Hindi / Devanagari Script — LGM v1.0


Script Basics

  • Written left-to-right.
  • Each consonant has an inherent vowel /ə/, modified by vowel diacritics or suppressed by the virāma (halant, ्).
  • Vowels can appear as independent letters (word-initial or standalone) or as diacritic signs attached to consonants.

Independent Vowels (10 core)

GlyphLatin ChainPhoneme (IPA)Notes
a/ə/inherent vowel
ā/aː/long a
i/ɪ/short i
ī/iː/long i
u/ʊ/short u
ū/uː/long u
/r̩/syllabic r
e/eː/long e
ai/ɛː/diphthong
o/oː/long o
au/ɔː/diphthong

Consonants (Grouped by Articulation)

Velars

GlyphLatin ChainPhonemeNotes
ka/k/
kha/kʰ/aspirated
ga/ɡ/
gha/ɡʱ/breathy voiced
ṅa/ŋ/velar nasal

Palatals

GlyphLatin ChainPhonemeNotes
ca/t͡ʃ/
cha/t͡ʃʰ/aspirated
ja/d͡ʒ/
jha/d͡ʒʱ/breathy voiced
ña/ɲ/palatal nasal

Retroflex

| ट | ṭa | /ʈ/ |
| ठ | ṭha | /ʈʰ/ | aspirated
| ड | ḍa | /ɖ/ |
| ढ | ḍha | /ɖʱ/ | breathy voiced
| ण | ṇa | /ɳ/ | retroflex nasal

Dentals

| त | ta | /t̪/ |
| थ | tha | /t̪ʰ/ | aspirated
| द | da | /d̪/ |
| ध | dha | /d̪ʱ/ | breathy voiced
| न | na | /n/ |

Labials

| प | pa | /p/ |
| फ | pha | /pʰ/ | aspirated
| ब | ba | /b/ |
| भ | bha | /bʱ/ | breathy voiced
| म | ma | /m/ |


Semi-vowels & Approximants

| य | ya | /j/ |
| र | ra | /r/ |
| ल | la | /l/ |
| व | va/wa | /ʋ/ |

Sibilants & Fricatives

| श | śa | /ʃ/ | palatal
| ष | ṣa | /ʂ/ | retroflex
| स | sa | /s/ | dental
| ह | ha | /ɦ/ | voiced glottal


Diacritic Vowel Signs (Matras)

Attach to consonants to replace inherent vowel /ə/:

MatraLatin ChainPhonemeExample (क)Result
◌ाā/aː/का
◌िi/ɪ/किki
◌ीī/iː/की
◌ुu/ʊ/कुku
◌ूū/uː/कू
◌ृ/r̩/कृkr̩
◌ेe/eː/केke
◌ैai/ɛː/कैkai
◌ोo/oː/कोko
◌ौau/ɔː/कौkau

Special Signs

  • Virāma (्) — removes inherent vowel (क् = /k/)
  • Anusvara (ं) — nasalization; place of articulation depends on following consonant
  • Visarga (ः) — voiceless glottal fricative after vowel
  • Chandrabindu (ँ) — nasalization of vowel

System Integration Notes

  • All Devanagari graphemes can be decomposed into base consonant + vowel flag.
  • For our lattice, this makes Hindi fully compatible with how we handled Baybayin, Geʽez, and other abugidas.
  • Aspirated vs. unaspirated and breathy voiced vs. plain voiced are phonemic — these must be separate phoneme nodes, not just features.
  • Nasalization can be a vowel feature flag in the lattice.