Bilingual Pages vs Lerna.ai

What each tool is

Lerna.ai

A voice-based AI conversation partner for language learners. You speak to the AI in your target language; it responds, corrects, and adapts to your level. The premise is that the bottleneck for adult learners is speaking practice — finding the time, courage, and patient interlocutors — and that an always-available AI removes the barrier. Web and mobile.

Bilingual Pages

A reading app with bilingual editions and tap-to-translate for imported EPUBs. The premise is that input volume — exposure to real language at scale — is what builds vocabulary, grammar intuition, and reading comprehension. Reading is the primary mechanism.

Side-by-side comparison

The cognitive science of input vs output

Modern second-language acquisition research distinguishes two systems that develop on different timelines and respond to different kinds of practice.

The receptive system: built by input

Vocabulary recognition, comprehension, grammar intuition, listening comprehension — the receptive side — develops through exposure. Reading, listening, and watching in the target language all contribute. The active ingredient is volume of comprehensible input; the more material you process, the more the system grows. This is what Krashen called the input hypothesis, and it's now well-established as the dominant mechanism for receptive skill development.

The productive system: built by output

Speaking and writing — producing language — depend on a different mechanism. Merrill Swain's output hypothesis (1985) and subsequent work argues that producing language forces a different kind of processing: you have to retrieve, structure, and articulate, which exposes gaps that input alone doesn't reveal. Output practice with feedback is how passive recognition turns into active production.

You need both — in roughly that order

The serious learner converges on a workflow: input first, then input-plus-output. You cannot produce what you have never encountered, so trying to speak before building a receptive base is inefficient. But once you have something to say, producing it is what cements it into active use. The two systems support each other.

What Lerna.ai is good at

Removing the social cost of speaking practice

The biggest barrier to adult speaking practice has always been the human one: finding patient partners, paying for tutors, tolerating the embarrassment of producing broken sentences. An AI removes all of this. You speak when you want, the AI doesn't judge you, and you can repeat the same exchange ten times without anyone getting bored. The lowered barrier translates directly into more practice volume.

Immediate corrective feedback

Modern AI tutors can give targeted corrections in real time — pronunciation, grammar, word choice, naturalness. This is one of the things human tutors do that other forms of practice (reading, watching, listening) can't. The feedback loop tightens the connection between attempt and refinement.

Topic-flexible conversation

Unlike scripted courses, AI conversation adapts to what you want to talk about. You can practise the vocabulary you actually need (work, hobbies, travel destinations) rather than the scenarios a textbook chose for you.

Where Lerna.ai has limits

AI conversation isn't native conversation

Current AI models produce fluent, grammatically clean target-language output, but they don't produce the messy, idiomatic, register-shifting speech of real native speakers. The vocabulary distribution and conversational rhythm are subtly off. This is fine — and usefully so — for practice. It isn't a substitute for real human conversation when that becomes available.

Limited input volume

A conversation exposes you to maybe 500–1,000 words in 30 minutes. A book exposes you to 4,000–6,000 words in the same time, with more lexical variety and grammatical complexity. AI conversation can't deliver the input volume that reading delivers; if you only have time for one, and you're below B1, reading is the better investment.

Requires a quiet space and confidence to speak

You can't do Lerna on the bus or in bed. The need to speak aloud is a real logistical constraint, and the energy required to do so is non-trivial — many learners find they can read for an hour but can only do focused conversation for 15–20 minutes.

What Bilingual Pages is good at

Everything on the input side. Reading is the highest-volume, lowest-friction way for adults to encounter their target language at scale. The bilingual layout and tap-to-translate workflow let you read material above your monolingual level without comprehension collapsing — which means you can be reading real literature within months of starting, rather than years.

But Bilingual Pages doesn't train output. You can't practise speaking with it. You can't get pronunciation feedback. The vocabulary you absorb sits in the receptive system; converting it to active use requires output practice that the app doesn't provide.

The combined workflow

For learners using both tools, a typical weekly shape:

1. Daily: 25–40 minutes of reading. The input engine. Build vocabulary and grammar intuition through bilingual editions or tap-to-translate.
2. Three to five times a week: 15–30 minutes of AI conversation. Active practice of what your reading is feeding into you. The conversations should pull from topics related to what you've been reading — vocabulary you encountered in books shows up in conversation, which cements it.
3. Periodically: real human conversation when accessible. Native speakers, tutors, language exchanges. The AI is preparation for these, not a replacement.

When Lerna.ai is the primary tool

• You're at B1+ already and the bottleneck is speaking, not understanding. Many adults can read better than they can speak; output practice is the asymmetric improvement.
• You have an upcoming need to speak — a trip, a job, a relationship. AI practice produces fluency faster than reading does.
• You don't enjoy reading. The whole input-via-reading premise only works if you actually do the reading; if you don't, conversation is your best alternative input source.

When Bilingual Pages is the primary tool

• You're at A2–B1 and the bottleneck is vocabulary and comprehension. Most learners are here for longer than they expect; reading is the dominant skill at this stage.
• You don't have a quiet space to speak aloud regularly. Reading is silent and portable.
• You enjoy literature and want to read books in your target language. AI conversation can't deliver this.
• You're building toward later conversation practice — the reading creates the vocabulary that the conversation later activates.

The verdict

Lerna.ai and Bilingual Pages aren't competing — they're the two halves of an adult-learner workflow. Bilingual Pages builds the receptive system through input. Lerna activates that system through output. Using both produces faster overall progress than either alone, and the right ratio shifts with your level: input-heavy at the start, output-heavy as you approach fluency. Pick one if you must, but the serious answer is that they're complementary tools for the same goal.