Guided Path vs Flashcards for Learning Chinese
One format decides what you study and runs out. The other never runs out and decides nothing. Most people need both, in a specific order.
Ask what to use for Chinese characters and you'll get two answers that don't engage with each other. Course apps and flashcard systems solve different problems, and picking the wrong one for your current problem is the most common reason people stall.
The distinction that matters: a guided path answers what should I learn next. A spaced-repetition system answers what should I review today. These are separate questions and a tool that's excellent at one is usually mediocre at the other.
Below is what each format actually does, where each one fails, and a rule for deciding which problem you currently have.
Two Different Jobs
| Guided path / course | Flashcards / SRS | |
|---|---|---|
| Answers | What to learn next | What to review today |
| Sequencing | Designed for you | Yours to design |
| Ceiling | Ends, usually around HSK 3-4 | None |
| Setup cost | Near zero | Hours, then ongoing |
| Failure mode | You finish it and stop | You never start, or the deck rots |
| Best at | Getting moving; the first 1,000 characters | Retention; the long tail past 1,500 |
| Worst at | Anything past its last unit | Telling a beginner where to begin |
Read the last two rows together and the answer to most people's question falls out. If you don't know where to start, a deck is the wrong tool. If you know exactly what you need and keep forgetting it, a course is the wrong tool.
What Guided Paths Get Right
- The sequencing problem is solved for you. Deciding what order to learn 3,000 characters in is genuinely hard, and getting it wrong is expensive.
- Progress is legible. Unit 40 of 391 is a fact. "I have 2,400 cards in my deck" is not progress information.
- Nothing arrives cold. A well-built path introduces a character with its meaning, sound, and structure before it tests you.
- Momentum is the actual product. Most people quit Chinese in the first three months, and a path's real function is getting you past that.
A path's value is highest at the start and declines steadily. By the time you have 1,500 characters, you mostly know what you need to learn next -- and the path's ceiling starts costing you more than its sequencing gains you.
What Flashcards Get Right
Spaced repetition is the best-evidenced technique in the whole of language learning. Scheduling a review just before you'd forget something is measurably more efficient than reviewing on a fixed timetable, and the effect is large.
- No ceiling. A deck goes to 5,000 characters as happily as 500.
- Retention is the hard part, and this is the fix. Learning a character takes minutes; keeping it takes years of small, well-timed reviews.
- Total control over card format. You decide whether you're tested on meaning, reading, writing, or all three.
- It works for anything. Vocabulary, measure words, sentence patterns, tone pairs.
The counterpart is that a flashcard system is a mechanism, not a curriculum. It will faithfully schedule whatever you put in, including a badly chosen 3,000-character list in a useless order.
The Failure Modes, Honestly
How Paths Fail
- They end. You reach the last unit somewhere around HSK 3 and have nowhere to go, which is the single most common complaint about course apps.
- They test recognition. Four-option multiple choice and tile-matching feel like testing and are much easier than producing an answer from nothing.
- They teach words, not characters. You learn 电话 as a unit and can't read 电 in an unfamiliar compound.
- They won't let you skip. Most trees are linear, so knowing HSK 1 already doesn't save you from grinding through 一.
How Decks Fail
- The blank-deck problem. You wanted to study Chinese and now you're designing a syllabus. Most people never finish the design.
- Review debt. Twenty new cards a day compounds into an hour of reviews within a month, and then you skip a week and it becomes unbearable.
- Bad cards drill the wrong thing. A character-to-English card trains recognition only, and leaves you unable to produce anything.
- No teach step. A card whose answer you've never seen is a guaranteed miss, and repeated guaranteed misses are how decks get abandoned.
The review-debt spiral is the killer. It isn't a discipline problem -- it's arithmetic. Adding new cards faster than you retire old ones has exactly one outcome, and capping daily new cards is the only real fix.
Which Problem Do You Have Right Now?
| Your situation | Your actual problem | Reach for |
|---|---|---|
| Starting from zero | You don't know what to learn or in what order | A guided path. Don't touch a deck yet. |
| Finished a course app, ~700 characters | Coverage stopped; recall was never trained | A path that continues past HSK 3, plus recall-based review |
| 1,500+ characters, forgetting constantly | Retention, not sequencing | SRS, with a hard cap on new cards |
| Reading but stuck on vocabulary | Encounter order left gaps | A systematic list to fill gaps, reviewed by SRS |
| Exam in three months | Neither -- you have a syllabus already | SRS over the official list, plus practice papers |
Notice that only the first row is a pure answer. Every other row wants both mechanisms, which is why the versus framing is a bit false.
How to Combine Them
The combination that works treats the path as the input and the SRS as the storage. New material comes from a sequence someone designed; keeping it is the scheduler's job.
Path decides what
SRS decides when
Production, not recognition
- Cap new material. Five to ten new characters a day is sustainable. Twenty is not, whatever the app suggests.
- Never add a card for something you haven't been taught. Encode first, schedule second.
- Order reviews by how overdue they are, not by how easy they are. A 30-day card a month late is more urgent than a 1-day card due this morning.
- Let mature cards retire. Without a graduation rule, the deck grows without bound and eventually eats your whole study session.
- Skip what you know. A placement test or a skippable checkpoint is worth hours.
If your tool doesn't let mature cards leave rotation, add that rule yourself -- retire anything you've got right twice at your longest interval. An unbounded deck is the single biggest cause of abandonment.
What This Looks Like in Practice
HanziFeed is built as the combined shape rather than one or the other: a 391-unit path supplies the sequence, and everything it teaches grades through a seven-bucket review schedule ordered by relative overdueness, with cards graduating out after two clean reviews at the top interval. Checkpoints every fifth unit can be sat cold to skip a whole group.
That's one implementation. You can assemble the same thing from a course app plus Anki, and our comparison with Anki covers where Anki's flexibility is genuinely worth the setup cost. The important part is having both halves, not which tools supply them.
Frequently Asked Questions
Can I just use flashcards from day one?
Is Anki better than a built-in review system?
How many new characters a day is realistic?
What happens if I take two weeks off?
Do I need both if I'm only preparing for an exam?
Are multiple-choice reviews useless?
A guided sequence and a review engine, in one place
HanziFeed's 391-unit path feeds a seven-bucket review schedule that surfaces what you're about to forget -- six review modes including handwriting and typed pinyin, free.