One-handed layout and font evidence
Jianlai candidate-size scopes: separate labels, annotations and composition from candidate text
Test the separately listed candidate-text, candidate-number, annotation and encoding-area size controls one at a time, holding font and theme fixed. Deliver a scope ledger without requiring every scheme to display annotations or numbers or treating setting values as measured screen pixels.
On this page
Test the separately listed candidate-text, candidate-number, annotation and encoding-area size controls one at a time, holding font and theme fixed. Deliver a scope ledger without requiring every scheme to display annotations or numbers or treating setting values as measured screen pixels.
1. Establish a repeatable input card
Confirm Jianlai identity, Chinese scheme, font and theme in a private ordinary note, then enter and save the complete card below. Use a separate line for candidates, producing the same Chinese target through a familiar scheme while recording actual input and visible choices. Do not install an unknown dictionary merely to create annotations or assume every scheme shares one code. Committed baseline text and pending candidate composition are different states. Recreate conditions after settings navigation rather than assume composition survives or reuse an old screenshot as a new-size observation.
山河 F34
2. Map four display categories by their actual labels
The official image lists candidate text, candidate numbers, candidate annotations and encoding-area sizes separately, with key-font size explicitly labelled as a percentage. Record installed labels and original values, then map them to real regions. An absent category is not applicable; document line numbers are not candidate numbers. Eighteen and twelve in the image are examples, and unlabelled units must not become invented pixels or points. Labels suggest scope, but effects need observation in the extension. A changed settings number does not establish every region grew. Keep font, key size and theme unchanged.
3. Change only candidate text for the first trial
Choose one actually available neighboring candidate-text size and apply it, then regenerate the target in the original host. Record outline size, visible word extent and occupied area, separately checking numbers and composition where observable. A changed candidate order is not automatically a size algorithm. Text size may affect fitting, so fewer visible choices do not establish a smaller dictionary. Keep setting and observation columns separate. Restore the recorded original and verify it before another category, rather than accumulate later changes and lose attribution.
4. Give number and annotation trials their own prerequisites
Only if candidate numbers are actually visible, change their size alone and inspect number/text distinction. A larger number does not establish larger candidate wording. Restore it before an annotation trial. If recognizable annotations exist, change that size alone and record location. Otherwise mark the trial not applicable rather than invent pinyin, meanings or translations. Similar size controls do not share an object; recreate actual candidates and conditions each round. Follow labels when installed column order differs from the image, not old coordinates. Restore a known value after an accidental change before continuing.
5. Observe composition separately from document text
If an encoding/composition region is recognizable, change only its size, enter the target’s actual code without committing and record readability and location. Committed 山河 in the host is not the code, and not every scheme displays Latin composition in the same keyboard position. Record actual locations instead of inventing a row. Commit afterward and check the card’s text, separating host styling from composition-region observations. Restore the original code size and recheck known candidate, number and annotation settings; one regional comparison is not a complete size restoration.
6. Use numeric ratios as paper controls only
On paper, eighteen to twenty adds two, about eleven-point-one percent relative to eighteen; twelve to fourteen also adds two but about sixteen-point-seven percent. Equal increments differ proportionally, yet those control-value ratios are not measured screen glyph heights or touch areas. Substitute actual values where the examples are unavailable rather than search for hidden numeric entry. Key-font one hundred percent and candidate value eighteen are unlike units, not an eighty-two-pixel difference. Keep calculations separate from actual settings as a control for independent parameters, not an unmeasured app scaling formula.
18 -> 20: +2; relative increase 2/18 = 1/9
12 -> 14: +2; relative increase 2/12 = 1/6
Control-value ratios are not measured glyph-height ratios.
7. Diagnose absent objects separately from unapplied settings
For no visible change, first check whether the controlled object exists in this scheme, then extension identity, font and application. No annotation is not a failed annotation-size setting. Different input content calls for scheme/sample checks before every size is changed. Record crowded regions and word length and restore a reference value rather than clear dictionaries or themes. Correct one item per round and retain committed source. Deleting long choices for a tidy screenshot is not improved layout evidence. Unobserved effects remain unknown, with no claims about prediction accuracy, speed or eyesight protection.
8. Restore the four values and deliver a scope matrix
Recheck restored values for every performed candidate-text, number, annotation and encoding trial, with key percentage unchanged, then commit the same card in an ordinary note. Pass reopening and setting checks only if performed, preserving not-applicable explanations. Deliver original values, one change per trial, actual objects and restoration, distinguishing known units from unmeasured ones. Another scheme may expose different objects and needs a new matrix. The outcome separates four display roles rather than enlarging everything uniformly or providing an automatic keyboard-accessibility evaluator.
Related guides and sources
Font-size basics; Input and composition boundaries.
Checked on 18 November 2026: Apple’s Finland lookup and China lookup both list Jianlai 1.0.5, requiring iOS / iPadOS 16.0 or later. The official store describes themes, backgrounds, fonts and multilingual input. Use controls actually present in the installed version; system editing and host processing are distinguished separately. These are repeatable instructions and paper expectations, not claims of completed device input, host compatibility or format-parser tests.
Method references: Official Apple keyboard-settings image at full reading size.