Modal-dialog localization is the work of translating dialogs, drawers, bottom sheets, pop-up panels, titles, buttons, helper text and focus states so users in every language understand what the interface is asking them to do and can complete or dismiss the interaction safely. People searching for how to localize modal dialogs, translate bottom sheets, internationalize drawers or adapt pop-up panels are solving an action-and-focus problem: words and layout can change, but the user must remain in the same interaction state.
A fluent translation can still break a modal flow if it makes the primary and secondary actions sound equivalent, turns Close into Cancel when data is saved differently, hides whether a sheet can be swiped away, changes which field receives focus, pushes the main action off screen or causes a screen reader to read background content before the dialog. These are interaction defects, not just wording problems.
This guide explains a practical system for professional dialog, drawer and bottom-sheet localization: define the interaction purpose, preserve primary and secondary action meaning, keep dismissal behavior clear, localize titles and helper text without changing state, manage focus and keyboard order, handle long labels responsively, support mobile gestures and right-to-left layouts, test accessibility semantics, and verify that the same action produces the same result in every locale.
1. Dialog purpose
For dialog purpose, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test dialog purpose with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
2. Modal state
For modal state, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test modal state with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
3. Nonmodal panels
For nonmodal panels, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test nonmodal panels with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
4. Dialog titles
For dialog titles, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test dialog titles with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
5. Primary actions
For primary actions, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test primary actions with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
6. Secondary actions
For secondary actions, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test secondary actions with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
7. Close actions
For close actions, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test close actions with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
8. Cancel actions
For cancel actions, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test cancel actions with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
9. Save actions
For save actions, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test save actions with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
10. Apply actions
For apply actions, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test apply actions with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
11. Done actions
For done actions, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test done actions with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
12. Back actions
For back actions, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test back actions with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
13. Dismissal rules
For dismissal rules, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test dismissal rules with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
14. Outside click
For outside click, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test outside click with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
15. Escape key
For escape key, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test escape key with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
16. Swipe dismissal
For swipe dismissal, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test swipe dismissal with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
17. Drawers
For drawers, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test drawers with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
18. Bottom sheets
For bottom sheets, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test bottom sheets with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
19. Full-screen sheets
For full-screen sheets, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test full-screen sheets with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
20. Form fields
For form fields, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test form fields with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
21. Initial focus
For initial focus, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test initial focus with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
22. Focus trap
For focus trap, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test focus trap with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
23. Focus return
For focus return, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test focus return with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
24. Keyboard order
For keyboard order, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test keyboard order with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
25. Screen-reader labels
For screen-reader labels, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test screen-reader labels with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
26. Aria descriptions
For aria descriptions, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test aria descriptions with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
27. Long titles
For long titles, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test long titles with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
28. Long buttons
For long buttons, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test long buttons with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
29. Button wrapping
For button wrapping, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test button wrapping with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
30. Small screens
For small screens, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test small screens with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
31. Large text
For large text, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test large text with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
32. Scrolling content
For scrolling content, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test scrolling content with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
33. Sticky actions
For sticky actions, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test sticky actions with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
34. Loading state
For loading state, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test loading state with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
35. Success state
For success state, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test success state with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
36. Error state
For error state, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test error state with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
37. Empty state
For empty state, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test empty state with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
38. Nested dialogs
For nested dialogs, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test nested dialogs with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
39. Confirmation dialogs
For confirmation dialogs, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test confirmation dialogs with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
40. Information dialogs
For information dialogs, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test information dialogs with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
41. Selection dialogs
For selection dialogs, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test selection dialogs with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
42. Right-to-left layout
For right-to-left layout, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test right-to-left layout with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
43. Worked example
For worked example, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test worked example with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
44. Acceptance testing
For acceptance testing, localization should begin with the interaction state and expected outcome. Identify what opened the panel, what information it contains, which actions commit a change, which actions merely dismiss the surface and where focus should move afterward. A polished translation is useful only when it preserves those relationships.
A common failure is to choose a natural synonym that changes the action model. Close, Cancel, Back, Done, Save and Apply can look similar in a short dialog, but they may persist different data or return users to different places. Review the target wording together with the actual button behavior rather than as isolated strings.
Test acceptance testing with pointer, keyboard and assistive technology, and repeat on narrow screens where translated text wraps. Confirm which control receives focus, whether background content remains unavailable while a modal is active, and where focus returns after dismissal. This exposes interaction errors that a static translation review cannot see.
The practical standard is state equivalence. The same user intention should lead to the same committed change, cancellation, selection or return point in every language. Layout can adapt and button labels can become longer, but localization should not change the interaction contract or make the intended action harder to discover.
45. Final Operating Checklist
Define each dialog’s purpose, state and dismissal rules. Keep primary, secondary, close, cancel, save, apply and done actions distinct. Preserve focus entry, focus trapping and focus return, test long text and sticky action areas, verify mobile gestures, keyboard navigation, screen-reader semantics and right-to-left layout, and compare resulting state after every action. Any translation that changes what a dialog action does or causes users to lose their place should be treated as a high-severity defect.