Readings ISO 15919

Readings and ISO 15919

Source of truth: this wiki page.

How words are written in Latin letters: ISO 15919

Every word in an Indian language has a reading, its Latin letters, shown on the card. The readings are based on the letters of ISO 15919, the international standard for writing Indian scripts in Latin letters, and they depart from it wherever a learner is better served (ADR-0025). They are not a strict ISO 15919 transliteration: the list of departures follows the table of letters.

ISO 15919 was chosen because its marks show what plain letters cannot: పాలు, milk, is pālu and పలు, many, is palu; పాట, song, is pāṭa and పాత, old, is pāta. One standard covers all the Indian scripts, so a learner of a second Indian language meets the same letters again.

ISO 15919 writes letters, so कितना is kitanā letter for letter. The decks write each word as it is said, in ISO 15919's letters: kitnā, without the a that Hindi does not say. Where a language says two letters alike, the reading writes the sound: Bengali ঈ is i, as ই is.

Letters	Sound	Example
a	the short vowel every consonant carries: u in but in Hindi, a in Telugu	कल kal
ā ī ū	long a, i, u: held about twice as long	పాలు pālu, नहीं nahīm̐
e ē, o ō	short and long e and o, where a language has both (Telugu, Kannada)	నేను nēnu
ē ō	the long e and o of Hindi, which has no short ones	मेरा mērā
ai au	the vowels of ऐ and औ	है hai
ô	Bengali's and Assamese's own vowel, as in law	কথা kôthā
ê	the vowel of cat, in Bengali, and in words from English	ব্যাগ bêg
ṭ ṭh ḍ ḍh ṇ	tongue curled back to the roof of the mouth (retroflex)	పాట pāṭa, अंडा aṇḍā
t th d dh n	tongue on the back of the upper teeth (dental)	పాత pāta
kh gh ch jh th dh ph bh	an h after a letter is a puff of breath, never one sound: th is not thin, ph is not phone	खाना khānā
c ch	ch of church; ch with a puff of breath	चाय cāy, छह chah
ś ṣ	sh; ṣ with the tongue curled back	शादी śādī
ṅ ñ	ng of sing; ny of canyon	বাংলা bāṅlā
ḷ	an l with the tongue curled back	ನಾಳೆ nāḷe
ṛ ṛh	a flap with the tongue curled back	लड़का laṛkā
ṁ	the anusvara where it nasalises the vowel before y, r, l, v, s or h	संसार saṁsār
m̐	the vowel before is said through the nose	हाँ hām̐
k͟h q ġ z f	sounds from Persian, Arabic and English, written with a dot (nukta)	ख़त्म k͟hatm
x	Assamese's own sound, as in Scottish loch; ISO 15919 has no letter for it	অসম ôxôm

Where the readings depart from ISO 15919

Each row is a deliberate departure, made because the learner is better served by the sound than by the spelling. "ISO 15919 writes" is the letter-for-letter reading of the script, as the tables in tools/transcribe.py give it. The last column is the file that shows it: a card id, or the language's decks/<code>/<code>-romanisation.yaml, which says the same in a paragraph.

What	ISO 15919 writes	Fluenough writes	Why	Shown by
Inherent a, where it is not said (Hindi, Marathi, Gujarati, Bengali, Assamese)	कितना kitanā	kitnā	Letter for letter suggests a syllable nobody says	hi-0163; hi-romanisation.yaml
Inherent a in Bengali and Assamese	কত kata	kôto: ô where it is said [ɔ], o where it is said [o]	The vowel is not [a]. ô is ISO 15919's letter for ऑ, reused	bn-0417, as-0377; bn-romanisation.yaml, as-romanisation.yaml
য-ফলা, ব-ফলা and ম-ফলা (Bengali, Assamese); ê for [æ] (Bengali)	স্যার syāra, আত্মীয় ātmīẏa	sêr, āttiyo	The mark holds the consonant or changes the vowel; it does not say y, v or m	bn-0187, bn-0443; bn-romanisation.yaml
Conjuncts said otherwise than their letters	ज्ञ jña, क्ष kṣa	ज्ञ: Hindi gya, Marathi dnya, Gujarati gna, Bengali gg; क्ष: Hindi kśa, Gujarati kśa, Bengali and Assamese kkh (Marathi क्ष stays kṣa, as ISO 15919 writes it)	The letters are not how the conjunct sounds	hi-0083, hi-0085, mr-0089, gu-0085, bn-0081, as-0085
Anusvara ं and the nasal vowel	always ṁ: अंडा aṁḍā, नहीं nahīṁ	the nasal said, aṇḍā; m̐ for a nasal vowel, nahīm̐; ṁ only before y, r, l, v, a sibilant or h, kiṁvā; Bengali and Assamese ং is ṅ, bāṅlādeśi. The same in Gujarati, Telugu and Kannada: dukāṇam, tiṅgaḷu, navembar (ISO 15919: ṁ)	Anusvara is whichever nasal comes next	hi-0258, hi-0156, mr-0577, bn-0422, te-0246, kn-0451, kn-0464; hi-romanisation.yaml, bn-romanisation.yaml, te-romanisation.yaml, kn-romanisation.yaml, gu-romanisation.yaml, as-romanisation.yaml
Telugu's half nasal ఁ	teravam̐baḍalēdu	teravabaḍalēdu, left out	It is no longer said	te-en-reading-diddubatu.yaml; te-romanisation.yaml
Long ī ū, where the language does not tell them from i u (Marathi, Gujarati, Bengali, Assamese)	मी mī, কী kī	mi, ki	A mark would suggest a contrast the language does not make. Bengali ঈ is i	mr-0206, bn-0193, gu-0310
ए and ओ in those four languages	ē ō: आहे āhē, মোড় mōṛa	e, o: āhe, moṛ	No short e or o to tell them from	mr-0205, bn-0314
Bengali ঐ ঔ; Assamese ঐ ঔ and the vowel sign ো	ai au; Assamese মোৰ mōra	Bengali oi, ou; Assamese ôi, ôu, and u for ো in words, said [ʊ] (mur, muk). The letter ও alone still reads o	They are said so	bn-0775, bn-0011, as-0188, as-0167, as-2064
Bengali স and ষ (শ is ś in both)	s and ṣ: সে se, পরিষ্কার pariṣkāra	ś where said sh: śe, poriśkār; s where said s (strī)	Said alike, as sh or as s, by the word	bn-0126, bn-0683; bn-romanisation.yaml
Assamese স শ ষ	s ś ṣ: দেশ dēśa	x: dex	One sound, as in Scottish loch; ISO 15919 has no letter for it	as-0173
Other Assamese letters said alike	ছাৰ chāra, জানুৱাৰী jānuvārī, ভাইটি bhāiṭi, ভণ্টি bhaṇṭi	sār, zānuwāri, bhāiti, bhonti: চ ছ as s; জ ঝ য as z; ৱ as w; ṭ ṭh ḍ ḍh as t th d dh; ণ as n	Assamese says them alike	as-0162, as-0393, as-0155, as-0156
Other Bengali letters said alike	কারণ kāraṇa; য y, ঢ় ṛh, য় ẏ	kāron; j for য, ṛ for ড় and ঢ়, y for য়	Bengali says them alike	bn-0365, bn-0739; bn-romanisation.yaml
ष in Hindi and Gujarati	भाषा bhāṣā, શિક્ષક śikṣaka	bhāśā, śikśak	Said as श	hi-0197, gu-0322
ऋ	r̥: कृपया kr̥payā	Hindi kri, Marathi, Gujarati, Kannada and Telugu ru, Bengali and Assamese ri	How it is said, in letters every keyboard and font has	hi-0616, mr-0157, te-0007, as-0007
Kannada ೞ; Kannada ಱ and Telugu ఱ	ಮೞೆ maḻe; ṟ	maḷe; r	No longer said differently from ಳ and ర	kn-0580; kn-romanisation.yaml, te-romanisation.yaml
ಫ in words from English	ಕಾಫಿ kāphi	kāfi	Said f	kn-0494

Each language's file in decks/<code>/<code>-romanisation.yaml says how its readings are written, in a paragraph. Each card also records how the word is said in the IPA, the International Phonetic Alphabet: /paːlu/. The app does not show it yet.

A typed answer needs none of the marks. palu, paalu and pālu are all right for పాలు, as kitna and kitnaa are for कितना: answers are compared after the marks are taken away, and each romanisation file lists the spellings people type for one sound. To write a reading and IPA for a new card, run python3 tools/transcribe.py <code> "<word>" <how it is typed>.

See also: Learner-Guide · Home