Merge branch 'upstream' into development

Conflicts:
	dictsource/dict_phonemes
	phsource/compile_report
	phsource/ustop/th_rfx.wav

dictsource/bn_list

 _1	'&k
 _2	d'uj
 _3	t'in
-_4	tSh'ar
+_4	tS#'ar
 _5	p'a~tS
 _6	tS#'Oj
 _7	S'at
 _11	'&garo
 _12	b'arO
 _13	t'erO
-_14	tSh'oddO
+_14	tS#'oddO
 _15	p'OnerO
 _16	S'olO
 _17	S'OterO

dictsource/bn_rules

 
 
 .group ছ
-        ছ        tShV
-        ছ  (B    tSh
+        ছ        tS#V
+        ছ  (B    tS#
 
         
 .group জ

dictsource/de_list

 pochs           pOxs
 präsident        $3
 prämie           $alt
-privatisier     pri:vati:z'i:r
-problematisier  pro:ble:m'a:tIzi:r
+privatisier     $4
+problematisier  $3 //pro:ble:m'a:tIzi:r
 
 reliquie         $alt
 risik           $1

dictsource/en_list

 imo	$abbrev
 inlb	$abbrev
 ios	$abbrev
+ipa	$abbrev
 irc	$abbrev
 irs	$abbrev
 isbn	$abbrev
 d'ye		dj@
 dynamo		$1
 
+earlobe		i@3loUb
 early		3:lI
 eavesdrop	i:vzdr0p
 ebay		i:beI
 takel		tak@L
 talent		tal@nt
 tallyho		talI'hoU
+timbre		tamb3
 tampon		tamp0n
 tangerine	tandZ@r'i:n
 taoiseach	ti:S@x
 vestige		vEstI2dZ
 vex		vEks        // for vexed
 viagra		vaI'agr@
+vibrato		vI2brA:toU
 vicar		vIk3
 vigilante	vIdZI2l'anti
 vignette	vInj'Et
 Abu		abu:
 Abraham		eIbr@ham
 Ada		eId@  $onlys
-Andrea		$2
 Adrian		eIdri@n
 Agatha		aga#T@
-Angela		andZ@l@
-Angelika	$2
 Aggie		agi
 Agnes		agn@s
 Aguilera	agi:l'e@r@
 Andre		0ndreI
 Andrea		andri@
 Andreas		A:ndr'eI@s
-Angela		ang@l@
+Angela		andZ@l@
+Angelika	$2
 Anita		$alt3
 Anthony		ant@ni
 Anton		ant0n

dictsource/en_rules

 ?3       _) ar (istocrat  a#r
        log) ar (i        @r
       marg) ar (AtA_     3r'
+        _l) ar (yn       ar
          m) ar (it       ar
          p) ar (i        ar
          p) ar (in       e@
        &sp) ass (_       @s
         rr) ass          @s
         &v) ass (_       @s
+            as (al       eIz
+         b) as (al       as
             ascii        askI
          f) asc (ia      eIS
          f) asci (s      aSI
             pro (voc     pr%0#
             pro (vok     pr@
             pro (sec     pr0
+         _) prose (l     pr0s
             pro (fil     proU
             pro (gra     proU
             pro (log     proU

dictsource/hi_list

 
 // Numbers
 // devanagari numbers are changed to latin characters before translation
-_0        s'u:n.jV
+_0        S'u:nj@-
 _1        'e:k
 _2        d'o:
 _3        t'i:n
-_4        c:'a:r:
+_4        c'a:r
 _5        'pa~c
-_6        c#'@2
+_6        c#'VH
 _7        s'a:t
 _8        'a:t.#
 _9        n'O:
 ?3  _9    n'aU
 
 _10       d'Vs
-_11       gj'a:r-ah
-_12       b'a:rah
-_13       t'e:r@2h
-_14       c#'O:d@2h
-_15       p'Vndr@2h
-_16       s'o:l@2h
-_17       sVtr@2h
-_18       a:t#:arah
+_11       gj'a:*VH
+_12       b'a:rVH
+_13       t'e:rVH
+_14       c'O:dVH
+_15       p'VndrVH
+_16       s'o:lVH
+_17       sVtrVh
+_18       Vt.#'a:rVH
 _19       Unn'Is
 
 _20       b'i:s
-_21       Ik:'Is
-_22       b'a:Is
-_23       t'ue:Is
-_24       c'O:bIs
-_25       p@c:'Is
-_26       c#@bb'i:rs
-_27       s@t:'a:Is
-_28       at#:'a:Is
-_29       ,Unt'Is
+_21       Ik:'is
+_22       b'a:is
+_23       t'e:is
+_24       c'O:bis
+_25       p'Vc:is
+_26       c#Vbb'is
+_27       s@t:'a:is
+_28       at#:'a:is
+_29       'Untis
 _30       t'i:s
 
-_31       Ik@t:'Is
-_32       b'at:'I'S
-_33       tE~t'Is      // E~ ?
-_34       cO~t'Is
-_35       pE~t'Is
-_36       c#aT'tIs
-_37       sE~t'Is
-_38       art'Is
-_39       Unta:l'Is
-
-_40       ca:l'Is
-_41       Ikta:l'Is
-_42       baIa:l'Is
-_43       tE~ta:l'Is
-_44       cO:wa:l'Is
-_45       paI'nta:l'Is
-_46       c#hIja:l'Is
-_47       sE~ta:l'Is
-_48       arta:l'Is
-_49       'Un^ca:s
-
-_50       p@c'a:s
+_31       Ik@t'is
+_32       b'at:is
+_33       t'E~tis      // E~ ?
+_34       c'O~tis
+_35       p'E~tis
+_36       c#'Vt:is
+_37       s'E~tis
+_38       'Vrtis
+_39       Unt'a:lis
+
+_40       c'a:lis
+_41       Ikt'a:lis
+_42       baI'a:lis
+_43       tE~t'a:lis
+_44       cO:v'a:lis
+_45       paInt'a:lis
+_46       c#Ij'a:lis
+_47       sE~t'a:lis
+_48       Vrt'a:lis
+_49       'Unca:s
+
+_50       pVc'a:s
 _51       Ikj'a:wVn
 _52       b'a:wVn
-_53       t'IrpVn
-_54       c'Ou:wVn
-_55       p'ac#pVn
-_56       c#h@p:'Vn
-_57       santa:'w@n
-_58       at#:a:'w@n
-_59       Uns'Vt#
+_53       t'Ire:pVn
+_54       c'O:wVn
+_55       p'VcpVn
+_56       c#'Vp:Vn
+_57       sVt'a:w@n
+_58       at.#:'a:w@n
+_59       'UnsVt.#
 
 _60       s'a:t.#
-_61       'Iks@t.#
-_62       b'a:s@t.#
-_63       t'Irs@t.#
-_64       cO:'s@t.#
-_65       paI's@t.#
-_66       c#'Ija:s@t.#
-_67       s'Vrs@t.#
-_68       ars@t.#
-_69       Un@h,at:'Vr:
-
-_70       sat:'Vr:
-_71       Ik#at:'Vr:
-_72       b@hat:'Vr:
-_73       tIhat:'Vr:
-_74       cO:hat:'Vr:
-_75       pac#,at:'Vr:
-_76       c#Ihat:'Vr:
-_77       s@t@hat:'Vr:
-_78       at#hat:'Vr:
-_79       ,Un^Ia:s'i:
-
-_80       ,Vss'i:
-_81       Ikja:s'i
-_82       baIa:s'i
-_83       tIra:s'i
-_84       cO:ra:s'i
-_85       pI'ca:s'i
-_86       c#Ija:s'i
-_87       sVt:a:s'i
-_88       at#:a:s'i
-_89       nVwa:s'i
-
-_90       nVb:'e:
-_91       Ikj'a:n@we:
-_92       b'a:n@we:
-_93       tIr'a:n@we:
-_94       cO:r'a:m@we:
-_95       p@c'a:n@we:
-_96       c#Ij'a:n@we:
-_97       s@t:'a:n@we:
-_98       at#:'a:n@we:
-_99       nIn^j'a:n@we:
+_61       'IksVt.#
+_62       b'a:sVt.#
+_63       t'Ire:sVt.#
+_64       c'O~sVt.#
+_65       p'E~sVt.#
+_66       c#Ij'a:sVt.#
+_67       s'Vr.sVt.#
+_68       Vr.sVt.#
+_69       'UnVht:Vr
+
+_70       s'Vt:Vr
+_71       'IkHVt:Vr
+_72       'bVhVt:Vr
+_73       t'IhVt:Vr
+_74       cO:H'Vt:Vr
+_75       pVcH'Vt:Vr
+_76       c#IH'Vt:Vr
+_77       sVtH'Vt:Vr
+_78       VtH'Vt:Vr
+_79       Un'a:si
+
+_80       'Vssi:
+_81       Ikj'a:si
+_82       bVj'a:si
+_83       tIr'a:si
+_84       cO:r'a:si
+_85       pVc'a:si
+_86       c#Ij'a:si
+_87       sVt:'a:si
+_88       at#:'a:si
+_89       nVw'a:si
+
+_90       n'Vb:e:
+_91       Ik:j'a:n@be:
+_92       b'a:n@be:
+_93       tIr'a:n@be:
+_94       cO:r'a:m@be:
+_95       pVnc'a:n@be:
+_96       c#Ij'a:n@be:
+_97       sVt:'a:n@be:
+_98       Vt#:'a:n@be:
+_99       nIn^j'a:n@be:
 
 _0C	s'O:
-_0M1	h@J,a:r
+_0M1	h'VJa:r
 _0M2	l'a:k#
-_0M3	k@ro:r.V
-_0M4	'Vr@b
-_0M5	k#'Vr@b
-_dpt      _d@s@ml'o:_        // ?? what is Hi for "." ?
+_0M3	k@r'o:r.V
+_0M4	'VrVb
+_0M5	k#'VrVb
+_dpt      _d@s@ml'o:_   
 
 // Single consonants
 क      k@

dictsource/hi_rules

 
 .group ै
         ै          E:        // ?? [aI] [E:] [&:]
+        ैं         E~ 
         ै (ँ       E~ 
         ैं (_      E~
         ैँ (_      E~
         क़         qV
         क़ (B      q
 
+        क्क         k:V   // kk -> k:
+        क्क (B      k:
         क्ष          kSV
         क्ष (B       kS
         
         ट          t.V
         ट (B       t.
         ट (्य      t.:
+        ट्ट          t.:V    // t.t. -> t.:
+        ट्ट (B       t.:
+        ट्ठ          t.#:V    // t.t.# -> t.#:
+        ट्ठ (B       t.#:
 
 .group ठ
         ठ          t.#V
         ठ (B       t.#
-
+        
+        ठ्ठ          t.#:V   // t.#t.# -> t.#:
+        ठ्ठ (B       t.#:
+        
 .group ड
         ड          d.V
         ड (B       d.
         त (B       t
         त (्य      t:
 
+        त्त          t:V   // tt -> t:
+        त्त (B       t:
+        
 .group थ
         थ          t#V
         थ (B       t#
         प          pV
         प (B       p
         प (्य      p:
+        
+        प्प         p:V      // pp -> p:
+        प्प (B      p:
 
 .group फ
         फ          p#V

dictsource/ne_rules

-// Nepali Language.
+
+// Nepali Language.
 // This ne_rules file is prepared by Him Prasad Gautam < [email protected] >.
 // This file is UTF-8 encoded.
 
 .group आ
 	आ	a:
 	आँ	a~''
-	आइ	a:I
-	आउ	a:U
-	आऊ	a:U'
 
 .group इ
 
 .group ा
 	ा	a:,k'a::r@ // If not followed to a Consonant.
 	C) ा	a:
-	ाव	aU
 	C) ाँ	a~'
 
 .group ि
 
 // The Latin letter a-z are automatically translated into English.
 
-

dictsource/sl_list

 w	dv#'ojniv,@
 x	iks
 y	'ipsilon
-ć	m'ehkitS,@
+ć	m'ehki||tS,@
 đ	dZ@
 
 ä	$accent
 _1	'Ena
 _2	dv#'a
 _3	tR'i
-_4	St'iRi
+_4	St'i*i
 _5	p'et
-_6	S'est
-_7	s'edem
-_8	'osem
-_9	d'evet
-_10	d'eset
+_6	S'Est
+_7	s'edEm
+_8	'Os@m
+_9	d@v'Et
+_10	d@s'Et
 _11	'e'najst
 _12	dv#'anajst
 _13	tR'inajst
 _15	p'etnajst
 _16	S'estnajst
 _17	s'edemnajst
-_18	'osemnajst
-_19	d'evetnajst
-_2X	dv#'ajset
-_3X	tR'ideset
-_4X	St'iRideset
-_5X	p'etdeset
-_6X	S'estdeset
-_7X	s'edemdeset
-_8X	'osemdeset
-_9X	d'evetdeset
+_18	'Osemnajst
+_19	d'evEtnajst
+_2X	dv#'ajsEt
+_3X	tR'id@sEt
+_4X	St'i*id@sEt
+_5X	p'etd@sEt
+_6X	S'estd@sEt
+_7X	s'edemd@sEt
+_8X	'Os@md@sEt
+_9X	d'evEtd@sEt
 _2C	dv'esto
 _0C	sto
 _0M1	t'isotS

dictsource/sl_rules

 .group č
         č        tS
 
+.group ć
+        ć        tS;
+        
 .group d
         d        d
         dz       dz
 .group r
         r        R
         rr       R
+     A) r (A     *
 
      K) r (K     @R
 
         v (-A+   v    // preposition 'v' before a vowel
      A) v (K     w
      K) v (K     %u
+     K) v (r     v
    L01) v (A     v#
 
 

phsource/ph_bengali

 phoneme a
   vowel starttype #a endtype #a
   length 225
+  IF thisPh(isWordEnd) AND thisPh(isUnstressed) THEN
+    FMT(vowel/a#_2)
+  ENDIF
   FMT(vowel/a_3)
 endphoneme
 
 
 
 phoneme tS
-  import_phoneme base/tS
+  vls pla afr sibilant
+  voicingswitch dZ
+  lengthmod 2
+  Vowelin f1=0  f2=2300 200 400  f3=-100 80
+
+  IF nextPh(isPause2) THEN
+    WAV(ustop/tsh_)
+  ENDIF
+  WAV(ustop/tsh_)
 endphoneme
 
 
 phoneme tS#
-  import_phoneme base/tS
+  vls pla afr sibilant
+  voicingswitch dZ
+  lengthmod 2
+  Vowelin f1=0  f2=2300 200 400  f3=-100 80
+
+  IF nextPh(isPause2) THEN
+    WAV(ustop/tsh_asp)
+  ENDIF
+  WAV(ustop/tsh_asp, 120)
 endphoneme
 
 
+
+
+
+
 phoneme ~
   virtual
 endphoneme

phsource/ph_german

     VowelStart(w/wi2)
   ENDIF
   
-  IF nextPh(r) AND next2Ph(isNotVowel) THEN
-    ChangePhoneme(iR)
-  ENDIF
+//  IF nextPh(r) AND next2Ph(isNotVowel) THEN
+//    ChangePhoneme(iR)
+//  ENDIF
   FMT(vowel/i_2)
 endphoneme
 
   vowel starttype #i endtype #@
   ipa iɐ
   flag1
-  length 230
+  length 240
   FMT(vdiph2/i@_3)
 endphoneme
 

phsource/ph_hindi

   IF nextPh(H) OR prevPh(H) THEN
     FMT(vowel/@_low)
   ENDIF
+  IF nextPh(j) THEN
+    FMT(vowel/V_3)
+  ENDIF
   FMT(vowel/@_bck)
 endphoneme
 
 
 phoneme V
   vowel flag1 starttype #@  endtype #@
-  length 130
+  length 140
 
   IF thisPh(isWordStart) THEN
     FMT(vowel/V_3)
   ENDIF
 
   IF thisPh(isNotStressed) THEN
-    IF prevPh(H) THEN
-      ChangePhoneme(@)
-    ENDIF
+    length 130
 
-    IF thisPh(isWordEnd) THEN
+    IF thisPh(isWordEnd) AND NOT thisPh(isFirstVowel) THEN
       ChangePhoneme(NULL)
     ENDIF
 
+    IF prevPh(H) THEN
+      ChangePhoneme(@)
+    ENDIF
+
     IF prev2PhW(isVowel) AND NOT nextVowel(isFlag1) THEN
       IF nextPhW(isNotVowel) AND nextPh(isWordEnd) THEN
         // don't delete schwa before a word-final consonant
         ChangePhoneme(NULL)
       ENDIF
     ENDIF
-  ENDIF
 
-  IF prevPh(isPalatal) THEN
-    ChangePhoneme(@3)
-  ELSEIF prevPh(r) THEN
-    ChangePhoneme(@2)
-  ELSE
-    IF thisPh(isNotStressed) THEN
+    IF prevPh(isPalatal) AND NOT nextPh(H) THEN
+      ChangePhoneme(@3)
+    ELSEIF prevPh(r) THEN
+      ChangePhoneme(@2)
+    ELSE
       ChangePhoneme(@)
     ENDIF
+  ENDIF
+  
+  IF nextPh(H) OR prevPh(H) THEN
+    FMT(vowel/@_low)
+  ENDIF
 
-    IF nextPh(H) OR prevPh(H) THEN
-      FMT(vowel/@_low)
-    ENDIF
+  IF prevPh(w) THEN
+    FMT(vowel/V_4)
   ENDIF
 
   FMT(vowel/V_3)
 phoneme i:
   vowel long starttype #i endtype #i
   length 230
+  IF thisPh(isUnstressed) THEN
+    ChangePhoneme(i)
+  ENDIF
   IfNextVowelAppend(;)
   FMT(vowel/i_fnt)
 endphoneme
   vowel starttype #i endtype #i
   length 230
   IfNextVowelAppend(;)
-  FMT(vowel/i_6)
+  FMT(vowel/ii_6)
 endphoneme
 
 phoneme I
 
 phoneme e
   vowel starttype #e endtype #e
-  length 130
+  length 140
   FMT(vowel/e_2)
 endphoneme
 
 
 phoneme E
   vowel starttype #e endtype #e
-  length 130
+  length 140
   FMT(vowel/e_mid)
 endphoneme
 
 
 phoneme a:
   vowel long starttype #a endtype #a
-  length 190
+  length 200
   IF nextPhW(j) THEN
     FMT(vowel/a_5)
   ENDIF
 
 phoneme a
   vowel long starttype #a endtype #a
-  length 130
+  length 140
   FMT(vowel/a#_3)
 endphoneme
 
 
 phoneme o
   vowel starttype #o endtype #o
-  length 130
+  length 140
   FMT(vowel/o)
 endphoneme
 
 
 phoneme O
   vowel starttype #o endtype #o
-  length 130
+  length 140
   FMT(vowel/0)
 endphoneme
 
 
 phoneme U
   vowel starttype #u endtype #u
-  length 120
+  length 140
   FMT(vowel/uu_bck)
 endphoneme
 
   voicingswitch d.#
   Vowelin f1=0  f2=1800 -300 300  f3=-400 80
   Vowelout f1=0 f2=1800 -300 250  f3=-400 80  rms=20 colr=2
-//  WAV(ustop/th_rfx)
-  WAV(ustop/t)
+  
+  IF nextPhW(t.#) THEN
+    ChangePhoneme(t.)
+  ENDIF
+  WAV(ustop/th_rfx)
 endphoneme
 
 phoneme d.
 // also try [tS] and [dZ]
 
 phoneme c
-  import_phoneme base/c
+  vls pla afr sibilant
+  voicingswitch dZ
+  lengthmod 2
+  Vowelin f1=0  f2=2300 200 400  f3=-100 80
+
+  IF nextPh(isPause2) THEN
+    WAV(ustop/tsh_)
+  ENDIF
+  WAV(ustop/tsh_unasp)
 endphoneme
 
 
+
 phoneme c#
   CALL base/tS;
   ipa cU+02B0

phsource/ph_slovenian

   vowel starttype #i endtype #i
   length 140
   IfNextVowelAppend(;)
+  IF thisPh(isStressed) AND NOT nextPhW(isRhotic) THEN
+    FMT(vowel/ii_6)
+  ENDIF
   FMT(vowel/ii_3)
 endphoneme
 
   FMT(vowel/e_2)
 endphoneme
 
+phoneme e:
+  vowel starttype #e endtype #e
+  length 240
+  FMT(vowel/e)
+endphoneme
+
 
 phoneme E
   vowel starttype #e endtype #e

phsource/phonemes

   lengthmod 3
   Vowelout f1=3 f2=1400 -400 300  f3=-400 80 rms=35 len=15 colr=2
   Vowelin f1=2  f2=1400 -400 300  f3=-400 80 len=20
+  
+  IF nextPh(isNotVowel) THEN
+    FMT(r3/@tap_rfx_)
+  ENDIF
   FMT(r3/@tap_rfx)
 endphoneme
 

platforms/windows/make_espeak.iss

 
 [Setup]
 AppName=eSpeak
-AppVerName=eSpeak version 1.47.08
+AppVerName=eSpeak version 1.47.09
 AppCopyright=Licensed under GNU General Public License version 3.   (See file License.txt for details).
 WindowVisible=yes
 
 ne.v2=अथवा यसमा निर्धारण गरिएको आवाजलाई नै चयन गर्ने हो भने ईन्टर कुञ्जीलाई दबाउनु होस् । 
 ne.v3=आवाजको नाम प्रविष्टी गर्नु होस् , जस्तै: नेपालीका लागि ne, अथवा स्वरको पनि चयन गर्ने हो भने ne+f3
 
+hi.v1=स्थापना करने के लिए पसन्दिदा आवाजों को चयन करें ।
+hi.v2=अगर आप कि चयन इसमें निर्दारित आवाज हि  हो तो केवल  ईन्टर कुञ्जी दबाए । 
+hi.v3=आवाज कि   नाम प्रविष्टी करें, जैसा कि हिन्दी के लिए hi, और किसि स्वर भि चयन करना चाहते हो तो hi+f3
+
 [Code]
 var
   UILanguage: Integer;

platforms/windows/make_espeakedit.iss

 
 [Setup]
 AppName=eSpeakEdit
-AppVerName=eSpeakEdit version 1.47.08
+AppVerName=eSpeakEdit version 1.47.09
 DefaultDirName={pf}\eSpeak
 DefaultGroupName=eSpeak
 OutputBaseFilename=setup_espeakedit

src/dictionary.cpp

 			*p2++ = (acc << (8-bits));
 		}
 		*p2 = 0;
-		strcpy(text, buf);
-		return((p2 - buf) | 0x40);  // bit 6 indicates compressed characters
+		ix = p2 - buf;
+		memcpy(text, buf, ix);
+		return(ix | 0x40);  // bit 6 indicates compressed characters
 	}
 	else
 	{

src/extras.cpp

 
 	while(!feof(f_in))
 	{
-		c = fgetc(f_in);
+		c = fgetc(f_in) & 0xff;
 		if(c >= 0xa0)
 			c = translator->charset_a0[c-0xa0];
 
 
 	wxLogStatus(_T("Written to: ")+fname+_T("_1"));
 
-}  // end of ConvertToItf8
+}  // end of ConvertToUtf8
 
 //******************************************************************************************************
 

src/readclause.cpp

   0x061f,  // Arabic ?
   0x06d4,  // Arabic .
 
+  0x0df4,  // Singhalese Kunddaliya
   0x0f0d,  // Tibet Shad
   0x0f0e,
 
   CLAUSE_QUESTION,   // Arabic question mark
   CLAUSE_PERIOD,     // Arabic full stop
 
+  CLAUSE_PERIOD+0x8000,     // Singhalese period
   CLAUSE_PERIOD+0x8000,     // Tibet period
   CLAUSE_PARAGRAPH,
 

src/synthdata.cpp

 #include "translate.h"
 #include "wave.h"
 
-const char *version_string = "1.47.08  26.Apr.13";
-const int version_phdata  = 0x014701;
+const char *version_string = "1.47.09  29.Apr.13";
+const int version_phdata  = 0x014709;
 
 int option_device_number = -1;
 FILE *f_logespeak = NULL;

src/tr_languages.cpp

     {"_ar",    OFFSET_ARABIC,   0x600, 0x6ff,  0, 0},
     {"_dv",    OFFSET_THAANA,   0x780, 0x7bf,  0, 0},
     {"_hi",    OFFSET_DEVANAGARI, 0x900, 0x97f,L('h','i'), AL_WORDS},
-    {"_bn",    OFFSET_BENGALI,  0x0980, 0x9ff, L('b','n'), 0},
+    {"_bn",    OFFSET_BENGALI,  0x0980, 0x9ff, L('b','n'), AL_WORDS},
     {"_gur",   OFFSET_GURMUKHI, 0xa00, 0xa7f,  L('p','a'), AL_WORDS},
     {"_gu",    OFFSET_GUJARATI, 0xa80, 0xaff,  0, 0},
     {"_or",    OFFSET_ORIYA,    0xb00, 0xb7f,  0, 0},
     {"_te",    OFFSET_TELUGU,   0xc00, 0xc7f,  L('t','e'), 0},
     {"_kn",    OFFSET_KANNADA,  0xc80, 0xcff,  L('k','n'), AL_WORDS},
     {"_ml",    OFFSET_MALAYALAM,0xd00, 0xd7f,  L('m','l'), AL_WORDS},
-    {"_si",    OFFSET_SINHALA,  0xd80, 0xdff,  0, 0},
+    {"_si",    OFFSET_SINHALA,  0xd80, 0xdff,  L('s','i'), AL_WORDS},
     {"_th",    OFFSET_THAI,     0xe00, 0xe7f,  0, 0},
     {"_lo",    OFFSET_LAO,      0xe80, 0xeff,  0, 0},
     {"_ti",    OFFSET_TIBET,    0xf00, 0xfff,  0, 0},