A bilingual award programme is usually scoped as a translation project. Translate the entry form, translate the guidelines, add a language toggle, done. That framing is why so many of them are quietly unfair.
Translation is the easy part. The hard part is that two languages do not carry information at the same density, judges do not read both equally, and a rule that is identical in wording can be materially different in effect. These are design decisions, not linguistic ones, and they have to be made before entries open.
On this page
- The word-limit trap
- Judging across languages
- What breaks in the interface
- A bilingual award programme checklist
The word-limit trap
“Maximum 500 words” looks like one rule applied evenly. It is not. Chinese is not word-delimited in the way English is, so a word count applied to Traditional Chinese is either meaningless or silently measuring something else entirely — usually whatever your text field decided to count.
The industry conversion is well established: roughly 1,000 English words corresponds to 1,300–1,800 Chinese characters, and 1,000 Chinese characters comes back as roughly 650–750 English words.
English entrant
500 words
A well-understood constraint. Writers know roughly what 500 words holds.
繁體中文 entrant
≈ 650–900 characters
The equivalent capacity. Impose “500” as a character limit instead and you have cut their submission to roughly a third.
The fix is to state limits separately and explicitly: a word count for English, a character count for Chinese, set at genuine equivalence rather than at the same number. Then enforce each in the field the entrant is actually using. This sounds obvious written down. It is wrong in a large share of the bilingual forms we have reviewed.
Judging across languages
This is the harder problem, and the one most programmes handle by not thinking about it.
If entries arrive in both languages and judges read whichever they can, an entry’s score depends partly on which judges were available to read it. If entries are translated for judging, some entries are assessed as written and others through an intermediary — and translation quality becomes a factor in the result. If judges are asked to work in their second language, fluency differences become scoring differences.
There is no costless option here. There are three defensible ones.
| Approach | What it costs |
|---|---|
| Language-matched panels Chinese entries judged by a Chinese-reading panel, English by an English-reading panel | Creates two panels that must be normalised against each other, since severity cannot be assumed equal. Needs anchor entries scored by both. |
| Professional translation before judging All entries presented to judges in one language | Expensive, slow, and introduces a translator between the entrant and the decision. Requires disclosure to entrants. |
| Fully bilingual panel Every judge reads both languages competently | Cleanest in principle. Sharply limits who you can recruit, which may cost you sector expertise. |
Language-matched panels are the common choice in Hong Kong, and they make cross-panel comparison unavoidable — which brings you straight into the territory covered in our piece on judge score normalisation, including the anchor-entry requirement. A bilingual programme with language-split panels and no linking design cannot defend its cross-language rankings.
Whichever you choose, publish it. Entrants are entitled to know whether their submission will be read in the language they wrote it in. This belongs in the programme rules alongside the scoring criteria, not discovered afterwards by someone who lost.
What breaks in the interface
A set of failures that appear reliably, usually late, usually at deadline.
Layout breaks on short strings, not long ones. The W3C’s guidance on text size in translation makes the counter-intuitive point that expansion is proportionally worst on the shortest strings — button labels, field names, table headers. These are exactly the elements sitting in fixed-width containers. Long paragraphs reflow; a button does not.
PDF generation fails on CJK fonts. Entry confirmations, judge score sheets and certificate exports routinely render Chinese as boxes or blanks, because the PDF library defaults to a font with no CJK coverage. This is invisible in testing if the test data is all English — and test data usually is.
Sorting produces nonsense. Alphabetical sorting of Chinese names by byte order is meaningless to a human. Decide deliberately whether to sort by stroke count, by romanisation, or by entry number, and expect that any choice will feel wrong to somebody.
Search misses across scripts. An administrator searching for an organisation by its English name will not find the entry filed under its Chinese name unless you have deliberately indexed both. In practice most institutional entrants have both, and staff use them interchangeably.
Notification templates drift. The English templates get updated when a deadline moves; the Chinese ones do not. Then Chinese-language entrants receive the old date. This is a process failure rather than a technical one, and it is the most common of the lot.
A bilingual award programme checklist
- Length limits stated separately for each language, set at real equivalence, enforced per field
- Your cross-language judging approach chosen, documented, and published in the rules
- Anchor entries scored across language-split panels, if you use them
- Test data in Traditional Chinese, used throughout testing rather than added at the end
- Every PDF export verified with Chinese content before launch
- Both name fields indexed for administrator search
- A single owner responsible for keeping both template sets in sync
- Your privacy notice available in both languages, as covered in our PDPO compliance guide
Genuine bilingual capability is one of the reasons Hong Kong programmes have historically concluded that international platforms do not fit them, and commissioned bespoke builds instead — a pattern we documented in our audit of the local market. The conclusion was reasonable when the alternatives genuinely could not handle Traditional Chinese properly. It is worth re-testing rather than assuming.
AwardScience is built bilingual rather than translated, with English and Traditional Chinese treated as equals throughout the entrant and judging experience. Book a live demo or see pricing.
