partners. Access to new languages when they become available.
$49/month
FAQ
Fair questions
How is Parla different from Duolingo?
Duolingo is very good at building a habit of using Duolingo. Parla is built for a different outcome: holding a conversation. That means speaking out loud from your first session, a tutor that talks back and corrects you, and progress measured by what you can produce — not by streaks, gems, or leagues. If you want five relaxing minutes of tapping, keep the owl. If you want to order dinner in Rome, come speak.
I’m a complete beginner. Can I really speak from day one?
Yes — because you start with chunks, not grammar tables. Ready-made pieces like «come si dice…?» (how do you say…?) let you hold a real exchange in your first session, and the tutor keeps everything it says ~95–98% within words you already know, so you’re never lost.
What if speaking makes me anxious?
That’s exactly who we built this for — anxiety is one of the strongest predictors of language-learning outcomes, and a private, endlessly patient tutor removes the audience. You can also switch to push-to-talk, so nothing is recorded until you press the button. Low stakes, real accountability.
What’s free, and what’s in Pro?
Structured lessons, chat, and vocabulary review are free. Parla Pro unlocks the live voice tutor for unlimited real-time speaking practice, with a 7-day free trial.
Say your first sentence today
Not after 300 lessons. Not once the tree is golden. Today — out loud, in a language you love.
7-day free trial on Pro · Cancel anytime · No streak guilt, ever