adjustments to the (Un/)aspirated Voiceless Consonants rules

[eroux]

The rules like

SUBSTITUTE ("གཐ(.*)"r) ("གཏ$1"v) TARGET (σ);

etc. seem fine, but there seems to be a few exceptions (especially to this one), for instance here's what I find in the Monlam dictionary:

• གཐུམ -> འཐུམ (not གཏུམ)
• གཐོས -> ཐོས (not གཏོས)
• གཐྲ -> གཏར (not གཏྲ, which would be invalid)
• གཆེག -> ཙེག (not གཅེད)
• གཐང -> ཐང (not གཏང) (note that I'm not so sure about this one)

I think this should be double checked but I just want to record it here to start the discussion

[FChrispz]

@eroux this one is not very clear to me - we actually want to change གཐོས -> གཏོས and same for the other cases you listed - in fact in the grammar we said:

Aspirated Voiceless Consonants
Background: In Tibetan, Aspirated and Unaspirated Voiceless Consonants can appear only at syllable onset. In old Tibetan aspirated and unaspiratad voiceless initials does not have a phonemic distinction.
Rule: At the beginning of the syllable make the following changes:

SUBSTITUTE ("མཀ(.)"r) ("མཁ$1"v) TARGET (σ);
SUBSTITUTE ("མཅ(.)"r) ("མཆ$1"v) TARGET (σ);
SUBSTITUTE ("མཏ(.)"r) ("མཐ$1"v) TARGET (σ);
SUBSTITUTE ("མཙ(.)"r) ("མཚ$1"v) TARGET (σ);
SUBSTITUTE ("འཀ(.)"r) ("འཁ$1"v) TARGET (σ);
SUBSTITUTE ("འཅ(.)"r) ("འཆ$1"v) TARGET (σ);
SUBSTITUTE ("འཏ(.)"r) ("འཐ$1"v) TARGET (σ);
SUBSTITUTE ("འཔ(.)"r) ("འཕ$1"v) TARGET (σ);
SUBSTITUTE ("འཙ(.*)"r) ("འཚ$1"v) TARGET (σ);

[eroux]

what I mean is that in order to lemmatize better, you should first apply the rules indicated above. Otherwise your rules will replace
གཐུམ with གཏུམ, but in that case the equivalent in Classical Tibetan is འཐུམ, not གཏུམ.