HSK Preparation: Organizing Chinese Grammar in Notion

Most HSK candidates fail the grammar section not because the patterns are hard, but because their notes are a mess. Grammar points live in textbook margins, screenshots, and three different apps. When the exam approaches, there's no single place to review them by level. A Notion database fixes this. Here's the exact structure I use to prep for HSK, and how to run it so the patterns actually stick.

Why a database beats linear notes

A notebook forces grammar into the order you happened to learn it. But HSK is organized by level, and review is most efficient when you can slice points by what you're weak on. A database lets you do both: store each grammar point once, then view it filtered by level, by mastery, or by the structure it teaches.

The core idea is one row per grammar point. Not one row per lesson, not one row per textbook chapter. One pattern, one row. That granularity is what makes filtering and spaced review possible later.

The properties that matter

Create a new database (full page) and add these properties. Resist the urge to add more until you've used it for a week.

• Pattern (title): the structure itself, e.g. 是…的, 把 + object + verb, 越来越 + adjective.
• HSK Level (select): 1 through 6. This is the spine of the whole system.
• Function (text): what the pattern does in plain English, e.g. "emphasizes time, place, or manner of a completed action."
• Example (text): one Chinese sentence using the pattern, with pinyin and translation.
• Mastery (select): New / Learning / Solid. This drives your review queue.
• Confusable with (relation, self-referencing): links to other patterns you mix up, like 了 vs 过.
• Last reviewed (date): so spaced repetition has something to sort on.

The self-referencing relation is the underrated one. Most HSK grammar mistakes are confusion between two similar patterns, not ignorance of one. Linking 在 to 正在, or 才 to 就, surfaces those pairs every time you open either row.

Building the level views

Once you have rows, create one filtered view per HSK level you're targeting. For HSK 4, the view filters HSK Level is 4. Use a board (kanban) view grouped by Mastery so each level becomes three columns: New, Learning, Solid. You literally drag a card right as a pattern moves from "I just met this" to "I can produce it under time pressure."

This board is the single most motivating part of the setup. Watching the Solid column fill up as the exam approaches is far more concrete than a percentage in an app.

A review workflow that fits HSK

The database is only useful if you touch it daily. Here's the loop:

1. Add as you learn. Every new pattern from a textbook, video, or tutor goes in as a row the same day, marked New.
2. Daily review (10 minutes). Open a view filtered to Mastery is Learning sorted by Last reviewed ascending. Review the oldest first. For each, cover the example, recall the structure, then check. Update Last reviewed.
3. Weekly promotion pass. Move patterns you can produce without hesitation from Learning to Solid. Demote anything you blanked on back to Learning.
4. Pre-exam sweep. Two weeks out, filter to your target level and Mastery is not Solid. That's your shortlist. Everything else is maintenance.

The Last reviewed date gives you crude spaced repetition without a plugin. Sorting ascending always shows you the patterns you've neglected longest, which is exactly where forgetting happens.

Capturing confusable pairs

When you get a practice question wrong, don't just note the right answer. Open both patterns and link them via Confusable with, then add a one-line note in each Example field on how to tell them apart. For 了 vs 过: 了 marks completion or change of state; 过 marks past experience ("have done at least once"). Writing the distinction in your own words, in the row, is what converts a wrong answer into a fixed weakness.

Over a few weeks these links form a small graph of your personal trouble spots. Filtering to rows that have a Confusable relation gives you a targeted drill set that's worth more than any generic worksheet.

Keeping it from becoming a graveyard

The failure mode for any study database is over-building. People spend an afternoon on formulas and color-coding, then never add a real grammar point. Avoid this:

• Start with the seven properties above. Add nothing else for two weeks.
• Don't import a giant grammar list on day one. Add patterns as you encounter them in study, so each row carries a memory of context.
• One example per row is enough. More examples feel thorough but slow down review.

Adapting it across levels

The same database scales the entire HSK ladder. As you advance, you don't rebuild anything; you just add higher-level patterns and create a new level view. By HSK 5–6 the Confusable relations become the heart of your prep, because at that stage almost every error is a fine distinction between near-synonymous structures.

If you want a head start, you can replicate this structure in about fifteen minutes from the property list above. The point isn't the template; it's the discipline of one pattern per row, reviewed oldest-first, with confusable pairs linked. Do that consistently and the grammar section stops being the part of HSK you dread.