Leaked source code of windows server 2003
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

775 lines
48 KiB

  1. /******************************************************************************
  2. * MiscData.cpp *
  3. *--------------*
  4. * This file stores the const data used in various non-normalization front-end
  5. * code
  6. *------------------------------------------------------------------------------
  7. * Copyright (C) 1999 Microsoft Corporation Date: 05/02/2000
  8. * All Rights Reserved
  9. *
  10. ****************************************************************** AARONHAL ***/
  11. /******************************************************************************
  12. !!!!! IMPORTANT !!!!!
  13. Compiler optimizations need to be turned off for this file!!! If they
  14. are not, the duplicate strings in the abbreviation table below will
  15. reference the same memory, and their initialization will FAIL!!! In
  16. addition, this requires turning off precompiled headers for this file.
  17. DO NOT CHANGE THESE SETTINGS!!!
  18. ******************************************************************************/
  19. #include"stdsentenum.h"
  20. CComAutoCriticalSection g_AbbrevTableCritSec;
  21. bool g_AbbrevTableInitialized = false;
  22. const BrillPatch g_POSTaggerPatches [] =
  23. {
  24. { MS_Adj, MS_Noun, CAP, MS_Unknown, MS_Unknown, NULL, NULL },
  25. { MS_Verb, MS_Noun, PREV1T, MS_Adj, MS_Unknown, NULL, NULL },
  26. { MS_Verb, MS_Noun, CAP, MS_Unknown, MS_Unknown, NULL, NULL },
  27. { MS_Noun, MS_Verb, PREV1T, MS_VAux, MS_Unknown, NULL, NULL },
  28. { MS_Conj, MS_Adv, NEXT2T, MS_Conj, MS_Unknown, NULL, NULL },
  29. { MS_Adj, MS_Adv, NEXT1T, MS_Verb, MS_Unknown, NULL, NULL },
  30. { MS_Adj, MS_Adv, PREV1TNEXT1T, MS_Noun, MS_Unknown, NULL, NULL },
  31. { MS_Verb, MS_Noun, PREV2T, MS_Prep, MS_Unknown, NULL, NULL },
  32. { MS_Noun, MS_Verb, PREV1T, MS_SubjPron, MS_Unknown, NULL, NULL },
  33. { MS_Noun, MS_Verb, PREV1T, MS_Pron, MS_Unknown, NULL, NULL },
  34. { MS_Noun, MS_Verb, PREV1T, MS_Adv, MS_Unknown, NULL, NULL },
  35. { MS_Verb, MS_Noun, NEXT1T, MS_VAux, MS_Unknown, NULL, NULL },
  36. { MS_Adj, MS_Adv, PREV1TNEXT1T, MS_Verb, MS_Adj, NULL, NULL },
  37. { MS_Verb, MS_Noun, PREV1TNEXT1T, MS_Noun, MS_Verb, NULL, NULL },
  38. { MS_Noun, MS_Adj, NEXT1T, MS_Adj, MS_Unknown, NULL, NULL },
  39. { MS_Verb, MS_Noun, NEXT1T, MS_CConj, MS_Unknown, NULL, NULL },
  40. { MS_Adj, MS_Verb, PREV1T, MS_VAux, MS_Unknown, NULL, NULL },
  41. { MS_Noun, MS_Verb, PREV1TNEXT1T, MS_Prep, MS_Adv, NULL, NULL },
  42. { MS_Noun, MS_Verb, PREV1TNEXT1T, MS_Prep, MS_Adj, NULL, NULL },
  43. { MS_VAux, MS_Noun, CAP, MS_Unknown, MS_Unknown, NULL, NULL },
  44. { MS_Adv, MS_Prep, PREV1TNEXT1T, MS_Noun, MS_Prep, NULL, NULL },
  45. { MS_Adv, MS_Prep, PREV1TNEXT2T, MS_Verb, MS_Noun, NULL, NULL },
  46. { MS_Adj, MS_Adv, NEXT1T, MS_Adv, MS_Unknown, NULL, NULL },
  47. { MS_Noun, MS_Verb, NEXT1T, MS_ObjPron, MS_Unknown, NULL, NULL },
  48. { MS_Adv, MS_Adj, PREV1TNEXT1T, MS_Prep, MS_Noun, NULL, NULL },
  49. { MS_Noun, MS_Verb, PREV1TNEXT1T, MS_Interr, MS_Prep, NULL, NULL },
  50. { MS_Adj, MS_Adv, PREV1TNEXT1T, MS_Adv, MS_Adj, NULL, NULL },
  51. { MS_Prep, MS_Verb, PREV1T, MS_VAux, MS_Unknown, NULL, NULL },
  52. { MS_Adv, MS_Conj, NEXT1T, MS_SubjPron, MS_Unknown, NULL, NULL },
  53. { MS_Adj, MS_Noun, PREV1TNEXT1T, MS_Adj, MS_Prep, NULL, NULL },
  54. { MS_Noun, MS_Verb, PREV1TNEXT2T, MS_Interr, MS_Noun, NULL, NULL },
  55. { MS_Noun, MS_Verb, PREV1T, MS_VAux, MS_Unknown, NULL, NULL },
  56. { MS_Verb, MS_Noun, PREV1TNEXT1T, MS_Unknown, MS_Verb, NULL, NULL },
  57. { MS_Adj, MS_Verb, PREV1TNEXT1T, MS_Verb, MS_Adj, NULL, NULL },
  58. { MS_Conj, MS_Adv, PREV1TNEXT1T, MS_Unknown, MS_Unknown, NULL, NULL },
  59. { MS_Adv, MS_Prep, PREV1TNEXT2T, MS_Prep, MS_Unknown, NULL, NULL },
  60. { MS_Adj, MS_Adv, PREV2TNEXT1T, MS_Verb, MS_Adj, NULL, NULL },
  61. { MS_Adv, MS_Prep, NEXT1T, MS_Noun, MS_Unknown, NULL, NULL },
  62. { MS_Verb, MS_Noun, PREV1TNEXT1T, MS_Conj, MS_Verb, NULL, NULL },
  63. { MS_Adj, MS_Noun, PREV1TNEXT1T, MS_Prep, MS_Prep, NULL, NULL },
  64. { MS_Noun, MS_Verb, PREV1TNEXT1T, MS_Prep, MS_Pron, NULL, NULL },
  65. { MS_Adj, MS_Noun, PREV1TNEXT1T, MS_Noun, MS_Verb, NULL, NULL },
  66. { MS_Adj, MS_Adv, PREV1T, MS_VAux, MS_Unknown, NULL, NULL },
  67. { MS_Adj, MS_Noun, NEXT1T, MS_VAux, MS_Unknown, NULL, NULL },
  68. { MS_Adj, MS_Adv, NEXT1T, MS_VAux, MS_Unknown, NULL, NULL },
  69. { MS_Adj, MS_Noun, PREV1TNEXT1T, MS_Unknown, MS_Prep, NULL, NULL },
  70. { MS_Verb, MS_Noun, PREV2TNEXT1T, MS_Unknown, MS_Verb, NULL, NULL },
  71. { MS_Noun, MS_Verb, PREV1T, MS_SubjPron, MS_Unknown, NULL, NULL },
  72. { MS_Adj, MS_Verb, PREV1TNEXT1T, MS_Prep, MS_Adv, NULL, NULL },
  73. { MS_Adv, MS_Adj, PREV1TNEXT1T, MS_Conj, MS_Noun, NULL, NULL },
  74. { MS_Conj, MS_Adv, PREV1TNEXT1T, MS_Prep, MS_Adj, NULL, NULL },
  75. { MS_Adv, MS_Prep, NEXT1T, MS_Verb, MS_Unknown, NULL, NULL },
  76. { MS_Noun, MS_Adj, NEXT1T, MS_Adj, MS_Unknown, NULL, NULL },
  77. { MS_Adv, MS_Conj, NEXT1T, MS_Pron, MS_Unknown, NULL, NULL },
  78. { MS_Conj, MS_Adv, PREV1T, MS_VAux, MS_Unknown, NULL, NULL },
  79. { MS_Noun, MS_Verb, PREV1T, MS_Adv, MS_Unknown, NULL, NULL },
  80. { MS_Verb, MS_Noun, PREV2TNEXT1T, MS_CConj, MS_Unknown, NULL, NULL },
  81. { MS_Noun, MS_Verb, PREV1T, MS_VAux, MS_Unknown, NULL, NULL },
  82. { MS_Noun, MS_Verb, PREV1T, MS_Pron, MS_Unknown, NULL, NULL },
  83. { MS_Adv, MS_Adj, PREV2TNEXT1T, MS_Prep, MS_Noun, NULL, NULL },
  84. { MS_Noun, MS_Adj, PREV1TNEXT1T, MS_Adv, MS_Noun, NULL, NULL },
  85. { MS_Conj, MS_Adv, PREV1TNEXT1T, MS_Adv, MS_Adj, NULL, NULL },
  86. };
  87. //--- IMPORTANT!!! This list must remain alphabetized for the binary search algorithm to work properly!!!
  88. AbbrevRecord g_AbbreviationTable[] =
  89. {
  90. //--- Initial - never EOS
  91. { L"a", L"EY 1", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  92. { L"amp", L"AE 1 M P", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  93. { L"app", L"AE 1 P", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  94. { L"apr", L"EY 1 - P R AX L", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  95. { L"apt", L"ax - p aa 1 r t - m ax n t", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  96. { L"assoc", L"ax - s ow 2 - s iy - ey 1 - sh ax n", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  97. { L"asst", L"ax - s ih 1 s - t ax n t", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0 -1 },
  98. { L"aug", L"ao 1 - g ax s t", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  99. { L"ave", L"ae 1 v - ax - n uw 2", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  100. //--- Initial - never EOS
  101. { L"b", L"B IY 1", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  102. { L"bldg", L"b ih 1 l - d ih ng", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  103. { L"blvd", L"b uh 1 l - ax - v aa 2 r d", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  104. //--- SingleOrPluralAbbreviation
  105. { L"bu", L"b uh 1 sh - ax l", MS_Noun, L"b uh 1 sh - ax l Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  106. //--- Initial - never EOS
  107. { L"c", L"S IY 1", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  108. //--- SingleOrPluralAbbreviation
  109. { L"cal", L"k ae 1 l - ax - r iy", MS_Noun, L"k ae 1 l - ax - r iy Z", MS_Noun, L"K AE 1 L", MS_Noun, 0, 0 },
  110. { L"cals", L"k ae 1 l - ax - r iy Z", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  111. { L"ch", L"ch ae 1 p - t er", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  112. //--- SingleOrPluralAbbreviation
  113. { L"cl", L"s eh 1 n - t ax - l iy 2 - t er", MS_Noun, L"s eh 1 n - t ax - l iy 2 - t er Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  114. //--- SingleOrPluralAbbreviation
  115. { L"cm", L"s eh 1 n - t ax - m iy 2 - t er", MS_Noun, L"s eh 1 n - t ax - m iy 2 - t er Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  116. { L"cms", L"s eh 1 n - t ax - m iy 2 - t er Z", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  117. { L"cntr", L"s eh 1 n - t er", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  118. //--- IfEOSNotAbbreviation
  119. { L"co", L"k ah 1 m - p ax - n iy", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 1, -1 },
  120. { L"cont", L"k ax n - t ih 1 n - y uw D", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  121. { L"corp", L"k ao 2 r - p ax - r ey 1 - sh ax n", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  122. { L"ctr", L"s eh 1 n - t er", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  123. { L"cu", L"k y uw 1 - b ih k", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  124. //--- Initial - never EOS
  125. { L"d", L"D IY 1", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  126. //--- SingleOrPluralAbbreviation
  127. { L"db", L"d eh 1 s - ax - b ax l", MS_Noun, L"d eh 1 s - ax - b ax l Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  128. { L"dec", L"d ih - s eh 1 m - b er", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  129. //--- SingleOrPluralAbbreviation
  130. { L"deg", L"d ih - g r iy 1", MS_Noun, L"d ih - g r iy 1 z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  131. { L"dept", L"d ih - p aa 1 r t - m ax n t", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  132. { L"dist", L"d ih 1 s - t r ax k t", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  133. { L"doc", L"D AA 1 K", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  134. //--- SingleOrPluralAbbreviation
  135. { L"doz", L"d ah 1 z - ax n", MS_Noun, L"d ah 1 z - ax n Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  136. //--- DoctorDriveAbbreviation
  137. { L"dr", L"d aa 1 k - t er", MS_Noun, L"D R AY 1 V", MS_Noun, NULL, MS_Unknown, 0, 1 },
  138. //--- Initial - never EOS
  139. { L"e", L"IY 1", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  140. //--- IfEOSNotAbbreviation
  141. { L"ed", L"eh 2 jh - ax - k ey 1 - sh ax n", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 1, -1 },
  142. { L"esq", L"eh 1 s - k w ay 2 r", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  143. { L"est", L"ax - s t ae 1 b - l ax sh T", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  144. { L"etc", L"EH T & s eh 1 t - er - ax", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  145. //--- IfEOSNotAbbreviation
  146. { L"ex", L"ih g - z ae 1 m - p ax l", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 1, -1 },
  147. { L"ext", L"ih k - s t eh 1 n - sh ax n", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  148. //--- Initial - never EOS
  149. { L"f", L"EH 1 F", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  150. { L"feb", L"f eh 1 b - r uw - eh 2 r - iy", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  151. //--- IfEOSNotAbbreviation
  152. { L"fig", L"f ih 1 g - y er", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 1, -1 },
  153. //--- IfEOSNotAbbreviation
  154. { L"figs", L"f ih 1 g - y er Z", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 1, -1 },
  155. { L"fn", L"f uh 1 t - n ow 2 t", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  156. { L"freq", L"f r iy 1 - k w ax n - s iy", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  157. { L"fri", L"f r ay 1 - d ey", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  158. //--- SingleOrPluralAbbreviation
  159. { L"ft", L"F UH 1 T", MS_Noun, L"F IY 1 T", MS_Noun, L"F AO 1 R T", MS_Noun, 0, 0 },
  160. { L"fwd", L"f ao 1 r - w er d", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  161. //--- SingleOrPluralAbbreviation
  162. { L"g", L"G R AE 1 M", MS_Noun, L"G R AE 1 M Z", MS_Noun, L"JH IY 1", MS_Noun, 0, 0 },
  163. //--- SingleOrPluralAbbreviation
  164. { L"gal", L"g ae 1 l - ax n", MS_Noun, L"g ae 1 l - ax n Z", MS_Noun, L"G AE 1 L", MS_Noun, 0, 0 },
  165. //--- TITLE - never EOS
  166. { L"gen", L"jh eh 1 n - er - ax l", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  167. //--- IfEOSNotAbbreviation
  168. { L"gov", L"g ah 1 v - er - n ER", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 1, -1 },
  169. //--- Initial - never EOS
  170. { L"h", L"EY 1 CH", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  171. //--- IfEOSNotAbbreviation
  172. { L"hr", L"AW 1 ER", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 1, -1 },
  173. { L"hrs", L"AW 1 ER Z", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  174. { L"ht", L"H AY 1 T", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  175. { L"hz", L"H ER 1 T S", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  176. //--- Initial - never EOS
  177. { L"i", L"AY 1", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  178. //--- SingleOrPluralAbbreviation
  179. { L"in", L"IH 1 N CH", MS_Noun, L"IH 1 N CH AX Z", MS_Noun, L"IH 1 N", MS_Prep, 0, 0 },
  180. { L"inc", L"ih n - k ao 1 r - p ax - r ey 2 - t AX D", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  181. { L"incl", L"ih n - k l uw 1 - d AX D", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  182. { L"intl", L"ih 2 n - t er - n ae 1 sh - ax - n ax l", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  183. //--- IfEOSAndLowercaseNotAbbreviation
  184. { L"is", L"ay 1 - l ax n d", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 2, -1 },
  185. //--- SingleOrPluralAbbreviation
  186. { L"j", L"JH UW 1 L", MS_Noun, L"JH UW 1 L Z", MS_Noun, L"JH EY 1", MS_Noun, 0, 0 },
  187. //--- IfEOSNotAbbreviation
  188. { L"jan", L"jh ae 1 n - y uw - eh 2 r - iy", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 1, -1 },
  189. { L"jr", L"jh uw 1 n - y er", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  190. { L"jul", L"jh uh - l ay 1", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  191. { L"jun", L"JH UW 1 N", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  192. //--- Initial - never EOS
  193. { L"k", L"K AY 1", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  194. //--- SingleOrPluralAbbreviation
  195. { L"kb", L"k ih 1 l - ax - b ay 2 t", MS_Noun, L"k ih 1 l - ax - b ay 2 t S", MS_Noun, NULL, MS_Unknown, 0, 0 },
  196. //--- SingleOrPluralAbbreviation
  197. { L"kcal", L"k ih 1 l - ax - k ae 2 l - ax - r iy", MS_Noun, L"k ih 1 l - ax - k ae 2 l - ax - r iy Z", MS_Noun, L"K AE 1 L", MS_Noun, 0, 0 },
  198. //--- SingleOrPluralAbbreviation
  199. { L"kg", L"k ih 1 l - ax - g r ae 2 m", MS_Noun, L"k ih 1 l - ax - g r ae 2 m Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  200. { L"kgs", L"k ih 1 l - ax - g r ae 2 m Z", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  201. { L"khz", L"k ih 1 l - ax - h er 2 t s", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  202. //--- SingleOrPluralAbbreviation
  203. { L"kj", L"k ih 1 l - ax - jh uw 2 l", MS_Noun, L"k ih 1 l - ax - jh uw 2 l z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  204. //--- SingleOrPluralAbbreviation
  205. { L"km", L"k ih - l aa 1 m - ih - t er", MS_Noun, L"k ih - l aa 1 m - ih - t er Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  206. //--- SingleOrPluralAbbreviation
  207. { L"kw", L"k ih 1 l - ax - w aa 2 t", MS_Noun, L"k ih 1 l - ax - w aa 2 t S", MS_Noun, NULL, MS_Unknown, 0, 0 },
  208. //--- SingleOrPluralAbbreviation
  209. { L"l", L"l iy 1 - t er", MS_Noun, L"l iy 1 - t er Z", MS_Noun, L"EH 1 L", MS_Noun, 0, 0 },
  210. //--- SingleOrPluralAbbreviation
  211. { L"lb", L"P AW 1 N D", MS_Noun, L"P AW 1 N D Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  212. { L"lbs", L"P AW 1 N D Z", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  213. { L"lg", L"L AA 1 R JH", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  214. //--- Title - never EOS
  215. { L"lt", L"l uw - t eh 1 n - ax n t", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  216. { L"ltd", L"l ih 1 m - ih - t AX D", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  217. //--- SingleOrPluralAbbreviation
  218. { L"m", L"M IY 1 T ER", MS_Noun, L"M IY 1 T ER Z", MS_Noun, L"EH 1 M", MS_Noun, 0, 0 },
  219. //--- IfEOSNotAbbreviation???
  220. { L"mar", L"M AA 1 R CH", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  221. //--- SingleOrPluralAbbreviation
  222. { L"mb", L"m eh 1 g - ax - b ay 2 t", MS_Noun, L"m eh 1 g - ax - b ay 2 t S", MS_Noun, NULL, MS_Unknown, 0, 0 },
  223. { L"mfg", L"m ae 2 n - y ax - f ae 1 k - ch er - IH NG", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  224. //--- SingleOrPluralAbbreviation
  225. { L"mg", L"m ih 1 l - ax - g r ae 2 m", MS_Noun, L"m ih 1 l - ax - g r ae 2 m Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  226. { L"mgr", L"m ae 1 n - ih - jh ER", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  227. { L"mgs", L"m ih 1 l - ax - g r ae 2 m Z", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  228. { L"mhz", L"m eh 1 g - ax - h er 2 t s", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  229. //--- SingleOrPluralAbbreviation
  230. { L"mi", L"M AY 1 L", MS_Noun, L"M AY 1 L Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  231. { L"mic", L"M AY 1 K", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  232. //--- SingleOrPluralAbbreviation
  233. { L"min", L"m ih 1 n - ax t", MS_Noun, L"m ih 1 n - ax t S", MS_Noun, L"m ih 1 n - ax - m ax m", MS_Noun, 0, 0 },
  234. { L"misc", L"m ih 2 s - ax - l ey 1 - n iy - ax s", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  235. //--- SingleOrPluralAbbreviation
  236. { L"ml", L"m ih 1 l - ax - l iy 2 - t er", MS_Noun, L"m ih 1 l - ax - l iy 2 - t er Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  237. //--- SingleOrPluralAbbreviation
  238. { L"mm", L"m ih 1 l - ax - m iy 2 - t er", MS_Noun, L"m ih 1 l - ax - m iy 2 - t er Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  239. { L"mon", L"m ah 1 n - d ey", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  240. //--- TITLE - never EOS
  241. { L"mr", L"M IH 1 S - T ER", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  242. //--- TITLE - never EOS
  243. { L"mrs", L"M IH 1 S - AX Z", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  244. //--- IfEOSNotAbbreviation
  245. { L"ms", L"M IH 1 Z", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 1, -1 },
  246. //--- SingleOrPluralAbbreviation
  247. { L"msec", L"m ih 2 l - ax - s eh 1 k - ax n d", MS_Noun, L"m ih 2 l - ax - s eh 1 k - ax n d Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  248. { L"msecs", L"m ih 2 l - ax - s eh 1 k - ax n d Z", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  249. { L"mt", L"M AW 1 N T", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  250. { L"mtn", L"m aw 1 n - t ax n", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  251. //--- Initial - never EOS
  252. { L"n", L"EH 1 N", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  253. //--- IfEOSNotAbbreviation
  254. { L"no", L"N UH 1 M - B ER", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 1, -1 },
  255. { L"nov", L"n ow - v eh 1 m - b er", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  256. //--- Initial - never EOS
  257. { L"o", L"OW 1", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  258. { L"oct", L"aa k - t ow 1 - b er", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  259. //--- IfEOSNotAbbreviation
  260. { L"op", L"OW 1 - P AX S", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 1, -1 },
  261. //--- SingleOrPluralAbbreviation
  262. { L"oz", L"AW 1 N S", MS_Noun, L"AW 1 N - S AX Z", MS_Noun, L"AA 1 Z", MS_Noun, 0, 0 },
  263. { L"ozs", L"AW 1 N - S AX Z", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  264. //--- AbbreviationFollowedByDigit
  265. { L"p", L"P EY 1 JH", MS_Noun, L"P IY 1", MS_Noun, NULL, MS_Unknown, 0, 2 },
  266. //--- IfEOSNotAbbreviation
  267. { L"pg", L"P EY 1 JH", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 1, -1 },
  268. { L"pgs", L"P EY 1 - JH AX Z", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  269. { L"pkg", L"p ae 1 k - ih jh", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  270. { L"pkwy", L"p aa 1 r k - w ey 2", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  271. { L"pl", L"P L EY 1 S", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  272. { L"pp", L"P EY 1 - JH AX Z", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  273. //--- TITLE - never EOS
  274. { L"pres", L"p r eh 1 z - ax - d ax n t", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  275. //--- TITLE - never EOS
  276. { L"prof", L"p r ax - f eh 1 - s ER", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  277. { L"pt", L"P OY 1 N T", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  278. //--- Initial - never EOS
  279. { L"q", L"K Y UW 1", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  280. //--- SingleOrPluralAbbreviation
  281. { L"qt", L"k w ao 1 r t", MS_Noun, L"k w ao 1 r t Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  282. //--- Initial - never EOS
  283. { L"r", L"AA 1 R", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  284. { L"rec", L"R EH 1 K", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  285. { L"rt", L"R UW 1 T", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  286. { L"rte", L"R UW 1 T", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  287. //--- Initial - never EOS
  288. { L"s", L"EH 1 S", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  289. //--- IfEOSAndLowercaseNotAbbreviation
  290. { L"sat", L"s ae 1 t - er - d ey", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 2, -1 },
  291. //--- IfEOSNotAbbreviation
  292. { L"sec", L"s eh 1 k - ax n d", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 1, -1 },
  293. //--- TITLE - never EOS
  294. { L"sen", L"s eh 1 n - ax - t ER", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  295. { L"sep", L"s eh p - t eh 1 m - b er", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  296. { L"sept", L"s eh p - t eh 1 m - b er", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  297. { L"sm", L"S M AO 1 L", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  298. { L"sq", L"S K W EH 1 R", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  299. //--- IfEOSNotAbbreviation
  300. { L"sr", L"s iy 1 n - y er", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 1, -1 },
  301. //--- DoctorDriveAbbreviation
  302. { L"st", L"S EY 1 N T", MS_Noun, L"S T R IY 1 T", MS_Noun, NULL, MS_Unknown, 0, 1 },
  303. //--- IfEOSNotAbbreviation
  304. { L"sun", L"s ah 1 n - d ey", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 1, -1 },
  305. //--- Initial - never EOS
  306. { L"t", L"T IY 1", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  307. //--- SingleOrPluralAbbreviation
  308. { L"tbsp", L"t ey 1 - b ax l - s p uw 2 n", MS_Noun, L"t ey 1 - b ax l - s p uw 2 n Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  309. { L"tech", L"T EH 1 K", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  310. { L"thu", L"th er 1 z - d ey", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  311. { L"thur", L"th er 1 z - d ey", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  312. { L"thurs", L"th er 1 z - d ey", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  313. //--- SingleOrPluralAbbreviation
  314. { L"tsp", L"t iy 1 - s p uw 2 n", MS_Noun, L"t iy 1 - s p uw 2 n Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  315. { L"tue", L"t uw 1 z - d ey", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  316. { L"tues", L"t uw 1 z - d ey", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  317. //--- Initial - never EOS
  318. { L"u", L"Y UW 1", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  319. { L"univ", L"y uw 2 - n ax - v er 1 - s ih - t iy", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  320. //--- Initial - never EOS
  321. { L"v", L"V IY 1", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  322. { L"ver", L"v er 1 - zh ax n", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  323. { L"vers", L"v er 1 - zh ax n", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  324. { L"vol", L"v aa 1 l - y uw m", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  325. { L"vs", L"v er 1 - s ax s", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  326. //--- Initial - never EOS
  327. { L"w", L"d ah 1 b - ax l - y uw", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  328. //--- IfEOSAndLowercaseNotAbbreviation
  329. { L"wed", L"w eh 1 n z - d ey", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 2, -1 },
  330. { L"wk", L"W IY 1 K", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  331. { L"wt", L"W EY 1 T", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  332. { L"wy", L"W EY 1", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  333. //--- Initial - never EOS
  334. { L"x", L"EH 1 K S", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  335. //--- Initial - never EOS
  336. { L"y", L"W AY 1", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  337. //--- SingleOrPluralAbbreviation
  338. { L"yd", L"Y AA 1 R D", MS_Noun, L"Y AA 1 R D Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  339. { L"yds", L"Y AA 1 R D Z", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  340. { L"yr", L"Y IY 1 R", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, 0, -1 },
  341. //--- Initial - never EOS
  342. { L"z", L"Z IY 1", MS_Noun, NULL, MS_Unknown, NULL, MS_Unknown, -1, -1 },
  343. };
  344. //--- IMPORTANT!!! This list must remain alphabetized for the binary search algorithm to work properly!!!
  345. AbbrevRecord g_AmbiguousWordTable[] =
  346. {
  347. //--- AllCapsAbbreviation
  348. { L"al", L"EY 1 & EH 1 L", MS_Noun, L"AE 1 L", MS_Noun, NULL, MS_Unknown, 0, 3 },
  349. //--- AllCapsAbbreviation
  350. { L"apr", L"ey 1 & p iy 1 & aa 1 r", MS_Noun, L"ey 1 - p r ax l", MS_Noun, NULL, MS_Unknown, 0, 3 },
  351. //--- AllCapsAbbreviation
  352. { L"as", L"EY 1 & EH 1 S", MS_Noun, L"AE 1 Z", MS_Conj, NULL, MS_Unknown, 0, 3 },
  353. //--- SingleOrPluralAbbreviation
  354. { L"bu", L"b uh 1 sh - ax l", MS_Noun, L"b uh 1 sh - ax l Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  355. //--- DegreeAbbreviation
  356. { L"c", L"s eh 1 l - s iy - ax s", MS_Noun, L"s iy 1", MS_Noun, NULL, MS_Unknown, 0, 6 },
  357. //--- SingleOrPluralAbbreviation
  358. { L"cal", L"k ae 1 l - ax - r iy", MS_Noun, L"k ae 1 l - ax - r iy Z", MS_Noun, L"K AE 1 L", MS_Noun, 0, 0 },
  359. //--- SingleOrPluralAbbreviation
  360. { L"cl", L"s eh 1 n - t ax - l iy 2 - t er", MS_Noun, L"s eh 1 n - t ax - l iy 2 - t er Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  361. //--- SingleOrPluralAbbreviation
  362. { L"cm", L"s eh 1 n - t ax - m iy 2 - t er", MS_Noun, L"s eh 1 n - t ax - m iy 2 - t er Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  363. //--- AllCapsAbbreviation
  364. { L"co", L"S IY 1 & OW 1", MS_Noun, L"k ah 1 m - p ax - n iy", MS_Noun, NULL, MS_Unknown, 0, 3 },
  365. //--- AllCapsAbbreviation
  366. { L"cu", L"S IY 1 & Y UW 1", MS_Noun, L"k y uw 1 - b ih k", MS_Noun, NULL, MS_Unknown, 0, 3 },
  367. //--- SingleOrPluralAbbreviation
  368. { L"db", L"d eh 1 s - ax - b ax l", MS_Noun, L"d eh 1 s - ax - b ax l Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  369. //--- SingleOrPluralAbbreviation
  370. { L"deg", L"d ih - g r iy 1", MS_Noun, L"d ih - g r iy 1 z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  371. //--- SingleOrPluralAbbreviation
  372. { L"doz", L"d ah 1 z - ax n", MS_Noun, L"d ah 1 z - ax n Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  373. //--- DoctorDriveAbbreviation
  374. { L"dr", L"d aa 1 k - t er", MS_Noun, L"D R AY 1 V", MS_Noun, NULL, MS_Unknown, 0, 1 },
  375. //--- DegreeAbbreviation
  376. { L"f", L"f ae 1 r - ax n - h ay 2 t", MS_Noun, L"eh 1 f", MS_Noun, NULL, MS_Unknown, 0, 6 },
  377. //--- AbbreviationFollowedByDigit
  378. { L"fig", L"f ih 1 g - y er", MS_Noun, L"F IH 1 G", MS_Noun, NULL, MS_Unknown, 0, 2 },
  379. //--- AbbreviationFollowedByDigit
  380. { L"figs", L"f ih 1 g - y er Z", MS_Noun, L"F IH 1 G Z", MS_Noun, NULL, MS_Unknown, 0, 2 },
  381. //--- SingleOrPluralAbbreviation
  382. { L"ft", L"F UH 1 T", MS_Noun, L"F IY 1 T", MS_Noun, L"F AO 1 R T", MS_Noun, 0, 0 },
  383. //--- SingleOrPluralAbbreviation
  384. { L"g", L"G R AE 1 M", MS_Noun, L"G R AE 1 M Z", MS_Noun, L"JH IY 1", MS_Noun, 0, 0 },
  385. //--- SingleOrPluralAbbreviation
  386. { L"gal", L"g ae 1 l - ax n", MS_Noun, L"g ae 1 l - ax n Z", MS_Noun, L"G AE 1 L", MS_Noun, 0, 0 },
  387. //--- DoctorDriveAbbreviation
  388. { L"gov", L"g ah 1 v - er - n ER", MS_Noun, L"G AH 1 V", MS_Noun, NULL, MS_Unknown, 0, 1 },
  389. //--- AllCapsAbbreviation
  390. { L"hi", L"EY 1 CH & AY 1", MS_Noun, L"H AY 1", MS_Interjection, NULL, MS_Unknown, 0, 3 },
  391. //--- HR hour hours
  392. { L"hr", L"AW 1 ER", MS_Noun, L"AW 1 ER Z", MS_Noun, L"EY 1 CH AA 1 R", MS_Noun, 0, 5 },
  393. //--- SingleOrPluralAbbreviation
  394. { L"in", L"IH 1 N CH", MS_Noun, L"IH 1 N CH AX Z", MS_Noun, L"IH 1 N", MS_Prep, 0, 0 },
  395. //--- SingleOrPluralAbbreviation
  396. { L"j", L"JH UW 1 L", MS_Noun, L"JH UW 1 L Z", MS_Noun, L"JH EY 1", MS_Noun, 0, 0 },
  397. //--- DegreeAbbreviation
  398. { L"k", L"k eh 1 l - v ax n", MS_Noun, L"k ey 1", MS_Noun, NULL, MS_Unknown, 0, 6 },
  399. //--- SingleOrPluralAbbreviation
  400. { L"kb", L"k ih 1 l - ax - b ay 2 t", MS_Noun, L"k ih 1 l - ax - b ay 2 t S", MS_Noun, NULL, MS_Unknown, 0, 0 },
  401. //--- SingleOrPluralAbbreviation
  402. { L"kcal", L"k ih 1 l - ax - k ae 2 l - ax - r iy", MS_Noun, L"k ih 1 l - ax - k ae 2 l - ax - r iy Z", MS_Noun, L"K AE 1 L", MS_Noun, 0, 0 },
  403. //--- SingleOrPluralAbbreviation
  404. { L"kg", L"k ih 1 l - ax - g r ae 2 m", MS_Noun, L"k ih 1 l - ax - g r ae 2 m Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  405. //--- SingleOrPluralAbbreviation
  406. { L"kj", L"k ih 1 l - ax - jh uw 2 l", MS_Noun, L"k ih 1 l - ax - jh uw 2 l z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  407. //--- SingleOrPluralAbbreviation
  408. { L"km", L"k ih - l aa 1 m - ih - t er", MS_Noun, L"k ih - l aa 1 m - ih - t er Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  409. //--- SingleOrPluralAbbreviation
  410. { L"kw", L"k ih 1 l - ax - w aa 2 t", MS_Noun, L"k ih 1 l - ax - w aa 2 t S", MS_Noun, NULL, MS_Unknown, 0, 0 },
  411. //--- SingleOrPluralAbbreviation
  412. { L"l", L"l iy 1 - t er", MS_Noun, L"l iy 1 - t er Z", MS_Noun, L"EH 1 L", MS_Noun, 0, 0 },
  413. //--- AllCapsAbbreviation
  414. { L"la", L"EH 1 L & EY 1", MS_Noun, L"L AH 1", MS_Noun, NULL, MS_Unknown, 0, 3 },
  415. //--- SingleOrPluralAbbreviation
  416. { L"lb", L"P AW 1 N D", MS_Noun, L"P AW 1 N D Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  417. //--- SingleOrPluralAbbreviation
  418. { L"m", L"M IY 1 - T ER", MS_Noun, L"M IY 1 - T ER Z", MS_Noun, L"EH 1 M", MS_Noun, 0, 0 },
  419. //--- AllCapsAbbreviation
  420. { L"ma", L"EH 1 M & AA 1", MS_Noun, L"M AA 1", MS_Noun, NULL, MS_Unknown, 0, 3 },
  421. //--- March Mar
  422. { L"mar", L"M AA 1 R CH", MS_Noun, L"M AA 1 R", MS_Verb, NULL, MS_Unknown, 0, 4 },
  423. //--- SingleOrPluralAbbreviation
  424. { L"mb", L"m eh 1 g - ax - b ay 2 t", MS_Noun, L"m eh 1 g - ax - b ay 2 t S", MS_Noun, NULL, MS_Unknown, 0, 0 },
  425. //--- AllCapsAbbreviation
  426. { L"me", L"EH 1 M & IY 1", MS_Noun, L"M IY 1", MS_ObjPron, NULL, MS_Unknown, 0, 3 },
  427. //--- SingleOrPluralAbbreviation
  428. { L"mg", L"m ih 1 l - ax - g r ae 2 m", MS_Noun, L"m ih 1 l - ax - g r ae 2 m Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  429. //--- SingleOrPluralAbbreviation
  430. { L"mi", L"M AY 1 L", MS_Noun, L"M AY 1 L Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  431. //--- SingleOrPluralAbbreviation
  432. { L"min", L"m ih 1 n - ax t", MS_Noun, L"m ih 1 n - ax t S", MS_Noun, L"m ih 1 n - ax - m ax m", MS_Noun, 0, 0 },
  433. //--- SingleOrPluralAbbreviation
  434. { L"ml", L"m ih 1 l - ax - l iy 2 - t er", MS_Noun, L"m ih 1 l - ax - l iy 2 - t er Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  435. //--- SingleOrPluralAbbreviation
  436. { L"mm", L"m ih 1 l - ax - m iy 2 - t er", MS_Noun, L"m ih 1 l - ax - m iy 2 - t er Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  437. //--- SingleOrPluralAbbreviation
  438. { L"mph", L"m ay 1 l & p er 1 & aw 1 er", MS_Noun, L"m ay 1 l z & p er 1 & aw 1 er", MS_Noun, L"eh 1 m & p iy 1 & ey 1 ch", MS_Noun, 0, 0 },
  439. //--- SingleOrPluralAbbreviation
  440. { L"msec", L"m ih 2 l - ax - s eh 1 k - ax n d", MS_Noun, L"m ih 2 l - ax - s eh 1 k - ax n d Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  441. //--- AllCapsAbbreviation
  442. { L"mt", L"EH 1 M & T IY 1", MS_Noun, L"M AW 1 N T", MS_Noun, NULL, MS_Unknown, 0, 3 },
  443. //--- AllCapsAbbreviation
  444. { L"oh", L"OW 1 & EY 1 CH", MS_Noun, L"OW 1", MS_Interjection, NULL, MS_Unknown, 0, 3 },
  445. //--- AllCapsAbbreviation
  446. { L"or", L"OW 1 & AA 1 R", MS_Noun, L"AO 1 R", MS_CConj, NULL, MS_Unknown, 0, 3 },
  447. //--- SingleOrPluralAbbreviation
  448. { L"oz", L"AW 1 N S", MS_Noun, L"AW 1 N S AX Z", MS_Noun, L"AA 1 Z", MS_Noun, 0, 0 },
  449. //--- AbbreviationFollowedByDigit
  450. { L"p", L"P EY 1 JH", MS_Noun, L"P IY 1", MS_Noun, NULL, MS_Unknown, 0, 2 },
  451. //--- AllCapsAbbreviation
  452. { L"pa", L"P IY 1 & EY 1", MS_Noun, L"P AA 1", MS_Noun, NULL, MS_Unknown, 0, 3 },
  453. //--- AllCapsAbbreviation
  454. { L"pg", L"P IY 1 & JH IY 1", MS_Noun, L"P EY 1 JH", MS_Noun, NULL, MS_Unknown, 0, 3 },
  455. //--- SingleOrPluralAbbreviation
  456. { L"qt", L"K W AO 1 R T", MS_Noun, L"K W AO 1 R T Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  457. //--- Saturday Sat
  458. { L"sat", L"s ae 1 t - er - d ey", MS_Noun, L"S AE 1 T", MS_Verb, NULL, MS_Unknown, 0, 4 },
  459. //--- SEC Second Seconds
  460. { L"sec", L"s eh 1 k - ax n d", MS_Noun, L"s eh 1 k - ax n d Z", MS_Noun, L"EH 1 S & IY 1 & S IY 1", MS_Noun, 0, 5 },
  461. //--- AllCapsAbbreviation
  462. { L"sr", L"EH 1 S & AA 1 R", MS_Noun, L"s iy 1 n - y er", MS_Noun, NULL, MS_Unknown, 0, 3 },
  463. //--- DoctorDriveAbbreviation
  464. { L"st", L"S EY 1 N T", MS_Noun, L"S T R IY 1 T", MS_Noun, NULL, MS_Unknown, 0, 1 },
  465. //--- Sunday Sun
  466. { L"sun", L"s ah 1 n - d ey", MS_Noun, L"S AH 1 N", MS_Noun, NULL, MS_Unknown, 0, 4 },
  467. //--- SingleOrPluralAbbreviation
  468. { L"tbsp", L"t ey 1 - b ax l - s p uw 2 n", MS_Noun, L"t ey 1 - b ax l - s p uw 2 n Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  469. //--- SingleOrPluralAbbreviation
  470. { L"tsp", L"t iy 1 - s p uw 2 n", MS_Noun, L"t iy 1 - s p uw 2 n Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  471. //--- AllCapsAbbreviation
  472. { L"us", L"Y UW 1 & EH 1 S", MS_Noun, L"AH 1 S", MS_ObjPron, NULL, MS_Unknown, 0, 3 },
  473. //--- Wednesday Wed
  474. { L"wed", L"w eh 1 n z - d ey", MS_Noun, L"W EH 1 D", MS_Verb, NULL, MS_Unknown, 0, 4 },
  475. //--- AllCapsAbbreviation
  476. { L"wy", L"d ah 1 b - ax l - y uw & W AY 1", MS_Noun, L"W EY 1", MS_Noun, NULL, MS_Unknown, 0, 3 },
  477. //--- SingleOrPluralAbbreviation
  478. { L"yd", L"Y AA 1 R D", MS_Noun, L"Y AA 1 R D Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  479. };
  480. //--- IMPORTANT!!! This list must remain alphabetized for the binary search algorithm to work properly!!!
  481. AbbrevRecord g_PostLexLookupWordTable[] =
  482. {
  483. //--- MeasurementDisambig
  484. { L"bu", L"b uh 1 sh - ax l", MS_Noun, L"b uh 1 sh - ax l Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  485. //--- MeasurementDisambig
  486. { L"cal", L"k ae 1 l - ax - r iy", MS_Noun, L"k ae 1 l - ax - r iy Z", MS_Noun, L"K AE 1 L", MS_Noun, 0, 0 },
  487. //--- MeasurementDisambig
  488. { L"cl", L"s eh 1 n - t ax - l iy 2 - t er", MS_Noun, L"s eh 1 n - t ax - l iy 2 - t er Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  489. //--- MeasurementDisambig
  490. { L"cm", L"s eh 1 n - t ax - m iy 2 - t er", MS_Noun, L"s eh 1 n - t ax - m iy 2 - t er Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  491. //--- MeasurementDisambig
  492. { L"db", L"d eh 1 s - ax - b ax l", MS_Noun, L"d eh 1 s - ax - b ax l Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  493. //--- MeasurementDisambig
  494. { L"deg", L"d ih - g r iy 1", MS_Noun, L"d ih - g r iy 1 z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  495. //--- MeasurementDisambig
  496. { L"doz", L"d ah 1 z - ax n", MS_Noun, L"d ah 1 z - ax n Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  497. //--- MeasurementDisambig
  498. { L"ft", L"F UH 1 T", MS_Noun, L"F IY 1 T", MS_Noun, L"F AO 1 R T", MS_Noun, 0, 0 },
  499. //--- MeasurementDisambig
  500. { L"g", L"G R AE 1 M", MS_Noun, L"G R AE 1 M Z", MS_Noun, L"JH IY 1", MS_Noun, 0, 0 },
  501. //--- MeasurementDisambig
  502. { L"gal", L"g ae 1 l - ax n", MS_Noun, L"g ae 1 l - ax n Z", MS_Noun, L"G AE 1 L", MS_Noun, 0, 0 },
  503. //--- MeasurementDisambig
  504. { L"hr", L"AW 1 ER", MS_Noun, L"AW 1 ER Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  505. //--- MeasurementDisambig
  506. { L"in", L"IH 1 N CH", MS_Noun, L"IH 1 N CH AX Z", MS_Noun, L"IH 1 N", MS_Prep, 0, 0 },
  507. //--- MeasurementDisambig
  508. { L"j", L"JH UW 1 L", MS_Noun, L"JH UW 1 L Z", MS_Noun, L"JH EY 1", MS_Noun, 0, 0 },
  509. //--- MeasurementDisambig
  510. { L"kb", L"k ih 1 l - ax - b ay 2 t", MS_Noun, L"k ih 1 l - ax - b ay 2 t S", MS_Noun, NULL, MS_Unknown, 0, 0 },
  511. //--- MeasurementDisambig
  512. { L"kcal", L"k ih 1 l - ax - k ae 2 l - ax - r iy", MS_Noun, L"k ih 1 l - ax - k ae 2 l - ax - r iy Z", MS_Noun, L"K AE 1 L", MS_Noun, 0, 0 },
  513. //--- MeasurementDisambig
  514. { L"kg", L"k ih 1 l - ax - g r ae 2 m", MS_Noun, L"k ih 1 l - ax - g r ae 2 m Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  515. //--- MeasurementDisambig
  516. { L"kj", L"k ih 1 l - ax - jh uw 2 l", MS_Noun, L"k ih 1 l - ax - jh uw 2 l z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  517. //--- MeasurementDisambig
  518. { L"km", L"k ih - l aa 1 m - ih - t er", MS_Noun, L"k ih - l aa 1 m - ih - t er Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  519. //--- MeasurementDisambig
  520. { L"kw", L"k ih 1 l - ax - w aa 2 t", MS_Noun, L"k ih 1 l - ax - w aa 2 t S", MS_Noun, NULL, MS_Unknown, 0, 0 },
  521. //--- MeasurementDisambig
  522. { L"l", L"l iy 1 - t er", MS_Noun, L"l iy 1 - t er Z", MS_Noun, L"EH 1 L", MS_Noun, 0, 0 },
  523. //--- MeasurementDisambig
  524. { L"lb", L"P AW 1 N D", MS_Noun, L"P AW 1 N D Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  525. //--- MeasurementDisambig
  526. { L"m", L"M IY 1 - T ER", MS_Noun, L"M IY 1 - T ER Z", MS_Noun, L"EH 1 M", MS_Noun, 0, 0 },
  527. //--- MeasurementDisambig
  528. { L"mb", L"m eh 1 g - ax - b ay 2 t", MS_Noun, L"m eh 1 g - ax - b ay 2 t S", MS_Noun, NULL, MS_Unknown, 0, 0 },
  529. //--- MeasurementDisambig
  530. { L"mg", L"m ih 1 l - ax - g r ae 2 m", MS_Noun, L"m ih 1 l - ax - g r ae 2 m Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  531. //--- MeasurementDisambig
  532. { L"mi", L"M AY 1 L", MS_Noun, L"M AY 1 L Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  533. //--- MeasurementDisambig
  534. { L"min", L"m ih 1 n - ax t", MS_Noun, L"m ih 1 n - ax t S", MS_Noun, L"m ih 1 n - ax - m ax m", MS_Noun, 0, 0 },
  535. //--- ReadDisambig
  536. { L"misread", L"m ih s - r iy 1 d", MS_Verb, L"m ih s - r eh 1 d", MS_Verb, NULL, MS_Unknown, 0, 2 },
  537. //--- MeasurementDisambig
  538. { L"ml", L"m ih 1 l - ax - l iy 2 - t er", MS_Noun, L"m ih 1 l - ax - l iy 2 - t er Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  539. //--- MeasurementDisambig
  540. { L"mm", L"m ih 1 l - ax - m iy 2 - t er", MS_Noun, L"m ih 1 l - ax - m iy 2 - t er Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  541. //--- MeasurementDisambig
  542. { L"msec", L"m ih 2 l - ax - s eh 1 k - ax n d", MS_Noun, L"m ih 2 l - ax - s eh 1 k - ax n d Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  543. //--- MeasurementDisambig
  544. { L"oz", L"AW 1 N S", MS_Noun, L"AW 1 N S AX Z", MS_Noun, L"AA 1 Z", MS_Noun, 0, 0 },
  545. //--- ReadDisambig
  546. { L"proofread", L"p r uw 1 f - r iy 2 d", MS_Verb, L"p r uw 1 f - r eh 2 d", MS_Verb, NULL, MS_Unknown, 0, 2 },
  547. //--- MeasurementDisambig
  548. { L"qt", L"K W AO 1 R T", MS_Noun, L"K W AO 1 R T Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  549. //--- ReadDisambig
  550. { L"read", L"R IY 1 D", MS_Verb, L"R EH 1 D", MS_Verb, NULL, MS_Unknown, 0, 2 },
  551. //--- MeasurementDisambig
  552. { L"sec", L"S EH 1 k - ax n d", MS_Noun, L"s eh 1 k - ax n d z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  553. //--- MeasurementDisambig
  554. { L"tbsp", L"t ey 1 - b ax l - s p uw 2 n", MS_Noun, L"t ey 1 - b ax l - s p uw 2 n Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  555. //--- TheDisambig
  556. { L"the", L"DH IY 2", MS_Det, L"DH AX 2", MS_Det, NULL, MS_Unknown, 0, 1 },
  557. //--- MeasurementDisambig
  558. { L"tsp", L"t iy 1 - s p uw 2 n", MS_Noun, L"t iy 1 - s p uw 2 n Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  559. //--- MeasurementDisambig
  560. { L"yd", L"Y AA 1 R D", MS_Noun, L"Y AA 1 R D Z", MS_Noun, NULL, MS_Unknown, 0, 0 },
  561. };
  562. WCHAR *g_pOfA = L"ah 2 v & ax 2 &";
  563. WCHAR *g_pOfAn = L"ah 2 v & ax 2 n &";
  564. const SentBreakDisambigFunc g_SentBreakDisambigTable[] =
  565. {
  566. CStdSentEnum::IsAbbreviationEOS,
  567. CStdSentEnum::IfEOSNotAbbreviation,
  568. CStdSentEnum::IfEOSAndLowercaseNotAbbreviation,
  569. };
  570. const PronDisambigFunc g_PronDisambigTable[] =
  571. {
  572. CStdSentEnum::SingleOrPluralAbbreviation,
  573. CStdSentEnum::DoctorDriveAbbreviation,
  574. CStdSentEnum::AbbreviationFollowedByDigit,
  575. };
  576. const PronDisambigFunc g_AmbiguousWordDisambigTable[] =
  577. {
  578. CStdSentEnum::SingleOrPluralAbbreviation,
  579. CStdSentEnum::DoctorDriveAbbreviation,
  580. CStdSentEnum::AbbreviationFollowedByDigit,
  581. CStdSentEnum::AllCapsAbbreviation,
  582. CStdSentEnum::CapitalizedAbbreviation,
  583. CStdSentEnum::SECAbbreviation,
  584. CStdSentEnum::DegreeAbbreviation,
  585. };
  586. const PostLexLookupDisambigFunc g_PostLexLookupDisambigTable[] =
  587. {
  588. CStdSentEnum::MeasurementDisambig,
  589. CStdSentEnum::TheDisambig,
  590. CStdSentEnum::ReadDisambig,
  591. };
  592. //--- IMPORTANT!!! This list must remain alphabetized for the binary search algorithm to work properly!!!
  593. // This is an alphabetized list of all non-proper-noun words which
  594. // appear within the list of the 200 most frequent first words
  595. // in sentences in both the Brown and WSJ corpora.
  596. const SPLSTR g_FirstWords[] =
  597. {
  598. DEF_SPLSTR( "A" ),
  599. DEF_SPLSTR( "About" ),
  600. DEF_SPLSTR( "According" ),
  601. DEF_SPLSTR( "After" ),
  602. DEF_SPLSTR( "Again" ), // PaulCa added 4/14/99 (Bug 107)
  603. DEF_SPLSTR( "All" ),
  604. DEF_SPLSTR( "Also" ),
  605. DEF_SPLSTR( "Although" ),
  606. DEF_SPLSTR( "Among" ),
  607. DEF_SPLSTR( "An" ),
  608. DEF_SPLSTR( "And" ),
  609. DEF_SPLSTR( "Another" ),
  610. DEF_SPLSTR( "Any" ), // PaulCa added 4/14/99 (Bug 107)
  611. DEF_SPLSTR( "Anyway" ), // PaulCa added 4/14/99 (Bug 107)
  612. DEF_SPLSTR( "Are" ), // PaulCa added 4/14/99 (Bug 107)
  613. DEF_SPLSTR( "As" ),
  614. DEF_SPLSTR( "At" ),
  615. DEF_SPLSTR( "Back" ), // PaulCa added 4/14/99 (Bug 107)
  616. DEF_SPLSTR( "Because" ),
  617. DEF_SPLSTR( "Before" ),
  618. DEF_SPLSTR( "Besides" ), // PaulCa added 4/14/99 (Bug 107)
  619. DEF_SPLSTR( "Both" ),
  620. DEF_SPLSTR( "But" ),
  621. DEF_SPLSTR( "By" ),
  622. DEF_SPLSTR( "Can" ), // PaulCa added 4/14/99 (Bug 107)
  623. DEF_SPLSTR( "Consequently" ), // PaulCa added 4/14/99 (Bug 107)
  624. DEF_SPLSTR( "Dear" ), // PaulCa added 4/14/99 (Bug 107)
  625. DEF_SPLSTR( "Despite" ),
  626. DEF_SPLSTR( "Did" ), // PaulCa added 4/14/99 (Bug 107)
  627. DEF_SPLSTR( "Do" ), // PaulCa added 4/14/99 (Bug 107)
  628. DEF_SPLSTR( "Does" ), // PaulCa added 4/14/99 (Bug 107)
  629. DEF_SPLSTR( "Don't" ), // PaulCa added 4/14/99 (Bug 107)
  630. DEF_SPLSTR( "During" ),
  631. DEF_SPLSTR( "Each" ),
  632. DEF_SPLSTR( "Early" ), // PaulCa added 4/14/99 (Bug 107)
  633. DEF_SPLSTR( "Even" ),
  634. DEF_SPLSTR( "Every" ), // PaulCa added 4/14/99 (Bug 107)
  635. DEF_SPLSTR( "Finally" ), // PaulCa added 4/14/99 (Bug 107)
  636. DEF_SPLSTR( "First" ),
  637. DEF_SPLSTR( "Following" ), // PaulCa added 4/14/99 (Bug 107)
  638. DEF_SPLSTR( "For" ),
  639. DEF_SPLSTR( "Four" ),
  640. DEF_SPLSTR( "From" ),
  641. DEF_SPLSTR( "Further" ), // PaulCa added 4/14/99 (Bug 107)
  642. DEF_SPLSTR( "Furthermore" ), // PaulCa added 4/14/99 (Bug 107)
  643. DEF_SPLSTR( "Generally" ), // PaulCa added 4/14/99 (Bug 107)
  644. DEF_SPLSTR( "Given" ), // PaulCa added 4/14/99 (Bug 107)
  645. DEF_SPLSTR( "Go" ), // PaulCa added 4/14/99 (Bug 107)
  646. DEF_SPLSTR( "Great" ), // PaulCa added 4/14/99 (Bug 107)
  647. DEF_SPLSTR( "Had" ), // PaulCa added 4/14/99 (Bug 107)
  648. DEF_SPLSTR( "Have" ), // PaulCa added 4/14/99 (Bug 107)
  649. DEF_SPLSTR( "Having" ), // PaulCa added 4/14/99 (Bug 107)
  650. DEF_SPLSTR( "He" ),
  651. DEF_SPLSTR( "Her" ),
  652. DEF_SPLSTR( "Here" ),
  653. DEF_SPLSTR( "His" ),
  654. DEF_SPLSTR( "How" ), // PaulCa added 4/14/99 (Bug 107)
  655. DEF_SPLSTR( "However" ),
  656. DEF_SPLSTR( "I" ),
  657. DEF_SPLSTR( "If" ),
  658. DEF_SPLSTR( "In" ),
  659. DEF_SPLSTR( "Indeed" ),
  660. DEF_SPLSTR( "Initially" ), // PaulCa added 4/14/99 (Bug 107)
  661. DEF_SPLSTR( "Instead" ),
  662. DEF_SPLSTR( "Is" ), // PaulCa added 4/14/99 (Bug 107)
  663. DEF_SPLSTR( "It" ),
  664. DEF_SPLSTR( "Its" ),
  665. DEF_SPLSTR( "Just" ), // PaulCa added 4/14/99 (Bug 107)
  666. DEF_SPLSTR( "Last" ),
  667. DEF_SPLSTR( "Later" ),
  668. DEF_SPLSTR( "Let" ), // PaulCa added 4/14/99 (Bug 107)
  669. DEF_SPLSTR( "Like" ),
  670. DEF_SPLSTR( "Many" ),
  671. DEF_SPLSTR( "Maybe" ), // PaulCa added 4/14/99 (Bug 107)
  672. DEF_SPLSTR( "Meanwhile" ),
  673. DEF_SPLSTR( "More" ),
  674. DEF_SPLSTR( "Moreover" ),
  675. DEF_SPLSTR( "Most" ),
  676. DEF_SPLSTR( "Much" ),
  677. DEF_SPLSTR( "My" ), // Added to fix bug #385
  678. DEF_SPLSTR( "Neither" ),
  679. DEF_SPLSTR( "Never" ), // PaulCa added 4/14/99 (Bug 107)
  680. DEF_SPLSTR( "Nevertheless" ),
  681. DEF_SPLSTR( "New" ),
  682. DEF_SPLSTR( "Next" ), // PaulCa added 4/14/99 (Bug 107)
  683. DEF_SPLSTR( "No" ),
  684. DEF_SPLSTR( "None" ), // PaulCa added 4/14/99 (Bug 107)
  685. DEF_SPLSTR( "Nonetheless" ), // PaulCa added 4/14/99 (Bug 107)
  686. DEF_SPLSTR( "Nor" ),
  687. DEF_SPLSTR( "Not" ),
  688. DEF_SPLSTR( "Nothing" ), // PaulCa added 4/14/99 (Bug 107)
  689. DEF_SPLSTR( "Now" ),
  690. DEF_SPLSTR( "Of" ),
  691. DEF_SPLSTR( "On" ),
  692. DEF_SPLSTR( "Once" ),
  693. DEF_SPLSTR( "One" ),
  694. DEF_SPLSTR( "Only" ),
  695. DEF_SPLSTR( "Or" ),
  696. DEF_SPLSTR( "Other" ),
  697. DEF_SPLSTR( "Others" ),
  698. DEF_SPLSTR( "Our" ), // PaulCa added 4/14/99 (Bug 107)
  699. DEF_SPLSTR( "Over" ),
  700. DEF_SPLSTR( "People" ),
  701. DEF_SPLSTR( "Perhaps" ), // PaulCa added 4/14/99 (Bug 107)
  702. DEF_SPLSTR( "Please" ), // PaulCa added 4/14/99 (Bug 107)
  703. DEF_SPLSTR( "Previous" ), // PaulCa added 4/14/99 (Bug 107)
  704. DEF_SPLSTR( "Recent" ), // PaulCa added 4/14/99 (Bug 107)
  705. DEF_SPLSTR( "Right" ), // PaulCa added 4/14/99 (Bug 107)
  706. DEF_SPLSTR( "Second" ), // PaulCa added 4/14/99 (Bug 107)
  707. DEF_SPLSTR( "See" ), // PaulCa added 4/14/99 (Bug 107)
  708. DEF_SPLSTR( "Several" ),
  709. DEF_SPLSTR( "She" ),
  710. DEF_SPLSTR( "Shortly" ), // PaulCa added 4/14/99 (Bug 107)
  711. DEF_SPLSTR( "Similarly" ), // PaulCa added 4/14/99 (Bug 107)
  712. DEF_SPLSTR( "Since" ),
  713. DEF_SPLSTR( "So" ),
  714. DEF_SPLSTR( "Some" ),
  715. DEF_SPLSTR( "Sometimes" ), // PaulCa added 4/14/99 (Bug 107)
  716. DEF_SPLSTR( "Soon" ), // PaulCa added 4/14/99 (Bug 107)
  717. DEF_SPLSTR( "Still" ),
  718. DEF_SPLSTR( "Subsequently" ), // PaulCa added 4/14/99 (Bug 107)
  719. DEF_SPLSTR( "Such" ),
  720. DEF_SPLSTR( "Take" ), // PaulCa added 4/14/99 (Bug 107)
  721. DEF_SPLSTR( "That" ),
  722. DEF_SPLSTR( "The" ),
  723. DEF_SPLSTR( "Their" ),
  724. DEF_SPLSTR( "Then" ),
  725. DEF_SPLSTR( "There" ),
  726. DEF_SPLSTR( "Thereafter" ), // PaulCa added 4/14/99 (Bug 107)
  727. DEF_SPLSTR( "Therefore" ), // PaulCa added 4/14/99 (Bug 107)
  728. DEF_SPLSTR( "These" ),
  729. DEF_SPLSTR( "They" ),
  730. DEF_SPLSTR( "This" ),
  731. DEF_SPLSTR( "Those" ),
  732. DEF_SPLSTR( "Though" ),
  733. DEF_SPLSTR( "Three" ), // PaulCa added 4/14/99 (Bug 107)
  734. DEF_SPLSTR( "Through" ), // PaulCa added 4/14/99 (Bug 107)
  735. DEF_SPLSTR( "Thus" ),
  736. DEF_SPLSTR( "To" ),
  737. DEF_SPLSTR( "Today" ),
  738. DEF_SPLSTR( "Two" ),
  739. DEF_SPLSTR( "Under" ),
  740. DEF_SPLSTR( "Unlike" ), // PaulCa added 4/14/99 (Bug 107)
  741. DEF_SPLSTR( "Until" ),
  742. DEF_SPLSTR( "Upon" ), // PaulCa added 4/14/99 (Bug 107)
  743. DEF_SPLSTR( "We" ),
  744. DEF_SPLSTR( "Well" ), // PaulCa added 4/14/99 (Bug 107)
  745. DEF_SPLSTR( "What" ),
  746. DEF_SPLSTR( "When" ),
  747. DEF_SPLSTR( "Where" ), // PaulCa added 4/14/99 (Bug 107)
  748. DEF_SPLSTR( "Whether" ), // PaulCa added 4/14/99 (Bug 107)
  749. DEF_SPLSTR( "Which" ), // PaulCa added 4/14/99 (Bug 107)
  750. DEF_SPLSTR( "While" ),
  751. DEF_SPLSTR( "Who" ), // PaulCa added 4/14/99 (Bug 107)
  752. DEF_SPLSTR( "Why" ), // PaulCa added 4/14/99 (Bug 107)
  753. DEF_SPLSTR( "Will" ), // PaulCa added 4/14/99 (Bug 107)
  754. DEF_SPLSTR( "With" ),
  755. DEF_SPLSTR( "Within" ),
  756. DEF_SPLSTR( "Without" ), // PaulCa added 4/14/99 (Bug 107)
  757. DEF_SPLSTR( "Yes" ), // PaulCa added 4/14/99 (Bug 107)
  758. DEF_SPLSTR( "Yet" ),
  759. DEF_SPLSTR( "You" ),
  760. DEF_SPLSTR( "Your" ),
  761. };