Skip to content

v0.6.3

Latest

Choose a tag to compare

@5w0rdf15h 5w0rdf15h released this 26 May 21:03

Fixed

  • Open centring vowels were over-segmented. Several open (coda-less)
    vowel frames split into spurious extra syllables; they now parse as a
    single syllable across every scheme:
    • เ◌ือ carrying a tone mark — เสื้อ, เนื้อ, เชื่อ,
      เมื่อ, เบื่อ — now one /ɯːa/ syllable (tlc seuua{F} and
      so on) rather than three.
    • เ◌อะ short /ɤ/ — เลอะ, เถอะ, เยอะ — now one syllable.
    • เ◌ียว after a leading เหลียว, เหนียว, เหมียว
      now one syllable (liaao etc.).
    • A preposed written before a leading consonant but belonging to a
      เ◌ือ frame on the next consonant — เสมือนsa-meuuan
      (leader plus the centring syllable), not three syllables.
  • Three bare consonants now read as a leader plus a closed syllable. A
    C₁C₂C₃ sequence with no written vowel — ถนน, ขนม, กมล
    parses as a leading consonant (C₁ with inherent /a/) plus a closed
    C₂C₃ syllable (C₂ onset, inherent /o/, C₃ coda): ถนน
    tha-nohn, not tha-na-na.
  • Word-final consonant clusters read as a coda. A true cluster
    C + ร ending a word after a vowel-bearing syllable is a coda — the
    first consonant closes the syllable and is silent — rather than an
    onset cluster: บัตรbat, สมัครsa-mak, เนตร
    naeht, วัตรwat. Words whose C has no preceding vowel
    (e.g. นคร) keep the cluster reading.
  • รร is now derived productively. Any C + รร word is read by
    rule rather than only the previously enumerated forms: C + รร plus a
    single final consonant gives /a/ + that coda (มรรคmak,
    สรรพsap); C + รร before another syllable gives /-an/
    (กรรชิดgan-chit, บรรจุban-joo).

Changed

  • morev distinguishes /oː/ from /ɔː/. /ɔː/ /ɔ/ now render as
    ɔ̄/ɔ (Latin small letter open O, U+0254) — the supplementary
    sign introduced in the source dictionary's transcription key
    alongside ə for /ɤ/. Cyrillic о̄/о (U+043E) is now
    reserved for close /oː/ /o/. The previous output collapsed both
    phonemes onto Cyrillic о̄, producing identical Morev strings for
    minimum pairs such as โอน /oːn/ ↔ อ่อน /ɔ̀ːn/ and โต
    ตอ. Examples of changed forms: ขอкхɔ̄´ (was
    кхо̄´), บอกбɔ̄кˆ (was бо̄кˆ), อ่อนɔ̄нˆ
    (was о̄нˆ), ฟุตบอลфутˇ-бɔ̄н (was футˇ-бо̄н).
  • tlc nucleus spellings for three open vowels, matching the scheme's
    established convention:
    • /uːa/ before a /j/ offglide spells ua (not uaa):
      สวยsuay, ด้วยduay, กล้วยgluay.
    • Open short /ɤ/ (เ◌อะ) spells uh: เลอะluh — the
      coda-bearing เ◌ิ form still spells eer (เดินdeern).
    • Open short /o/ (โ◌ะ) spells o (not oh): โต๊ะ
      dto — closed inherent /o/ still spells oh (นกnohk).

Notes for callers

  • These are accuracy corrections: many words — especially those with open
    centring vowels, leading-consonant (อักษรนำ) syllables, word-final
    clusters, or รร — now produce different, more correct output.
    Callers that snapshot transcriptions should re-baseline.