Metlivi Blog

How to Find One Chinese Character in a Mixed Chinese-English Journal

If your journal returns no results for a Chinese character, first check which app and search mode you are using. Then test a known note with a one-character query, a two-character term, and an English word. Compare the results with the app’s documented behavior before changing settings or assuming the character is missing. Search engines differ: some use word boundaries, some offer a separate basic search, and some documentation does not specify how Chinese text is matched.

September 30, 20264 min readEveryday Aesthetics & Self-ExpressionBy Metlivi Editorial Team
Section 1

Start with the app’s own search documentation

Search is an app feature, not a universal rule. A journal might use a full-text index, a basic text scan, or more than one mode. Even products built on SQLite can differ in their SQLite extension, tokenizer, and app-specific handling. For example, [Joplin documents its use of SQLite FTS4](https://joplinapp.org/help/apps/search/) and says it switches to a custom mode for languages that do not use Latin word boundaries, including Chinese. Its page also distinguishes exact-term matching from substring matching in the documented FTS mode. These details describe Joplin; they do not establish how another journal behaves.

Look for the exact product and platform version, then note whether the manual describes full-text search, a basic mode, token or word matching, substring matching, language-specific behavior, and indexed content such as attachments. Don’t infer that a search box scans every visible character just because it accepts Chinese input. Conversely, don’t assume that an index will reject Chinese text: the app may have a language-aware mode or another implementation.

Section 2

Build a small control note

Use one known note as a controlled check. For example, create a fictional note whose body is: “山水 — 北京 — journal”. Here, 山 is a known single character; 北京 is a known two-character term; and journal is an English control word. Save the note, allow any documented indexing delay, and search each item separately. Record whether the note appears for 山, 北京, and journal. These are proposed checks, not results from testing any real app.

If journal does not find the note either, first check that the note is saved, included in the selected account or notebook, and in the searched fields. If journal works but neither Chinese query does, the issue is more specific to CJK handling or to how the search terms are interpreted. If 北京 works but 山 does not, record that distinction: it may point to a minimum-term or token-boundary behavior, but the result alone does not identify the cause.

Section 3

Compare full-text, basic, and substring search carefully

In a full-text search, the app may split text into terms and match those indexed terms. A substring search instead looks for a character sequence inside larger text. These are different behaviors: matching a whole term does not necessarily mean matching every shorter sequence within it. Joplin explicitly documents this distinction for its FTS queries and says its CJK custom mode avoids reliance on Latin word boundaries. Its documented slash prefix switches to basic search for cases FTS handles poorly, though Joplin warns that this mode can be much slower. Use that syntax only in Joplin and only as described in its current manual.

Database documentation can explain why implementation details matter without predicting an app’s results. SQLite’s [FTS5 documentation](https://sqlite.org/fts5.html) describes configurable tokenizers, including Unicode61 and trigram options. Joplin’s cited documentation, however, names FTS4. FTS5 examples therefore cannot be used to claim that Joplin, or every journal, uses those tokenizers. Check the app’s own documentation before applying database-level explanations.

Section 4

Diagnose the pattern, not just the failed query

Use this sequence to narrow the problem:

Do not transfer one app’s search operators or conclusions to another. Apple’s [Notes user guide](https://support.apple.com/guide/notes/search-your-notes-not18ab658ed/mac) describes searching note bodies and several attachment types, and notes that available features can depend on the account. It does not specify Chinese token boundaries or guarantee single-character substring retrieval. When official documentation leaves that point unstated, the reliable conclusion is simply that the behavior is undocumented there.

Confirm that a known note appears for its English control word.
Search the known two-character Chinese term, then the known single character, as separate queries.
Check whether the current scope includes the note and its body, and whether the app indexes only certain content types.
Consult the manual for a basic-search mode, CJK-specific behavior, token matching, or substring support. Try an alternate mode only with the app’s documented syntax.
If only one-character searches fail, report that exact pattern to the app’s support channel with the platform, version, query, and a minimal reproducible note. Avoid sharing private journal content.
Section 5

What the control searches can tell you

The control note helps separate a missing note or wrong search scope from a language-specific search issue. It does not prove how the app indexes every note, attachment, account, or language. Keep the exact app, platform, version, search mode, and three query outcomes together; that compact record is more useful than saying only “Chinese search is broken.” If you need reliable one-character lookup for a real archive, confirm the behavior against your own representative notes before reorganizing or reimporting them.

Related reading

Keep exploring this topic