Short labels expand proportionally more than long passages

What it is

The short-string expansion effect describes the tendency for buttons, menu items, and status labels to grow by a larger proportion in translation than long prose. Turning one word into two or three produces a large ratio, whereas adding the same words to a paragraph barely changes its relative length. These strings are risky because their original allocation is usually tight, not because they are the longest text on the page.

Why it happens

Short labels have little base length over which to dilute translation overhead. A compressed compound, abbreviation, or omitted argument may require explicit grammatical relations, an object, or a politeness form in the target language. That minimum phrasing cost dominates a one-word label but not a paragraph. Controls are also commonly constrained to one line, so wrapping cannot absorb the increase.

Studying it

Pair production source strings with translations, bin them by source length, and compare character ratios, absolute rendered width, and overflow. A tiny denominator mathematically amplifies any ratio, so a high proportional change alone does not establish a linguistic mechanism; report absolute growth and UI failure alongside it. Stratify by control, font, and target language, use medians and upper percentiles, and flag abbreviations, names, and untranslated brand tokens.

Where it stops holding

This is a distributional tendency, not a promise about every label. A target language can be more compact, a brand term may remain unchanged, or an idiomatic verb may be shorter. A high ratio also does not imply the greatest absolute width: a long heading can still occupy more space despite expanding less proportionally. Thresholds observed for one language pair do not transfer automatically to another.

Applying it

• Reserve more proportional slack for actions, tabs, navigation, and badges than for flowing prose.
• Prefer layouts that can reflow; where space is genuinely scarce, work with translators on idiomatic equivalents without deleting the object or consequence.
• Expand short strings more aggressively in pseudolocalization while preserving recognizable word boundaries.
• Segment screenshot-test failures by source length and verify that the shortest bin has no systematic overflow or semantic loss.

Related

• Same group: S1.01.1 Translations of the same content vary substantially in length across languages · S1.01.3 Containers fixed to source-language length inevitably overflow · S1.01.4 Layouts must be tested against the longest language, not the average · S1.01.5 Truncation hides overflow; it does not solve it
• Adjacent: S1.07 Writing for translation · T2.01 Buttons and action copy
• Search terms: short string expansion · UI localization · rendered text width