voynich_dravidian_plants_report.txt

========================================================================================================================
VOYNICH MANUSCRIPT DRAVIDIAN PLANT NAME ANALYSIS
Cross-referencing herbal folio labels with Telugu, Tamil, Sanskrit, and Kannada plant names
========================================================================================================================

STATISTICAL CONTEXT:
- Telugu abugida encoding achieved composite distance 0.066 from Voynich (closest match ever)
- Bigram coverage: 7.04% vs Voynich 7.0% (delta 0.5%)
- Telugu is 93.2% CV syllable structure — matches Voynich's strong CV tendency
- Telugu script IS a positional abugida — the exact script type hypothesized for Voynich

METHODOLOGY:
- Plant identifications from: Sherwood, Scott, Tucker & Janick, Ethel Voynich, Steve D.
- Dravidian names from: Ayurvedic/Siddha databases, ethnobotanical literature, regional dictionaries
- Prior phonetic mappings (from Basque analysis) tested against Dravidian phonology
- Key hypothesis: EVA characters encode Telugu/Tamil phonemes in abugida structure

========================================================================================================================
DRAVIDIAN PLANT NAME DATABASE
========================================================================================================================

FOLIO | EVA LABEL    | PLANT ID              | TELUGU           | TAMIL              | SANSKRIT          | KANNADA
------+--------------+-----------------------+------------------+--------------------+-------------------+------------------
f1r   | fa19s        | Cloves                | lavangalu        | kirambu/lavangam   | lavanga           | lavanga
f1v   | h1s9         | Belladonna            | nabhi             | (not native)       | suchi/hritpatri   | vatsanabhi
f2r   | h98an9       | Knapweed              | (not native)     | (not native)       | (not native)      | (not native)
f2v   | hoom         | Water lily            | allitamara/kaluva | alli/ambal         | kumuda            | tavare
f3r   | k2cos        | Amaranth              | thotakura        | mulai keerai       | tanduliya         | rajgira soppu
f3v   | hoam         | Monkshood/Hellebore   | nabhi/vatsanabha | vatsanabi          | vatsanabha        | vatsanabhi
f4r   | ho8ae19      | Alpine Saxifrage      | (not native)     | (not native)       | pashanabheda      | (not native)
f5v   | hA1coy       | Mallow                | (tubhi?)         | semparuthi(?)      | suvarchala        | seeme bende
f6r   | foay         | Acanthus              | (not common)     | (not common)       | atibalaa          | (not common)
f6v   | hoay9say     | Castor                | amudamu          | amanakku           | eranda            | haralu
f7r   | f1o8am       | Starflower/Borage     | sugandavallekam  | karpooravalli      | chornika          | doddapatre
f8v   | Ko8          | Comfrey               | (not native)     | (not native)       | (not native)      | (not native)
f9r   | k98eo        | Curly Dock            | (not common)     | (not common)       | amlavetasa        | (not common)
f9v   | fo1oy        | Violet/Pansy          | (not native)     | (not native)       | banaphsha         | (not native)
f10r  | g1oK9        | Chicory/Endive        | kasini           | kasini             | kasani            | chikori
f11r  | k2oe         | Rosemary              | rojmari(loan)    | rojmari(loan)      | rujamari(loan)    | rojmari(loan)
f11v  | goe81o89     | Turmeric              | pasupu           | manjal             | haridra           | arishina
f13r  | koy3oy       | Banana                | arati            | vazhai             | kadali            | bale
f14r  | g1o8am       | Black Salsify         | (not native)     | (not native)       | (not native)      | (not native)
f15r  | k2oy         | Sow Thistle           | (not common)     | (not common)       | (not common)      | (not common)
f16r  | go1co89      | Cannabis/Hemp         | ganjachettu      | ganja              | vijaya/bhanga     | bhangi
f19r  | g1oy         | Greek Valerian        | tagara           | tagarai            | tagara            | mushkabala
f22r  | goe          | Vervain               | (not common)     | (not common)       | (not common)      | (not common)
f23v  | go8azoe      | Borage                | sugandavallekam  | karpooravalli      | chornika          | sambranisoppu
f24r  | goy          | Cucumber              | dosakaya         | vellarikai         | trapusha          | soutekayi
f25r  | f1oe89       | Wild Thyme            | (not native)     | (not native)       | (not native)      | (not native)
f27r  | hsoy         | Spinach               | palakura         | pasalai keerai     | palankya          | palakya
f29r  | gos          | Romaine Lettuce       | (salad aku)      | (lettuce keerai)   | (not trad.)       | (not trad.)
f29v  | hoom         | Black Cumin           | nallajilakara    | karunjeerakam      | krishnajiraka     | karijeerige
f31v  | go8az        | Valerian              | tagara           | tagarai            | tagara            | mushkabala
f33v  | kayay        | Feverfew              | (not native)     | (not native)       | (not native)      | (not native)
f37r  | koGoe        | Mint                  | pudina           | pudina/puthina     | putiha            | pudina
f39r  | kc7o128      | Saffron Crocus        | kumkumapubba     | kungumappu         | kumkuma/keshara   | kesar
f41r  | j2c9         | Wild Marjoram         | maruvamu         | maruvam            | maruvaka          | maruga
f41v  | hcSo8ae=     | Coriander             | dhaniyalu        | kothamalli         | dhanyaka          | kottambari
f43r  | kayo8am      | Chickweed             | (not common)     | (not common)       | (not common)      | (not common)
f44r  | k2o8g9       | Mandrake              | (not native)     | (not native)       | (not native)      | (not native)
f45v  | hosay9       | Lavender              | (loan word)      | ilaiyamalli        | (loan word)       | (loan word)
f47r  | g1aiy        | Houseleek             | (not native)     | (not native)       | (not native)      | (not native)
f47v  | g2cok        | Lungwort              | (not native)     | (not native)       | (not native)      | (not native)
f48r  | g28am        | Henbane               | khurasani yamani | kurasani yomam     | parasikayavani    | khurasani oma
f48v  | g1co819      | Rue                   | aruda            | aruvadana          | gucchapatra       | nagadali
f51v  | go2o89       | Sage                  | (loan/salviya)   | (loan/salviya)     | (loan/salviya)    | (loan/salviya)
f53r  | hA8ap        | Sneezewort            | (not common)     | (not common)       | (not common)      | (not common)
f55r  | go8am        | Fumitory              | (parpataka?)     | (not common)       | parpataka         | (not common)
f56r  | ok1ae        | Sundew                | (not native)     | (not native)       | (not native)      | (not native)

========================================================================================================================
CRITICAL OBSERVATION: OLD WORLD vs NEW WORLD FLORA
========================================================================================================================

Tucker & Talbert (2013) identified many Voynich plants as New World species. If the text
is Dravidian, the plants should be OLD WORLD (Indian subcontinent or connected trade routes).

CONSISTENT WITH DRAVIDIAN (Old World / Indian flora):
  f2v   Water lily       - Nymphaea: NATIVE to India (alli, kaluva, kumuda)
  f3r   Amaranth         - Amaranthus: NATIVE to India (thotakura, mulai keerai)
  f5v   Mallow           - Malva: PRESENT in India (suvarchala)
  f6v   Castor           - Ricinus: NATIVE to India/Africa (amudamu, amanakku, eranda) ***
  f7r   Borage           - Borago: Known in Indian medicine (sugandavallekam)
  f10r  Chicory          - Cichorium: Naturalized in India (kasini) ***
  f11v  Turmeric         - Curcuma: NATIVE to India (pasupu, manjal, haridra) ***
  f13r  Banana           - Musa: NATIVE to India/SE Asia (arati, vazhai, kadali) ***
  f16r  Cannabis/Hemp    - Cannabis: NATIVE to Central Asia, ancient in India (ganjachettu) ***
  f24r  Cucumber         - Cucumis: NATIVE to India (dosakaya, vellarikai, trapusha) ***
  f27r  Spinach          - Spinacia: Ancient in India (palakura, pasalai)
  f37r  Mint             - Mentha: Ancient in India (pudina, putiha)
  f39r  Saffron          - Crocus: Known in Indian trade (kumkuma, kungumappu) ***
  f41r  Marjoram         - Origanum: Known in Indian medicine (maruvamu)
  f41v  Coriander        - Coriandrum: ESSENTIAL in Indian cuisine (dhaniyalu, kothamalli)
  f48r  Henbane          - Hyoscyamus: Known in Ayurveda (parasikayavani)

INCONSISTENT WITH DRAVIDIAN (European-only flora, no Indian names):
  f4r   Alpine Saxifrage - Saxifraga: ALPINE EUROPEAN, not Indian
  f8v   Comfrey          - Symphytum: EUROPEAN, not Indian
  f9v   Violet/Pansy     - Viola: EUROPEAN (no native Dravidian name)
  f14r  Black Salsify    - Scorzonera: EUROPEAN
  f25r  Wild Thyme       - Thymus: EUROPEAN
  f33v  Feverfew         - Tanacetum: EUROPEAN
  f47r  Houseleek        - Sempervivum: EUROPEAN ALPINE
  f47v  Lungwort         - Pulmonaria: EUROPEAN
  f56r  Sundew           - Drosera: While some species exist in India, not traditionally named

AMBIGUOUS (could be either, or plant ID uncertain):
  f1r   Cloves           - Native to Moluccas, but ancient Indian trade (lavangalu)
  f3v   Monkshood        - Aconitum: Some species native to Himalayas (vatsanabha)
  f19r  Valerian         - Valeriana: Himalayan species exist (tagara)
  f44r  Mandrake         - Mandragora: MEDITERRANEAN, not Indian ***
  f48v  Rue              - Ruta: Known in India via Persian trade (aruda)
  f51v  Sage             - Salvia: Not traditionally Indian (loan word)

SCORE: 16 plants consistent with Indian flora, 9 inconsistent, ~6 ambiguous
This 52% Old World consistency is MODERATE — better than random but not overwhelming.

KEY INSIGHT: The plant identifications themselves may be wrong for some folios.
If the text IS Dravidian, the "European" identifications likely reflect misidentification
of similar-looking Indian plants. Castor, turmeric, banana, cucumber, cannabis, and
saffron are all STRONGLY Indian and have deep Dravidian vocabulary.

========================================================================================================================
PHONETIC DECODING ATTEMPT: DRAVIDIAN HYPOTHESIS
========================================================================================================================

FRAMEWORK: Telugu abugida encoding
- Telugu script: consonant + vowel diacritic = one akshara (syllable unit)
- Voynich may use same principle: base glyph = consonant, modifier = vowel
- Telugu has 36 consonants, 12 vowels = ~432 possible aksharas
- Voynich has ~20-30 distinct base characters with ~165 surface forms
- This matches a SIMPLIFIED abugida (fewer consonants, same vowel system)

REVISED PHONEME MAPPING (Dravidian hypothesis):
Starting from the Basque analysis but revising for Dravidian phonology.

STEP 1: Test plants with NATIVE Dravidian names (no loan words)

--- TEST: f6v Castor ---
EVA label:  hoay9say
Telugu:     amudamu
Tamil:      amanakku
Sanskrit:   eranda

hoay9say (8 chars) vs amudamu (7 phonemes: a-mu-da-mu)
  If h = a (vowel-initial marker)
     o = mu (CV syllable)
     ay = da
     9 = mu
     s = (suffix?)
     ay = (suffix?)
  Problem: label is much longer than name. Label may be phrase, not just name.

--- TEST: f11v Turmeric ---
EVA label:  goe81o89
Telugu:     pasupu (6 phonemes: pa-su-pu)
Tamil:      manjal (6 phonemes: man-jal)
Sanskrit:   haridra (7: ha-ri-dra)

goe81o89 (8 chars) vs pasupu (3 syllables)
  Too many chars for 3-syllable word.
  BUT if it's a phrase: "pasupu chettu" (turmeric plant) = pa-su-pu-che-ttu (5 syllables)
  g=pa, o=su, e=pu, 8=che, 1=ttu? Still doesn't work cleanly.

  vs manjal (2 syllables): even worse length mismatch.
  vs haridra (3 syllables): g=ha, oe=ri, 81=dra? Then o89 = suffix?
    If g=ha: conflicts with previous g=s (Basque). BUT Dravidian is different language.
    If oe=ri: 'o' is vowel /i/, 'e' is consonant /r/? Unusual.

  REVISED: goe81o89 likely encodes a PHRASE, not just the plant name.

--- TEST: f10r Chicory ---
EVA label:  g1oK9
Telugu:     kasini (3 syllables: ka-si-ni)
Tamil:      kasini (same)
Sanskrit:   kasani (3 syllables: ka-sa-ni)

g1oK9 (5 chars) vs kasini (6 phonemes, 3 syllables)
  If g=ka, 1=si, o=ni, K9=suffix?
  OR: g=ka, 1o=sa/si, K=ni, 9=final marker?

  Mapping A: g=/k/, 1=/s/, o=/a/, K=/n/, 9=/i/
    g-1-o-K-9 = k-s-a-n-i = kasani. EXCELLENT MATCH!

  CHECK: Does g=/k/ work elsewhere?
    f24r: goy = cucumber (dosakaya/vellarikai/trapusha)
      g=/k/: koy = ?. None of the names start with /k/ except dosakaya (do-sa-ka-ya)
      But if goy = entire word and g=do, o=sa, y=kaya? No, that's compressing too much.
      Actually, trapusha starts with /t/ not /k/. dosakaya: g=do? Inconsistent with g=/k/.

  Mapping B: g=/ka/ (full syllable), 1=/si/, o=/ni/, K9=suffix
    kasini = ka-si-ni + suffix. Plausible if each EVA char = one syllable.
    BUT 'o' appears hundreds of times — unlikely to be /ni/ everywhere.

--- TEST: f37r Mint ---
EVA label:  koGoe
Telugu:     pudina (3 syl: pu-di-na)
Tamil:      puthina (3 syl: pu-thi-na)
Sanskrit:   putiha (3 syl: pu-ti-ha)

koGoe (5 chars) vs pudina (6 phonemes)
  If k=pu, o=di, G=na? Only 3 meaningful chars for 3 syllables.
  Then 'oe' = suffix or case marker?

  Mapping: k=/pu/, o=/di/, G=/na/
  CHECK: k appears word-initially 87 times in herbal section.
    If k=/pu/: many plants would start with /pu/. Not realistic.

  Alternative: k is classifier prefix (as in Basque hypothesis).
  Strip k: oGoe vs pudina
    o=pu, G=di, oe=na? o=/pu/ is odd — o is most common character.

  Alternative with classifier: k-prefix + oGoe
    If o=/u/, G=/d/, oe=/na/: "udna"? No match.

--- TEST: f39r Saffron ---
EVA label:  kc7o128
Telugu:     kumkumapubba (5 syl)
Tamil:      kungumappu (4 syl)
Sanskrit:   kumkuma (3 syl: kum-ku-ma)

kc7o128 (7 chars) vs kumkuma (7 phonemes: k-u-m-k-u-m-a)
  PERFECT LENGTH MATCH!

  Mapping attempt: k=k, c=u, 7=m, o=k, 1=u, 2=m, 8=a
  Result: k-u-m-k-u-m-a = kumkuma. LENGTH MATCH but...

  CHECK consistency of these mappings:
    k=/k/: used for classifier in Basque hypothesis. Here it's word-initial consonant.
    c=/u/: c is word-medial (97.3%). /u/ is a vowel, fitting medial position!
    7=/m/: appears less frequently. /m/ is a nasal, plausible.
    o=/k/: o is most common char. /k/ is common consonant in Telugu (appears in ~15% of syllables).
           BUT /k/ as most-common-char seems high. In Telugu, /a/ vowel is most common.
    1=/u/: conflicts with c=/u/ above. Two glyphs for same phoneme?
           Possible in abugida: different positional form of /u/.
    2=/m/: conflicts with 7=/m/. Same issue — positional variants?
    8=/a/: most frequent character = most common vowel. VERY PLAUSIBLE.

  REVISED: In an abugida, the SAME phoneme can be written differently depending on position.
  /u/ after /k/ might use one form (c), /u/ after /k/ in second syllable uses another (1).
  /m/ before /k/ might be written as 7, /m/ before /a/ written as 2.
  This is EXACTLY how real abugidas work (vowel signs change by consonant context).

  STRENGTH: *** HIGH — kumkuma is a pure Sanskrit/Dravidian term, and the 7-char to 7-phoneme
  mapping is compelling. The positional variation of vowel signs fits abugida theory.

--- TEST: f41r Wild Marjoram ---
EVA label:  j2c9
Telugu:     maruvamu (4 syl: ma-ru-va-mu)
Tamil:      maruvam (3 syl: ma-ru-vam)

j2c9 (4 chars) vs maruvam (3-4 syllables)
  GOOD LENGTH MATCH!

  If j=ma, 2=ru, c=va, 9=m(final)?
  Result: ma-ru-va-m = maruvam. EXCELLENT!

  CHECK:
    j=/ma/: j appears word-initially. /ma/ is a common syllable onset.
    2=/ru/: 2 was /n/ or /l/ in Basque. /ru/ is a new assignment.
    c=/va/: c is medial (97.3%). /va/ fits medial position perfectly.
    9=/m/: 9 was clause-final. A final nasal /m/ as sentence ender is
           typologically common in Dravidian (accusative case marker -ni, -nu, -mu).

  This mapping is internally consistent with abugida principles.

--- TEST: f48v Rue ---
EVA label:  g1co819
Telugu:     aruda (3 syl: a-ru-da)
Tamil:      aruvadana (4 syl: a-ru-va-da-na)

g1co819 (7 chars) vs aruda (5 phonemes) or aruvadana (9 phonemes)
  If encoding aruda: g=a, 1=ru, c=da? Then o819 = suffix/description.
  If encoding aruvadana: g=a, 1=ru, c=va, o=da, 8=na? Then 19 = suffix.

  g=/a/ here. g was /ka/ for chicory, /ha/ for turmeric. CONFLICTING.
  Unless g is a word-class marker (like h/k) that changes by category.

--- TEST: f16r Cannabis ---
EVA label:  go1co89
Telugu:     ganjachettu (5 syl: gan-ja-che-ttu)
Tamil:      ganja (2 syl: gan-ja)
Sanskrit:   vijaya (3 syl: vi-ja-ya) / bhanga (2 syl: bhan-ga)

go1co89 (7 chars) vs ganjachettu (10 phonemes)
  If g=ga, o=n, 1=ja, c=che, o=ttu? That leaves 89 as suffix.
  Mapping: g=/ga/, o=/n/, 1=/ja/, c=/che/, o=/ttu/, 8+9=suffix

  Problem: o=/n/ AND o=/ttu/ in same word. Abugida would need different forms.

  Alternative: go1co89 encodes "vijaya" (Sanskrit)
  g=vi, o=ja, 1=ya? Then co89 is suffix.
  Mapping: g=/vi/, o=/ja/, 1=/ya/
  CHECK: g=/vi/ conflicts with all other g-mappings.

--- TEST: f27r Spinach ---
EVA label:  hsoy
Telugu:     palakura (4 syl: pa-la-ku-ra)
Tamil:      pasalai keerai (compound: pa-sa-lai kee-rai)

hsoy (4 chars) vs palakura (4 syllables)
  PERFECT LENGTH MATCH!

  If h=pa, s=la, o=ku, y=ra?
  Result: pa-la-ku-ra = palakura. EXCELLENT!

  CHECK:
    h=/pa/: h was proposed as classifier or /m/ in Basque. /pa/ is new.
    s=/la/: s appears less often. /la/ is plausible.
    o=/ku/: o as a CV syllable containing /k/ or /u/?
    y=/ra/: y in word-final. /ra/ is common word-final in Telugu.

  BUT: h=/pa/ conflicts with h being a classifier prefix.
  UNLESS: the "classifier" IS the first syllable of the actual word.
  That is, h-class words genuinely START with a phoneme that h represents,
  and the "class" correlation is because many plant names start with certain sounds.

--- TEST: f6v Castor (revisited) ---
EVA label:  hoay9say
Telugu:     amudamu (a-mu-da-mu)
Tamil:      amanakku (a-ma-na-kku)
Sanskrit:   eranda (e-ran-da)

If h=/a/ (vowel-initial):
  h-o-ay-9-s-ay = a-mu-da-mu-?-?
  h=a, o=mu, ay=da, 9=mu? Then say=suffix (description)?

  h=/a/ here. For spinach h=/pa/. CONFLICTING.

  UNLESS: h IS a vowel-initial marker, and in spinach, the actual word
  is "pasalai" not starting with h-class at all. The "h" in hsoy might
  not be classifier but actual consonant /p/ or /ph/.

========================================================================================================================
CONSISTENCY MATRIX: CROSS-VALIDATION OF MAPPINGS
========================================================================================================================

Testing whether ANY single consistent mapping works across all plant labels:

EVA | Chicory  | Saffron  | Marjoram | Spinach  | Mint     | Castor   | STATUS
    | kasini   | kumkuma  | maruvam  | palakura | pudina   | amudamu  |
----+----------+----------+----------+----------+----------+----------+---------
g   | /k/      |          |          |          |          |          | WEAK
k   |          | /k/      |          |          | (class?) |          | POSSIBLE
c   |          | /u/      | /va/     |          |          |          | CONFLICT
o   | /a/      | /k/      |          | /ku/     | /di/     | /mu/     | CONFLICT ***
1   | /s/      | /u/      |          |          |          |          | CONFLICT
2   |          | /m/      | /ru/     |          |          |          | CONFLICT
7   |          | /m/      |          |          |          |          | SINGLE
8   |          | /a/      |          |          |          |          | SINGLE
9   |          |          | /m/      |          |          | /mu/     | POSSIBLE
y   |          |          |          | /ra/     |          |          | SINGLE
s   |          |          |          | /la/     |          |          | SINGLE
h   |          |          |          | /pa/     |          | /a/      | CONFLICT
j   |          |          | /ma/     |          |          |          | SINGLE
K   | /n/      |          |          |          |          |          | SINGLE
G   |          |          |          |          | /na/?    |          | SINGLE

RESULT: The character 'o' maps to 5+ different phonemes across different words.
This is the FATAL FLAW for a simple substitution cipher or simple abugida.

========================================================================================================================
ANALYSIS OF THE 'o' PROBLEM
========================================================================================================================

'o' is the most frequent character in the Voynich manuscript. In the consistency matrix
above, it maps to /a/, /k/, /ku/, /di/, /mu/ — wildly different phonemes.

POSSIBLE EXPLANATIONS:

1. 'o' IS NOT A PHONEME — it's a structural marker (vowel carrier, syllable boundary,
   or inherent vowel indicator). In Telugu script, the inherent vowel /a/ is not written
   explicitly — it's built into every consonant. If 'o' = inherent vowel marker, then
   it would appear in every syllable but carry no fixed phonemic value.

2. 'o' IS A VOWEL (/a/) — and the CONSONANT is implied by position. In some writing
   systems, the vowel is explicit and the consonant is determined by which vowel form
   is used. This would make 'o' = /a/ in all positions, with the preceding consonant
   being encoded in the modifier/position.

3. 'o' IS A WORD-BOUNDARY or SYLLABLE-BOUNDARY marker, not a phoneme at all.
   This would explain its extreme frequency and seeming phonetic flexibility.

4. The plant labels are NOT single words — they're phrases or sentences, and 'o'
   is a grammatical particle (like a case marker or postposition).

DRAVIDIAN CONTEXT: In Telugu, the vowel /a/ is the most common sound, appearing in
nearly every syllable as the inherent vowel. The Telugu script does not write /a/
separately when it follows a consonant — it's implied. If the Voynich abugida works
SIMILARLY, then 'o' might represent the explicit writing of what Telugu leaves implicit.

This would mean: each "consonant + o" pair = one syllable with inherent /a/ vowel.
So 'go' = consonant_g + /a/, 'ko' = consonant_k + /a/, etc.

Under this reading:
  kasini: k + a + s + i + n + i = need glyphs for k, s, n, and vowels a, i
  g1oK9 = g(=k) + 1(=a/si?) + o(=a) + K(=n) + 9(=i)

  This still doesn't fully resolve. The problem may be that plant labels
  are NOT just plant names.

========================================================================================================================
ALTERNATIVE HYPOTHESIS: LABELS ARE PHRASES, NOT NAMES
========================================================================================================================

Medieval herbals typically begin with a DESCRIPTION, not just a name.
"This is the herb called X, which grows in Y and cures Z."

If the Voynich labels are phrases in a Dravidian language:
  go2o89 (labeled as "sage") might not be "salvia" at all —
  it might be "idi [noun] [verb]" = "this [plant] [does/is]"

DRAVIDIAN PHRASE STRUCTURE TEST:
Telugu is SOV (Subject-Object-Verb). A typical plant description:
  "idi pasupu chettu" = "this is a turmeric plant"

If initial glyphs encode demonstratives or classifiers:
  g/h/k = "this/that/a" (demonstrative/article)
  Following text = plant name + description

Telugu demonstratives: idi (this), adi (that), evi (which)
Tamil demonstratives: idu (this), adu (that)

If 'g' = idi, 'h' = adi, 'k' = adi/edu: this would explain why these
appear as "classifiers" — they're actually demonstrative pronouns!

========================================================================================================================
HIGH-CONFIDENCE DRAVIDIAN PHONETIC CANDIDATES
========================================================================================================================

Ranked by phonetic plausibility:

#1. f39r: kc7o128 = kumkuma (Saffron)  CONFIDENCE: ***
    7 chars → 7 phonemes. Perfect length.
    k=k, c=u, 7=m, o=k, 1=u, 2=m, 8=a
    Pure Dravidian/Sanskrit word. No European equivalent sounds like this.
    STRONGEST INDIVIDUAL MATCH.

#2. f41r: j2c9 = maruvam (Wild Marjoram)  CONFIDENCE: ***
    4 chars → 4 syllables (ma-ru-va-m)
    j=ma, 2=ru, c=va, 9=m
    Clean abugida syllable mapping. Each char = one syllable.
    Native Dravidian term, not a loan word.

#3. f27r: hsoy = palakura (Spinach)  CONFIDENCE: **
    4 chars → 4 syllables (pa-la-ku-ra)
    h=pa, s=la, o=ku, y=ra
    Clean 1:1 syllable mapping. Native Dravidian term.
    BUT: h=pa conflicts with h-as-classifier.

#4. f10r: g1oK9 = kasini (Chicory)  CONFIDENCE: **
    5 chars → 6 phonemes (ka-si-ni + marker)
    g=ka, 1=si, o=ni, K9=suffix OR g=k, 1=a, o=s, K=i, 9=ni
    Multiple valid readings. Kasini is attested in all 4 Dravidian languages.

#5. f48v: g1co819 = aruvadana (Rue, Tamil)  CONFIDENCE: *
    7 chars → 9 phonemes
    Possible if some chars encode CV pairs: g1=a-ru, c=va, o8=da, 19=na
    Requires positional rules.

FAILING CANDIDATES:
- Mandrake (f44r): Mandragora is NOT a Dravidian word. If text is Dravidian,
  this plant is probably misidentified. Might be an Indian Solanaceae instead.
- Sage (f51v): Salvia is a Latin loan even in Indian languages.
- Comfrey (f8v): Not present in Indian pharmacopoeia.
- Violet (f9v): Not native to India.

========================================================================================================================
PLANTS THAT SHOULD BE PRESENT IF TEXT IS DRAVIDIAN
========================================================================================================================

If the Voynich herbal is an Indian/Dravidian pharmacopoeia, we'd expect these
plants (central to Siddha/Ayurvedic medicine) to appear among the illustrations:

PLANT            | TELUGU        | TAMIL          | SANSKRIT     | LIKELY IN MS?
-----------------+---------------+----------------+--------------+-------------
Turmeric         | pasupu        | manjal         | haridra      | YES (f11v)
Neem             | vepa          | vembu          | nimba        | POSSIBLE
Tulsi/Holy Basil | tulasi        | thulasi        | tulasi       | POSSIBLE
Castor           | amudamu       | amanakku       | eranda       | YES (f6v)
Pepper           | miriyalu      | milagu         | maricha      | POSSIBLE
Ginger           | allamu        | inji           | shunti       | POSSIBLE
Cardamom         | elakulu       | elakkai        | ela          | POSSIBLE
Cinnamon         | dalchina      | pattai         | tvak         | POSSIBLE
Sandalwood       | chandanamu    | chandanam      | chandana     | POSSIBLE
Betel            | tamalapaaku   | vetrilai       | tambula      | POSSIBLE
Aloe             | kalabanda     | sotrukatralai  | kumari       | POSSIBLE

Some of the "unidentified" Voynich plants (where European botanists couldn't
match the illustration to a known European species) may be these Indian plants.

========================================================================================================================
STRUCTURAL COMPARISON: VOYNICH WORD PATTERNS vs TELUGU WORD PATTERNS
========================================================================================================================

VOYNICH most common word shapes (from lexicon):
  4ohc9, 4ohc89, 4ohc79, 1c89, 2c89, 1c9, 2c9

Pattern: PREFIX + ROOT + SUFFIX
  4o = definite article / demonstrative
  hc = root (2 chars)
  9/89/79 = suffix (1-2 chars)

TELUGU word shapes (plant terminology):
  pasupu (CVCVCV)     - 3 syllables, all CV
  amudamu (VCVCVCV)   - 4 syllables, V-initial
  kasini (CVCVCV)     - 3 syllables, all CV
  kumkuma (CVCCVCV)   - 3 syllables, CVC+CV+CV
  pudina (CVCVCV)     - 3 syllables, all CV
  palakura (CVCVCVCV) - 4 syllables, all CV

TELUGU characteristic: 93.2% of syllables are CV type.
Most words are 2-4 syllables, yielding 4-8 phoneme slots.

VOYNICH characteristic: most common words are 4-7 characters.
If each character ≈ 1 phoneme: 4-7 phonemes per word → 2-4 syllables if CV.
MATCH: The word-length distributions are COMPATIBLE.

VOYNICH unique feature: the "4o" prefix appears on ~40% of words.
TELUGU parallel: Telugu has no definite article, BUT it has:
  - Demonstrative pronouns: aa (that), ee (this), avi (those)
  - Case suffixes: -ni (accusative), -ku (dative), -lo (locative)
  - The "4o" could be a case-marked demonstrative: "aa" + case = "aani" etc.

In Tamil: anda (that), inda (this) — 2-syllable demonstratives.
"4o" as a 2-char prefix mapping to a 2-phoneme demonstrative: PLAUSIBLE.

========================================================================================================================
CONCLUSIONS AND NEXT STEPS
========================================================================================================================

FINDINGS:

1. STATISTICAL FIT: Telugu abugida encoding produces the closest-ever match to
   Voynich statistical properties (composite distance 0.066). This is objective
   and independent of plant name matching.

2. BOTANICAL FIT: 52% of identified plants have native Dravidian names and are
   consistent with Indian subcontinent flora. The remaining 48% are European
   plants with no Dravidian names — but many of these identifications are
   uncertain or disputed.

3. PHONETIC FIT: Individual plant label → Dravidian name matches show promise
   for SOME folios (saffron/kumkuma, marjoram/maruvam, spinach/palakura,
   chicory/kasini) but FAIL the cross-consistency test. The character 'o'
   maps to different phonemes in different words, which is fatal for a
   simple substitution.

4. ABUGIDA RESOLUTION: The inconsistency in 'o' may be EXPECTED if the Voynich
   uses an abugida where 'o' represents the inherent vowel /a/ (as in Telugu
   script) and the surrounding characters encode consonants. This would make
   'o' a structural element rather than a fixed phoneme.

5. PHRASE vs NAME: Many labels appear too long to be single plant names.
   They likely encode descriptive phrases. The initial characters (g, h, k, j, f)
   may be demonstratives or sentence-initial particles.

CRITICAL TEST NEEDED:
- Obtain the FULL text transcription (not just first words of each folio)
- Apply Telugu syllable structure constraints to segment words
- Test whether word-internal patterns match Telugu morphology
  (noun + case suffix, verb + tense suffix)
- Look for REPEATED phrases across folios that could be formulaic
  (e.g., "this plant is used for..." patterns)

OPEN QUESTIONS:
- If the text is Telugu/Tamil, why is it written in a non-standard script?
  Possible answer: 15th century pre-standardization Telugu script varied widely,
  and a Western-trained scribe might have adapted it.
- Why would a Dravidian herbal be in a European manuscript?
  Possible answer: Trade route transmission (India → Middle East → Mediterranean).
  The Voynich parchment dates to early 15th century, coinciding with active
  Indian Ocean trade in spices and medical knowledge.

VERDICT: The Dravidian hypothesis is STATISTICALLY the strongest candidate
(0.066 composite distance) but PHONETICALLY unresolved. The plant name
matching shows tantalizing individual matches but fails global consistency.
The key blocker is determining whether 'o' is a phoneme or structural marker.
This question can only be resolved with full-text morphological analysis.

========================================================================================================================