Verb-first labels are the most scannable

What it is

The verb-first action-label heuristic places the operation early and consistently when the target language and control context permit. In a command list for one object, “Export report,” “Archive report,” and “Delete report” expose their difference immediately. This is not a universal instruction to place a grammatical verb at the start of every language, nor does it cover headings, state labels, object navigation, or all controls. The aim is early visibility of action category and item difference, not one global word order.

Why it happens

When people inspect menus, toolbars, and bulk-action lists, they use leading words, length, icons, and parallel structure to narrow candidates instead of reading every item deeply. Repeating a shared object first delays the distinguishing information. Where a verb–object imperative is natural, leading with a concrete operation can create a clear parallel set. The benefit still depends on natural syntax, truncation, whether context names the object, icon familiarity, and list organization. In a verb-final language—or when objects must be distinguished first—mechanical fronting can make labels less natural and harder to understand.

Studying it

Create command-search tasks in the actual target language and compare verb-early, object-early, and other natural constructions. Measure finding time, wrong choices, fixation path, and consequence comprehension. Control length, item position, icons, familiarity, and risk, and separate same-object from mixed-object lists so information architecture is not mistaken for word order. Include supported word orders and writing directions, with native review of every variant. Results choose a pattern for that locale and component; one language, menu, or small sample does not establish a universal reading law.

Where it stops holding

Not every clickable label names an action. Nouns can naturally open object spaces such as “Billing”; switches often label a state; dialog actions form meaning with their prompt. Verb-final, richly inflected, and argument-dropping languages need local rules for information placement. A short button can rely on adjacent context for its object, but high-risk actions cannot omit scope or consequence merely to optimize the first word. Platform conventions and a complete accessible name can matter more than visual head-word consistency.

Applying it

• Classify control roles first. Evaluate action position only for controls that trigger operations, excluding object navigation, states, switches, and headings.
• In a same-object command set, try a natural parallel action structure that reveals differences before truncation. If objects differ or the language is verb-final, use object terms, qualifiers, icons, or grouping for early information scent.
• Maintain native-reviewed action-label patterns per locale instead of translating a source verb-first template word by word. Generate complete messages in local syntax rather than reordering concatenated fragments.
• Test recognition and misselection at production widths, zoom, keyboard, and screen-reader conditions. Adopt verb-first only where it helps that language and list without harming naturalness.

Related

• Same group: T1.02.2 Nominalization hides the action about to happen · T1.02.3 Verbs must be specific enough to judge consequences
• Adjacent: S1.06.1 Word order differs across languages · T3.01.1 Users scan documentation, they don't read it
• Search terms: verb-first action label · menu scanning · action-label syntax