Translate an XLIFF file into Korean

Particles, word order and inline tags

Korean is verb-final and attaches particles to nouns, and their form depends on the last syllable: 을/를 for objects, 이/가 for subjects, 으로/로 for direction. When the noun is a placeholder, its ending is unknown at translation time, so Delete <ph id="1"/>? cannot take a guaranteed particle. Two common Korean conventions handle this: the dual form <ph id="1"/>을(를) 삭제할까요?, or a phrasing with no particle at all, such as 삭제할 항목: <ph id="1"/>. Check which one was used and whether it suits your UI.

Word order is the second constraint. Inline tags are kept in the same order as in the source, so the Korean sentence is shaped around that order. For Click <g id="1">here</g> to open <g id="2">the report</g>, the most natural Korean would mention the report first; a version that respects tag order reads like <g id="1">여기</g>를 클릭하면 <g id="2">보고서</g>가 열립니다. Note that the particle sits after the closing tag.

Korean has a single CLDR plural category, other. Numbers take a counter word instead: 파일 3개 for items, 3명 for people, <x id="count"/>건의 새 메시지 for messages. If your XLIFF carries ICU messages with one and other branches, both will hold the same Korean text; only other is actually needed.

Speech level, punctuation and on-screen length

Korean verb endings encode politeness. The formal -습니다 style (저장되었습니다) suits system messages and business software; the polite everyday -요 style (저장했어요) is common in consumer apps. Buttons and menu items usually take a bare noun form: 저장, 취소, 설정, not a full sentence. Mixing styles within one interface is immediately noticeable to Korean users.

Manuce does not take style instructions, so the way to check the speech level is the free sample, then a read-through of the delivered file. Notes already in the file (<note> in 1.2, <notes> in 2.0) are passed to the translator as context: a note saying "button" or "error message" helps decide between 저장 and 저장하시겠습니까?.

Korean punctuation follows Western usage, with no space before ? or !; a French source such as Enregistrer ? becomes 저장할까요?. On length, a Korean string often has fewer characters than its English source, but each hangeul syllable block is wider than a Latin letter, so character counts alone will not tell you whether a label fits. Korean separates word groups with spaces; if your UI breaks lines mid-word, CSS word-break: keep-all is the usual fix on the code side.

What Manuce returns, and what it does not do

• The same XLIFF file, 1.2 or 2.0: every source translated into its target, with the state attribute updated.
• Indentation, element order, quotes, comments, line endings, encoding and BOM preserved; only the values change.
• Units marked translate="no" are left alone.
• <g>, <x/> and <ph> tags kept in the same order; a string that would lose one is rejected and flagged.
• Every Korean value is checked against its source for placeholders and tags; if a mismatch persists, the source value is kept and the report says so.
• A report listing translated strings, strings left untouched and the reason.

Not translated: IDs, empty values, URLs, email addresses, numbers, colours and date formats. A pattern such as yyyy. M. d. or yyyy년 M월 d일 therefore belongs to your locale handling (for example ko-KR), not to the translation.

Manuce does not test rendering in your app, does not move tags to follow natural Korean order, and does not provide sworn or certified translation. Pricing counts the words in values only, tags excluded, with a ×1.6 coefficient for Korean.

Preparing the file and checking the result

1. Encoding: the file's encoding is preserved, so it must be able to hold hangeul. ISO-8859-1 cannot; convert to UTF-8 before uploading.
2. Target language: set target-language="ko" on <file> in 1.2, or trgLang="ko" on <xliff> in 2.0, as your import tool expects.
3. Do-not-translate: mark brand and product names that must stay in Latin script with translate="no".
4. Context: add short notes describing ambiguous strings (button, window title, error).
5. Free sample: check the speech level and how particles after placeholders are handled before you order.

Once the file is back, read the report, then test in the app with real values: one ending in a vowel and one in a consonant (사진, for instance) quickly reveal an awkward particle. Also look for truncation in narrow buttons and for counters after numbers.

In practice

1. Convert the file to UTF-8 and set ko as the target language.
2. Mark fixed terms with translate="no" and add context notes.
3. Upload the XLIFF, review the quote and check the free Korean sample.
4. Download the file and report, then re-import it into your project.
5. Test the UI with real values: particles, line breaks, truncation.