Thai Vowel Length: Why น้ำ Is Long and เล่น Is Short

How often does Thai vowel length break from the spelling?

Rarely per word, often per page. On 26 October 2026 we took the 3,000 most frequent words in the Thai National Corpus as PyThaiNLP packages it and read each against its pronunciation respelling on English Wiktionary, which rewrites a word the way it is said: น้ำ as น้าม, เล่น as เล็่น. 2,644 of the 3,000 have one. Of the 1,101 that are a single syllable, 60 are said at a different length from the one they are written with.

Sixty is 5.4% of those words, about one in eighteen, but they are not rare words. Four are in the top 100: ได้, ต้อง, เช่น and แห่ง. Thirty-two are in the top 1,000. Together the 60 make up 3.04% of every word in the corpus, so a page of ordinary Thai prose has several.

The split is lopsided. Only 11 are written short and said long. The other 49 go the other way: written with a long vowel, said with a short one.

All 60, from the check of 26 October 2026: the spelling, the length Wiktionary's respelling gives, and each word's share of all words in the corpus. The same rows, with corpus counts and IPA, are in dataset.csv.

The eleven written short and said long

All eleven use one of the vowels that the long and short vowel chart treats as short: ไ- or ใ-, เ-า, and ◌ำ. Here they are in order of frequency: ได้, น้ำ, เจ้า, เปล่า, ไม้, ใต้, เช้า, เท้า, ไหว้, เก้า and เกล้า.

There is no spelling cue that picks them out. Every one carries a tone mark, but so do 41 words in the list with the same vowels that are short as written: 6 of the 21 tone-marked เ-า words are said long, 4 of the 24 tone-marked ไ- and ใ- words, and 1 of the 7 tone-marked ◌ำ words, น้ำ. ต่ำ, ค่ำ and ถ้ำ look exactly like it and are short. This group is eleven words to learn, and you will meet ได้ in almost every sentence.

Phayan, an app for learning to read Thai script letter by letter from zero, with audio on every word, teaches น้ำ as náam in the first round of its Unwritten rules level, as one of the famous exceptions, next to ใหญ่.

The forty-nine written long and said short

This is the bigger group, and it has a shape. All 49 end in a consonant, 41 carry a tone mark, and 46 are spelled with plain เ-, แ- or -อ. Wiktionary shows them respelled with ็, the mark that shortens a vowel: เล่น becomes เล็่น, ต้อง becomes ต็้อง.

How reliable is that shape? Among the one-syllable words in the list that end in a consonant, carry ่ or ้ and use one of these vowels:

• Plain เ-: 11 of 12 are said short. เช่น, เล่น, เน้น, เส้น, เล่ม, เก่ง, เร่ง, เต้น, เข้ม, เด่น and เว้น. The one that stays long is เอ่ย.
• แ-: 9 of 18. แห่ง, แบ่ง and แข่ง are short; แล้ว, แก้ว and แจ้ง are long. A coin toss.
• -อ: 19 of 37. ต้อง, ห้อง and ช่อง are short; ก่อน, น้อง and ร้อน are long.
• โ-: 0 of 4. โค้ง and โล่ง stay long.

So the one rule worth carrying is narrow: a closed syllable spelled with plain เ- and a tone mark is almost always said short. For แ- and -อ, learn the words. Eight in the group have no tone mark at all: เงิน, หรอก, แถว, เพชร, หอย, แชมป์, เปน and ถอย.

Does the wrong length change the tone?

Almost never. Length only feeds the Thai tone rules in a dead syllable, which Wikipedia's Thai phonology defines as one that ends in a stop or a short vowel, and 58 of the 60 words end in a sonorant. Each of those gets the same tone whichever length you read it with. We derived every word's tone from its spelling at both lengths and compared it with the IPA on its Wiktionary entry: read with the written length, 58 of the 60 come out right.

เพชร, a diamond, is the exception worth knowing. It is low class, dead, and written with a long vowel, which derives a falling tone: phêet. It is said short, and a short dead low-class syllable is high: phét. Here the spoken length is also the spoken tone. The other miss is แชมป์, champion, an English loan that Wiktionary gives a high tone where its spelling gives mid, whichever length you read.

Why vowel length matters at all

The two vowels in a Thai pair are the same sound, one held longer than the other. In Thai that separates words that share their consonants, their tone and their tone mark:

• เข้า and ข้าว. khâo, to enter, and khâao, rice. Both are high class, both carry ไม้โท, both are falling.
• ไต and ตาย. tai, kidney, and taai, to die.
• วัน and วาน. wan, day, and waan, as in วานนี้, yesterday.

Length also sorts a syllable into live or dead before the tone rules run. An open syllable, one with no final consonant, is live on a long vowel and dead on a short one, so คา is khaa, mid, and คะ is khá, high, with no tone mark on either. In a dead syllable on a low-class letter it decides the tone directly: พัด, short, is phát, and พาด, long, is phâat. RTGS romanisation writes both lengths with the same letter, so เข้า and ข้าว both come out as khao; the Thai spelling keeps them apart.

How to read these words

• Keep reading length from the spelling. It is right for about 95% of common one-syllable words, and only for เพชร among these 60 does the wrong length give the wrong tone.
• Learn the eleven long ones as words. ได้, น้ำ, เช้า and the rest have no spelling cue.
• Expect plain เ- with a tone mark to be short. เล่น, เต้น, เก่ง. It is right 11 times out of 12 in this list.
• Keep น้ำ long on its own. Wiktionary lists a short form, นั้ม-, only for the start of compounds.

How was this checked?

On 26 October 2026 a Python script using only the standard library downloaded the first 3,000 words of the Thai National Corpus frequency list as PyThaiNLP ships it and the kaikki.org extract of English Wiktionary's Thai entries, and compared each word's written vowel length with the length in its Wiktionary respelling. For the 60 that differ it also derived the tone from the spelling at both lengths and compared it with Wiktionary's IPA, and it re-reads every page this post quotes. The results are in dataset.csv and the script is check.py.

Frequently asked questions