Fixed phonetics code to strip post-suffix d (bug 800167 in

SourceForge).

  Implemented (somewhat kludgily) option for phonetics scheme to
  replace e with é iff it is the last letter of the last tsheg bar.
  This is required by the new THDL phonetics spec.

  New algorithm, per new THDL phonetics spec, for ba->wa processing.
  The heuristic is that it applies only to the last tsheg bar in
  multi-tsheg-bar words.  (Previously, ba always generated "?ba/wa?",
  which is maybe more correct but less attractive.)  This heuristic
  fails on, e.g., "tsheg bar".  Oh well.

  Rationalized format of phonetics file: > is used as separator in exceptions
  as well as rules.  (Previously, : was used in exceptions only.)

Aro phonetics scheme.txt

@@ -96,20 +96,14 @@ x ng
 ; pronunciation (which may also contain spaces).  A semicolon 
 ; precedes a comment.  Blank lines are OK.
 
-ba : wa                 ; mind you, ba (pronounced ba) means cow.  But that's much rarer than wa.
+rdo rje > dorjé
-bo : wo
+mkha' 'gro > khandro
-ba'i : wa'i
+sku mnye > kumnyé
-bo'i : wo'i
+sprul sku > tulku
-bar : ?bar/war?         ; bar = "middle"; could be either, so supply both and let user sort it out
+mtsho rgyal > tsogyèl
-bor : ?bor/wor?         ; bor = "cast away"; could be either, so supply both and let user sort it out
+rta mgrin> tamdrin
-rdo rje : dorjé
+dga' ldan > ganden
-mkha' 'gro : khandro
+dge 'dun  > gendün
-sku mnye : kumnyé
+a mdo > amdo
-sprul sku : tulku
+srid pa > sipa
-mtsho rgyal : tsogyèl
+pad ma > pèma
-rta mgrin: tamdrin
-dga' ldan : ganden
-dge 'dun  : gendün
-a mdo : amdo
-srid pa : sipa
-pad ma : pèma

Aro phonetics test cases.txt

@@ -40,7 +40,7 @@ klad pa > l
 glog > log
 le'u > lé'u
 pa'ang > pa'ang
-ba'i > wa'i
+bar ba'i > barwa'i
 rta mgrin > tamdrin
 
 ; Other tests, to exercise particular rules in the grammar that aren't covered in the rules above

THDL phonetics scheme.txt

@@ -18,16 +18,26 @@
 ; compactly.  For example, it would be difficult to capture the
 ; effects of preinitial consonants on tone (as in the scheme used
 ; in Joe Wilson's book, for instance).)  Also note that not even the
-; whole of the present scheme is implemented using these rules.  In
+; whole of the present scheme is implemented using these rules.  For
-; particular, the deletion of prefix and superscript consonants, 
+; example, the deletion of prefix and superscript consonants, 
 ; and of wa-zur, are done in program code, not using the rules here.
 
+; This makes e come out é only when the last letter in a "word" (*not*
+; syllable).  Our grammar engine is not nearly powerful enough to do
+; this in a clean way.
+<?Enable THDL final é kludge?>
+
 ; Miscellaneous prefix transformations
 g.      ; delete this (representing g prefix, used before root y only)
 dby y   ; must come before db->w, for dbyang
 dbr r   ; must come before db->w, for dbral
 db w    ; must come before by->j
 
+; Removal of confusing 'h's
+th t
+ph p
+tsh ts
+
 ; c and ch are both transcribed ch.  To get this we need a kludge
 ; (involving x), because the rule c -> ch would apply recursively.
 ch c
@@ -42,10 +52,10 @@ my ny
 
 ; Retroflexes
 kr tr
-khr thr
+khr tr
 gr dr
 pr tr
-phr thr
+phr tr
 br dr
 
 ; Other bad behavior from R
@@ -55,7 +65,7 @@ sr s
 ; Uniquely random case
 zl d
 
-; Umlaut of a, o, u followed by d, n, l, s 
+; Umlaut of a, o, u followed by d, n, l, s, and 'i
 ; Note: this must be done before suffix-stripping.
 ; Before actually doing the umlaut, we "hide" the n in ng, so that ng doesn't
 ;   induce umlaut.  This is gross; if we had a real grammar engine it wouldn't
@@ -65,17 +75,24 @@ ad e
 an en
 al el
 as e
+a'i e
 od ö
 on ön
 ol öl
 os ö
+o'i ö
 ud ü
 un ün
 ul ül
 us ü
+u'i ü
 ; restore ng
 x ng
 
+; Stripping of 'i from e'i 
+; (It is stripped from a, o, u by umlaut rules, and from i by vowel-doubling rule.)
+e'i e
+
 ; Stripping of suffix d,  s, and ' from i and e
 ; Note: this has already been done by the umlaut rules for some cases, 
 ;       which don't need to be repeated here.
@@ -112,22 +129,27 @@ ub up
 
 ; There is one exception per line.  Each exception consists of
 ; the transliteration (which may be several syllables separated
-; by spaces), followed by a space, a colon, a space, and the
+; by spaces), followed by a space, a greater-than, a space, and the
 ; pronunciation (which may also contain spaces).  A semicolon 
 ; precedes a comment.  Blank lines are OK.
 
-ba : wa                 ; mind you, ba (pronounced ba) means cow.  But that's much rarer than wa.
+mkha' 'gro > khandro
-bo : wo
+sprul sku > tulku
-ba'i : wai
+rta mgrin > tamdrin
-bo'i : woi
+dga' ldan > ganden
-bar : ?bar/war?         ; bar = "middle"; could be either, so supply both and let user sort it out
+dge 'dun  > gendün
-bor : ?bor/wor?         ; bor = "cast away"; could be either, so supply both and let user sort it out
+a mdo > amdo
-rdo rje : dorje
+bka' 'gyur > kangyur
-mkha' 'gro : khandro
+rgyu 'bras > gyundré
-sprul sku : tulku
+ngos 'dzin > ngöndzin
-rta mgrin: tamdrin
+chab mdo > chamdo
-dga' ldan : ganden
+dpal ldan > penden
-dge 'dun  : gendün
+dpal 'bar > pembar
-a mdo : amdo
+rig 'dzin > rindzin
-blo bzang : lobzang
+skyabs 'gro > kyamdro
-sbra nag zhol : banakzhöl
+'bri ru > biru
+sbra nag zhol > banakzhöl
+rdo rje > dorje
+o rgyan > orgyen
+lha rje > lharjé
+rgyal rtse > gyantsé

THDL phonetics test cases.txt

@@ -1,72 +1,120 @@
 ;
-; These examples come from the draft (8/21/03) THDL Phonetics document
+; These examples mostly come from the THDL Phonetics document (Jan 2004 draft)
 ;
-lha sa > lhasa
+dag pa > dakpa
+ring po > ringpo
+rin chen > rinchen
+lab > lap
+dum bu > dumbu
+dmar po > marpo
+ril bu > rilbu
 sa skya pa > sakyapa
-blo bzang > lobzang
+blo bzang > lozang
 rnying ma pa > nyingmapa
-rdo rje > dorje
+rdo rje > dorjé
 dge lugs pa > gelukpa
-gzhis ka rtse > zhikatse
+gzhis ka rtse > zhikatsé
-mar me > marme
+mar me > marmé
+dge bshes > geshé
 bcu >  chu
-lce > che
+gcig pa > chikpa
-rin chen bzang po > rinchenzangpo
 nag chu > nakchu
-bka' rgyud pa > kagyüpa
+'phag pa > pakpa
+gser thang > sertang
+khang tshan > khangtsen
+lce > ché
+rin chen bzang po > rinchenzangpo
+bka' rgyud > kagyü
 bsod nams> sönam
-thub bstan > thupten
+yul > yül
+dus tshod > dütsö
+bon po > bönpo
+sde dge > degé
+brgyad > gyé
+dge rgan > gegen
+ral pa can > relpachen
+tshe ring > tsering
+byes > jé
+bstan 'dzin > tendzin
 'jam dpal dbyangs > jampelyang
 dge legs > gelek
 kha btags > khatak
+sngags pa > ngakpa
+byang chub > jangchup
+thub bstan > tupten
+tabs > tap
 bka' shag > kashak
 sbra nag zhol > banakzhöl
-thabs > thap
+thabs > tap
 lha sa ba  > lhasawa
 jo bo > jowo
 dpa' bo > pawo
+gsal bar > selwar
+; nga'i deb > ngé dep -- can't do this one, it depends on word segmentation
+bar ba > barwa
 spyan ras gzig > chenrezik
+phyag > chak
 sbyin bdag > jindak
 smyong > nyong
+dmyal ba > nyelwa
 sgrol ma > drölma
 rten 'brel > tendrel
 'bras spungs  > drepung
-'phrin las > thrinle
+'phrin las > trinlé
-dbang > wang
+srung ma > sungma
-dbral > rel
+rdzun smra ba > dzünmawa
-dbyar kha > yarkha
-zla ba > dawa
 klad pa > lepa
 glog > lok
+zla ba > dawa
+lha sa > lhasa
+lho phyogs > lhochok
+lhun grub > lhündrup
+dbang > wang
+dbyar kha > yarkha
+dbral > rel
 le'u > leu
+khyi'u > khyiu
 pa'ang > pang
-ba'i > wai
+gri'i > dri
+'gro ba'i > drowé
+rgyal bu'i > gyelbü
+rin po che'i > rinpoché
+bdag po'i > dakpö
+le'u'i > leü
 rta mgrin > tamdrin
-
-; Other tests, to exercise particular rules in the grammar that aren't covered in the rules above
 g.yon > yön
 phyag > chak
 bkra shis > trashi
-khros ma > thröma
+khros ma > tröma
 sprul > trül
 mri tam ga > mitamga
 srid pa > sipa
 pad ma > pema
 pan chen > penchen
-ral pa can > relpachen
+thun > tün
-thun > thün
 dus gsum > düsum
-sbed > be
+sbed > bé
-ces > che
+ces > ché
-pa'i > pai
+btsan dbang > tsenwang
-che'i > chei
+tshong khang > tsongkhang
-gri'i > dri
+rdzong > dzong
-po'i > poi
+stabs > tap
-le'u'i > leui
+thug pa > tukpa
-rdzogs > dzok
+debs > dep
-thug pa > thukpa
-'debs > dep
 sib sib > sipsip
 lobs pa > loppa
 grub > drup
 kla col > lachöl
+spyan snga ba > chenngawa
+sems dpa'i > sempé
+bon po'i > bönpö
+rdzogs > dzok
+
+; Other random tests
+phreng > treng
+
+; Test of second-suffix d removal.  Made-up word because I don't know real ones.
+rand > ren
+; Test that we don't spazz out on single-letter words.
+a > a
+ai > ai