Writing fiction in Burmese
Writing fiction in Burmese: stacked diacritics, the Unicode and legacy font problem, line height, no spaces between words, broken word counts, publishing notes.
Published 27 Sept 2026 · 4 min read
Writing fiction in Burmese online means working with a script most writing software was never designed around. The Myanmar script is compact, circular and heavily decorated: consonants carry inherent vowels, stacks of marks rise above and fall below them, and nothing between words marks where one word ends and the next begins. None of that is a problem for the language, which has carried novels, poems and newspapers for a long time. It is a problem for tools, counters and platforms built for English, and it rewards a writer who knows exactly which assumptions break.
How the script works
Burmese is written with an abugida: each consonant letter carries an inherent vowel, and the other vowels are "indicated using separate letters or diacritics which appear above, below, in front of, after or around the consonant" (Omniglot). The language is tonal, and the tones are written too, "indicated in writing using diacritics or special letters". To suppress the inherent vowel or build a consonant cluster, the script stacks consonants vertically, so a single base character can end up with a column of marks above it and a stack below. That vertical inventory is the root of most of the formatting trouble below.
Unicode, fonts and the Zawgyi inheritance
The script has been in Unicode since version 3.0 in 1999, with extensions added over the years since. The complication is history: before Unicode fonts were widespread, Myanmar users wrote in hack fonts like Zawgyi and Win Innwa, which, in Wikipedia's summary, "use unallocated code points" in the Burmese block "to manually deal with shaping". Text in those encodings is not Unicode and renders as garbage or empty boxes on Unicode-only systems.
For a fiction author today the rule is simple: write in Unicode, full stop. Then choose a real Myanmar font. Pyidaungsu is the national standard font for Myanmar books, and Padauk, an open-source font, is specifically built so stacked consonants and tone marks render completely rather than being clipped. When you publish, test chapters on phones, not just your own laptop: font coverage on reader devices is uneven, and a platform whose font stack lacks Myanmar script shows your audience boxes where the story should be.
Line height, or why the diacritics get sliced
Because marks stack above and below the base consonant, Burmese needs more vertical room between lines than Latin text. At the line heights English web pages use by default, upper vowel signs and lower stacks collide with the line above or are cut off entirely. Kanonara's own typography research places Myanmar first among its eleven supported scripts in the vertical room it needs. Practically: never accept a Latin default line height for Burmese prose. Give it generous leading, and check the worst case, a consonant carrying every mark it can, not the average line.
No spaces, and what that breaks
Unicode's line-breaking specification groups Thai, Lao, Khmer and Myanmar into one "Complex Context Dependent (South East Asian)" class and says of these scripts that they "do not mark word boundaries in any way", so runs of text "require morphological analysis to determine break opportunities". For the writer this has three concrete consequences:
- Word counts are fiction. A counter that splits on spaces reports a whole paragraph as one "word". Minimum-length rules set in words, whether platform requirements, contest rules or your own targets, misbehave. Judge length in characters or reading time instead (see how long a web chapter should be).
- Line breaking is engine work. The reading app must segment syllables to find break points; where segmentation is weak, lines break in odd places. You cannot fix this at the text level; you choose tools whose rendering is good.
- Word-level tools misbehave. Search, spellcheck and replace built on space-delimited words are unreliable. Search for phrases, not words.
Publishing considerations
- Test the platform before committing. A web serial lives or dies on readers' phones, so post a test chapter and check it in the app, not only the web view, and check whether comments render correctly too.
- Mind mixed encodings. Some readers still write in Zawgyi. Your manuscript stays Unicode; convert anything you quote deliberately, and never let mixed-encoding text into your notes or story bible records.
- Check ebook and print exports the same way: the export font must cover the full mark inventory, and line height must be loosened before the stacked marks survive the trip.
The software question, as ever, is how to choose writing software: pick tools that render the script correctly, count it sensibly, and give you your files. For where Myanmar-language fiction can live online, start from the serial platform landscape and verify each platform's script support yourself.
Sources
- Unicode Standard Annex #14: Line Breaking Property Myanmar in the SA class; no marked word boundaries.
- Omniglot: Burmese script Diacritic positions, tones, rounded letters.
- Wikipedia: Burmese script Unicode history and the legacy font problem.
Craft claims above trace to these sources. Where a point is practitioner consensus rather than settled fact, the text says so.