Khmer Spell Checker

This free Khmer spell checker reads your Khmer text, splits it into words and underlines the ones that look wrong. Khmer is written without spaces between words, so the checker first works out where each word starts and ends, using a list of about 45,000 Khmer words from the Chuon Nath dictionary (វចនានុក្រមសម្ដេចសង្ឃរាជ ជួន ណាត) and modern Khmer–English word lists, with word frequencies to pick the most likely split.

Click an underlined word to see suggestions such as សម្រាប់ for សំរាប់ or កម្ពុជា for កម្ពជា, and replace it with one click. The tool also finds letters typed in the wrong Unicode order, which look right on screen but break search and sorting, counts Khmer words, and can insert zero-width spaces so long Khmer lines wrap neatly. Everything runs in your browser; your text is never uploaded.

How to use the Khmer Spell Checker tool

1. Type or paste your Khmer text into the box. An example with mistakes is filled in so you can try it straight away.
2. Read the checked text below: red words are not in the dictionary, blue words are real words that are often confused (such as និង and នឹង), and orange words are typed in the wrong Unicode order.
3. Click an underlined word and pick a suggestion to replace it. Tick Replace all to fix every copy of the same mistake, type your own correction, or press Ignore for names and new words.
4. Press Fix Unicode ordering to repair the order of vowels, subscripts and signs in the whole text.
5. Optionally press Insert zero-width spaces to add invisible word breaks, then Copy text to take your corrected text back.

Why use ToolOrca Khmer Spell Checker

• Khmer word segmentation with a 45,000-word dictionary
• Suggestions ranked by spelling, word frequency and context
• Finds often confused words: និង / នឹង, ដែល / ដែរ, ពី / ពីរ
• Fixes wrong Unicode order and doubled vowels or subscripts
• Khmer word count and word boundaries view
• Insert or remove zero-width spaces for line breaking
• Private: works offline in your browser, nothing is uploaded

How Khmer spell checking works

Khmer text has no spaces between words, so a spell checker cannot simply look up each word. This tool first cuts the text into letter clusters: a consonant or independent vowel together with its subscript consonant (្ + consonant), vowel sign and diacritics, so ស្ត្រី is one cluster made of ស + ្ត + ្រ + ី. It then finds the split into dictionary words that uses the fewest words (dynamic programming, like the Viterbi algorithm), and uses word frequencies to choose between splits of the same length.

Stretches that match no word are marked as unknown. A second pass looks at short or unusual pieces next to each other and asks whether together they are one small edit away from a real word: ស្រលាស់ is the real words ស្រ + លាស់, but it is almost certainly a typo for ស្រលាញ់. Suggestions come from an edit distance that works on whole subscripts and gives a lower cost to the mistakes Khmer writers really make, such as a missing subscript, swapped marks, ព/ភ, ល/ឡ, ិ/ី and ំ/ម្. Pairs of words seen together in the dictionary text help rank the suggestions.

Common Khmer spelling mistakes

Old or informal spellings. Many words are still often written with ំ where the standard spelling uses a subscript: សំរាប់ → សម្រាប់, តំលៃ → តម្លៃ, ចំងាយ → ចម្ងាយ. Others add or drop a letter: ភាសារ → ភាសា, ពត៌មាន → ព័ត៌មាន, អោយ → ឲ្យ or ឱ្យ.

Words that sound alike. និង (and) and នឹង (will), ដែល (which) and ដែរ (also), ពី (from) and ពីរ (two) are all real words, so only the context shows which one is right. The checker marks them in blue when the other word fits the neighbours much better, for example ខ្ញុំនិងទៅ → ខ្ញុំនឹងទៅ.

Unicode order. The same word can be typed in different orders that look identical: ស្រ្តី (subscript ro first) and ស្ត្រី, or nikahit before the vowel in ខ្ញំុ. Only one order is standard, and the other one is not found by search engines, dictionaries or Ctrl+F. Fix Unicode ordering puts every cluster in the standard order: base, subscripts (ro last), register shifter, vowel, then signs, and also joins split vowels and removes doubled marks and stray zero-width spaces inside words.

Zero-width spaces and Khmer line breaks

Because Khmer has no spaces, browsers, Word and phone apps may break a long line in the middle of a word, or not at all. A zero-width space (U+200B) is an invisible character that marks a word boundary: text looks the same, but lines wrap between words and some tools can select and count words correctly. Use Insert zero-width spaces before publishing a web page or subtitles, and Remove zero-width spaces to clean text that has too many. Turn on Show word boundaries to see how the text was split.