Bishnupriya Manipuri (BPM) Transcription Rules — Latin / ASCII / IPA

These rules are designed to convert Bishnupriya Manipuri (BPM) headwords into: Latin (IAST-like), ASCII Roman, and IPA. The main goals are:

• Detect & fix common OCR artifacts from scanned dictionaries.
• Use consistent transliteration mappings.
• Apply BPM-specific pronunciation logic (especially final schwa and Ba-fala behavior).
• Support a Tatsama option for Sanskritized spellings.

This page documents the rules; your conversion tool applies them programmatically. If you update your converter, update this page to match.

1. OCR cleanup (remove unwanted marks, normalize spacing and punctuation).
2. Core BPM normalization (standardize variants: nukta forms, ya-phala, etc.).
3. Ba-fala rewrite pass (rewrite BPM text so all outputs follow the same logic).
4. Generate Latin output.
5. Generate ASCII Roman output from Latin.
6. Generate IPA output directly from BPM (same normalization).

Apostrophe rule

Apostrophe ' is preserved as a grammatical marker (e.g., plural). It is not automatically converted into a final schwa. Example:

অকর   → ɔkɔr
অকর'  → ɔkɔr'

Latin output is IAST-like transliteration with inherent vowel behavior:

• Consonants carry an inherent a unless suppressed by ্ (virama) or replaced by a vowel sign.
• Vowel signs map to long/short vowels (e.g., া → ā).
• Anusvara, visarga:
  • ং → ṃ
  • ঃ → ḥ
• Ya/Ya-phala:
  • য় → y
  • ্য behaves like য় (y behavior)

Key consonant mapping highlights

ASCII romanization is derived from Latin by replacing diacritics:

ASCII rules are meant for search and typing convenience, not phonetics.

IPA output is generated from BPM with BPM-specific phonology rules. The default inherent vowel is represented as ɔ when applicable.

Core mapping notes

• Default inherent vowel: অ → ɔ.
• Long vowel signs: া → aː.
• Anusvara ং is rendered as ŋ (unless context rules are added later).

Special conjuncts

BPM words often drop the final inherent vowel, but many words must keep it. The tool uses these rules to decide whether to keep a final a (Latin) / ɔ (IPA):

Always keep final schwa

• Any single-letter word.
• Words ending with ্ + letter (final conjunct marker).
• Words ending with ঞ or হ.
• Words ending with these sequences:
  • ্ + মত
  • ্ + রত
  • ্ + রথ
  • ৃ + ত (i.e., ৃত)
  • ্ধ + ত (i.e., ্ধত)
  • ন্দ
• Words ending in ক্ষ or জ্ঞ keep final schwa.
• Project exception list: a curated list of words that must keep final schwa (maintained in the converter code).

Example

অর্ধাঙ্গ → ardhāṅga / ɔrdʱaːŋɡɔ (final schwa kept)

Ba-fala (্ব) behaves like Bengali with BPM-specific refinements. The converter applies a rewrite pass so Latin/ASCII/IPA follow the same behavior.

Rules

1. Beginning of word: usually silent (dropped).
2. Beginning + "া" (ā-kar): if the word begins with a Ba-fala conjunct and is followed by া, add a w-sound by rewriting ্ব → ্ৱ at the beginning only.
3. Middle/end: consonant is doubled (geminate), and the Ba-fala itself is dropped.
4. Exceptions where "b" remains: if attached to ম, ব, গ, হ, or in র্‌ব (rb), keep the b-sound.
5. Special conjunct behavior: শ্ব in the middle behaves like a doubled consonant (e.g., “śś / ʃʃ”).

Examples

অশ্ব      → aśsa / ashsha / ɔsʃɔ
বিশ্বাস   → biśsāsa / bishsaas / biʃsaːs
দ্বাদশ    → dwādaśa (beginning + ā-kar gives w-sound)

Some words are Sanskritized (Tatsama) and should follow different conjunct spellings:

• ক্ষ: Tatsama kṣ, non-Tatsama kkh (Latin) / kkʰ (IPA).
• জ্ঞ: Tatsama jñ, non-Tatsama gg (Latin/ASCII), IPA uses ɡɡ.
• স্ব: Tatsama sw (Latin) / sʋ (IPA), non-Tatsama behaves like শ.

Your UI can offer a checkbox “Tatsama” that switches these mappings.

Some examples depend on the evolving exception list for final schwa and other refinements. If an example differs from your current converter output, adjust either the converter or this page for consistency.