Why Thai romanisation lies (and what to use instead)
9 min read
Published
You'll spend your first six months reading romanised Thai. Some of those six months will be wasted because the romanisation you're reading is making promises it can't keep. Specifically: it's hiding the tones, suggesting wrong consonants, and giving you a confidence in your pronunciation that doesn't survive contact with a native speaker.
This isn't a reason to skip romanisation. It's a reason to know what it is and isn't, so you don't trust it more than it deserves.
The systems you'll encounter
Three systems compete for your attention as a Thai learner. They all look superficially similar — the same Thai word will read close enough across them that you'll miss the differences for a while.
RTGS (Royal Thai General System of Transcription). The official Thai government standard. You'll see it on road signs, official documents, and most older textbooks. It uses bare Latin letters with no diacritics. Example: ฝันดี renders as "fan di".
Phonetic-tone notation (the one ThaiDai uses, also taught in Thai language schools). Adds tone marks above vowels — ̌ for rising, ̂ for falling, ̀ for low, ́ for high, plain for mid. Long vowels are doubled or marked. ฝันดี renders as "fǎn dee".
Academic / Sanskrit-derived transliterations. Various systems used in scholarly and IPA-adjacent contexts. Heavy with diacritics like macrons, dots-below, and underscores. ฝันดี renders as something like "F̄ạn dī" depending on the exact system. (This is what Google Translate often emits.)
Quick comparison on the same word:
| System | ฝันดี (sweet dreams) | What's missing |
|---|---|---|
| RTGS | fan di | Tones, vowel length |
| Phonetic-tone | fǎn dee | (most things shown) |
| Academic | F̄ạn dī | Tones (the macrons mark vowel length, not tone) |
Why RTGS lies the most
RTGS is built for English-speakers to read Thai place names. It is not built for English-speakers to learn Thai.
It strips tones entirely. ฝันดี (sweet dreams, rising tone on ฝัน) and ฟ้าใด (which sky, falling tone on ฟ้า) both end up in RTGS as something like "fan di" or "fa dai" — the system genuinely cannot tell you what to say. The same is true for ขาว (white, rising) and เข้า (enter, falling) — both render close to "khao" without any tone signal.
It also normalises some consonant distinctions away. Thai has multiple consonants that map to the same RTGS letter — ส, ศ, and ษ all render as "s", but they have different histories and trigger different tone-class rules. For day-to-day pronunciation this doesn't matter (they sound the same), but the moment you start trying to read tones from the script, knowing the consonant class is everything.
For a learner: RTGS is how Thai is signed, not how Thai is taught.
Why academic systems lie subtly
Sanskrit-derived transliterations are precise about consonants and vowels, but they don't carry conversational tone information either. A macron over a vowel in those systems usually means "long vowel", not "high tone". You can't read the tone off the page.
They also feel professional, which is the worst kind of lie — it makes you trust them. A learner who sees ฝันดี written as "F̄ạn dī" assumes the diacritics encode pronunciation. They encode something. They don't encode the rising tone on the first syllable, which is the part you have to get right or you've said a different word.
Why phonetic-tone is the least bad
Phonetic-tone notation does the one thing that matters: it puts the tone on the page. fǎn dee tells you that ฝัน is rising-tone (the ̌ mark) and that the vowel in ดี is long (the doubled ee). Both are pieces of information you cannot pronounce without.
It's not perfect. There are several variants of phonetic-tone notation — ThaiDai uses one, Pimsleur uses another, your favourite Thai-language school in Bangkok uses a third. The tone marks sometimes look different (â vs ɑ̂ vs aa followed by a diacritic). For learners it doesn't matter much; you adapt to whichever system your primary resource uses.
What matters is the principle: the page shows you tone.
What romanisation cannot fix
Even good romanisation has a hard ceiling.
Length distinctions are slippery. A long vowel in Thai is roughly twice as long as a short one. Some systems double the letter (aa for long, a for short); some use a macron; some don't mark length at all. None of them communicate the feel of how long is long. Your ear has to learn it.
Consonant subtleties are flattened. ป (unaspirated p) and ผ (aspirated p) are different letters in Thai with different sounds — paa (ป้า, aunt) vs phaa (ผ้า, cloth). ผ is the high-class aspirated; พ is the low-class aspirated; both render as "ph" in romanisation but they trigger different tone-class rules. The English speaker reads "ph" as the f sound from "phone" and now there are three things wrong: the aspiration cue is misread, the wrong consonant is in their mouth, and they don't even know they made the mistake.
Glottal stops vanish. Thai sometimes has a glottal stop (the catch in the throat between syllables in "uh-oh") that romanisation rarely marks. Words that sound like "rao" with a glottal versus without are different words.
For these, the only fix is hearing native audio repeatedly. Romanisation is a launching pad; it's not a destination.
What to do as a learner
Use phonetic-tone notation as your primary romanisation. RTGS for street signs and place names; academic notation only if you're reading linguistic literature; phonetic-tone for everything else.
Treat romanisation as scaffolding. It's there to bootstrap your pronunciation while you learn the script. Delete it from your mental model the moment you can read the Thai script for a word — keep the audio, keep the meaning, drop the romanisation. Otherwise you're memorising a translation of a translation.
Read along while you listen. Romanisation + native audio together is much more powerful than either alone. The romanisation gives you a hook to remember the word; the audio corrects what the romanisation misses.
Start the script earlier than feels comfortable. Six weeks of Thai script lessons in your first three months saves a year of romanisation reliance. Once the script is automatic, you read tone directly off the page (consonant class + tone marks + vowel length tell you everything), and the romanisation crutch falls away.
We have a longer write-up on Thai tones for English speakers — same ground, different angle. If you're committed to learning the script seriously, the getting-started guide walks through the daily practice loop that makes script-reading stick.
Drill the tones, hear the difference.
Every word in the deck plays in a native-trained Thai voice. The minimal-pair drills (มา / ม้า / หมา) train your ear directly. Free tier covers the foundational set — get started in two minutes.
Read next
- 9 min read
Thai tones, explained for English speakers
Five tones, one syllable, five different meanings. A pragmatic guide to Thai's mid, low, falling, high, and rising tones — with examples, common pitfalls, and how to drill them into your ear.
- 6 min read
Getting started with ThaiDai
A practical 10-minute walkthrough — picking your first 100 Thai words, setting a daily goal, and turning the practice deck into something you actually use.
- 7 min read
The daily-practice loop
Why ten minutes a day beats two hours on Sunday, how spaced repetition actually works, and the small habit hacks that make Thai stick.