1 Automatically Generated Inflection Database (AGID)
5 Copyright 2000-2016 by Kevin Atkinson <kevina@gnu.org>
7 The file "infl.txt" is an automatically created database of the
8 inflected forms of words from a rather large word list.
10 The latest version can be found at http://wordlist.aspell.net
12 Entries are in the following form.
14 <word><sp><pos>[?]:<sp><inflected forms>
15 <word> := [[A-Za-z']]+
16 <sp> := <literal space>
18 <inflected forms> := <inflected form><sp>|<sp>...<sp>|<sp><inflected form>
19 <inflected form> := <individual entry>,<sp>...,<sp><individual entry>
20 <individual entry> := <word><word tags>[<sp><variant level>][<sp>{<explanation>}]
21 <word tags> := [~][<][!][?]
22 <explanation> := [<explanation text>][:<distinguishing number>]
23 <explanation text> := [[A-Za-z'_/]]+
25 where stuff between [ ] is optional, stuff between [[ ]] indicate a
26 range of possible characters for that entry. If a [[ ]] is followed by
27 a + it means the entry can consist of one or more characters in
28 that range. { } are literal.
30 A typical entry will look like
32 WORD V: WORDed, WORed 2, WORD {EXPL} | WORDing, WORing 2 | WORDs
34 <pos> is V for verb, N for noun, or A or adjective or adverb.
35 If <pos> is followed by a ? that means that the part-of-speech was not
36 in the part-of-speech database however the inflected forms of the word
37 where found in the word list.
39 The inflected forms are in the following order for verbs (except for
41 <past tense> [<past participle>] <-ing form> <-s form>
42 and for adjective or adverbs:
43 <-er form> <-est form>
44 Each form is seperated by a ' | '.
48 <past 1st & 3d singular> <2d singular, plural, past subjunctive>
49 <past participle> <present participle> <present 1st singular>
50 <2d singular> <3d singular> <plural present>
52 <past & past participle> <present participle> <present participle>
53 <present 1st & 3d singular> <2d singular> <plural present>
55 An absence of a variant level implies a variant level of 0. Two words
56 with the same whole number variant level are considered almost equal
57 with a slight preference given to the entry with a lower number. A
58 whole number variant level of 1 indicates a less preferred form of the
59 word. A whole number variant level of 2 indicates any number of
60 things. It could mean that it is from an archaic use of the word, or
61 a variant that is hardly ever used or for an extremely obscure meaning
62 of the word, or finally it could mean that the word looked like it
63 could possibly be a inflected form of the base word but I could not
64 find any evidence for them. If two words have the same variant level
65 and explanation it means that both inflections were found and the
66 script was not sure which one to use.
68 Sometimes the inflected form to use depends on the meaning of the
69 word. If this is the case the two entries will have different
70 explanations. If the distinction can be made in a few words it is
71 given with underscores (_) replacing spaces. Otherwise the two
72 entries will have different distinguishing numbers.
74 A < after a word means that there is a good change that this is an
75 inflected form of the word, a ~ after a word means that there is a
76 slight chance. A ! after a word indicates that the word is likely an
77 inflections of a similar word (generally one ending in e) and not the
78 current word. A ? after a word means that the word was not in the
79 word list but if it was it would be considered an inflected form of
82 This verson is now almost as accurate as Alan Beale's 2of12id file
83 distributed with the "Unofficial Alternate 12 Dicts Package" for the
84 base words which have an entry in 2of12id.txt with a few notable
85 exceptions. The most obvious one is the "person" entry. Alan Beale
86 considers, based on what his sources have told him, that "persons" is
87 the proper plural for "person" and "people" is considered a variant.
88 I however disagree and decided to consider "people" the primary form
89 and "persons" as the sligtly less perfered variant based on my own
90 experence and http://www.quinion.com/words/usagenotes/un-person.htm
93 The normal plural of person was persons ... However, there is
94 evidence from Chaucer onwards that some writers chose to use people
95 as a plural for person, not only in the generalised sense of 'an
96 uncountable or indistinct mass of individuals' but also in specific
97 countable cases. ... Though persons survives, it does so largely in
98 formal or legal contexts ...From the evidence, it seems that the
99 trend towards using people instead of persons is accelerating and
100 that it may not be so long before persons vanishes from the language
101 except in certain set phrases.
103 I considered making "persons" a variant (level 1), but I decided
104 against it as "persons" is for the most part perfectly acceptable and
105 probably considered the proper plural to use by some.
107 I also considered the -people ending the primary form for all words
108 ending in -person such as salesperson and the -persons entry the
109 slightly less preferred variant in spite of what 2of12id.txt said.
111 In some cases a variant of level 2 is listed in AGID where it is not
112 listed at all in 2of12id. In general this means that the script came
113 up with the possibility and, in spite it not being listed in 2of12id,
114 it seams logical to me.
116 The final case occurs when a word has two or more -s inflections used
117 as both noun and verb forms, and these forms would have different
118 variant levels in 2of12id. For example:
119 ditto N: dittos, dittoes 1
120 ditto V: dittoed | dittoing | dittos, dittoes 0.1
121 For purely technical reasons and because I do not feel that it matters
122 too much I have made the variant levels for the -s forms the same. For
123 example the ditto entries became:
124 ditto N: dittos, dittoes 0.1
125 ditto V: dittoed | dittoing | dittos, dittoes 0.1
126 The choice of the variant levels I used is somewhat arbitrary but I in
127 general went with the lower level.
129 Fell free to send me corrections to correct any of these questionable
130 words. I am mostly interested in the preferred form of the word when
131 the script was not able to decide or words marked with < or ~ that are
132 valid inflected forms of the words.
134 Also included in this version are the files "variant_0.lst",
135 "variant_1.lst", "variant_2.lst", and "variant.tab". The files
136 "variant_#.lst" include all of the inflected forms at the given level
137 found in infl.txt which are not generally considered to be some other
138 common word. The file variant.tab contains a cross reference of all
139 alternate forms of inflected form of words. The file variant-wroot.tab
140 is like variant.tab except that it also included the root form of the
143 Words are in mixed case but all accents have been striped thus words
144 like café are instead cafe.
146 The file "variant" contains a list of alternate inflections.
148 The file "irregular" contains extra information where a noun or verb
149 has irregular inflected forms.
151 The file "dontuse" contains a list of words not to consider an
152 inflected form of a word if more than one inflected form of a word is
155 The files "prefixes" and "suffixes" contains a list of common prefixes
156 and suffixes respectfully. These files are used by the script to
157 produce inflected forms for words that end in a word in the
158 "irregular" file. If the beginning appears in the word list or the
159 prefixes file and the ending appears in the irregular file I also
160 consider <prefix>+<irregular inflections>. If the prefix is 3 letters
161 or more OR appears in the prefixes file and the suffix is 4 letters or
162 more OR appears in the suffixes file I consider it the most likely
163 choice, otherwise I consider it as a possible candidate but not the
166 The file "make-infl" is the actual Perl script used to create the
169 The file "find-var" is the Perl script used to create the variant
170 lists and cross reference file.
172 The file "make-all" was used to create the word list used by the script.
176 From Ver 2014.08.11 to 2016.01.19
178 Avoid hard coding the input files in the make-infl script.
180 No changes to the data files.
182 From Rev 4 to Ver 2014.08.11
184 Misc. changes to sync up with what is being used by SCOWL.
186 From Revision 3a to 4 (January 2, 2003)
188 Added variant-wroot.tab
189 Update find-var script to also produce variant-wroot.tab.
191 From Revision 3 to 3a (April 04, 2001)
193 Fixed a bug in the find-var script which caused some common
194 words which are variants for one usage of a word but not
195 variants for any other common usage to improperly appear in
198 From Revision 2 to 3 (January 28, 2001)
200 Changed the format of infl.txt to something which is slightly harder
201 to read but a lot less ambiguous and easier to parse.
203 Update various files, including the actual script, so that the
204 output that is almost as accurate of Alan Beale 2of12id.txt
206 Eliminated Moby Words and ABLE from the word list used by the script
207 to give more accurate results.
209 From Revision 1 to 2 (August 18, 2000)
211 Classified variants as either almost equal, also used, or
214 The / is now used to indicate equal variants. "/?" is now used to
215 mean what "/" used to be.
217 Lots of additional rules added which greatly improved the results.
219 COPYRIGHT AND SOURCE:
221 The final product is under the following copyright, as well as any
222 copyrights mentioned below.
224 Copyright 2000-2014 by Kevin Atkinson
226 Permission to use, copy, modify, distribute and sell this database,
227 the associated scripts, the output created form the scripts and its
228 documentation for any purpose is hereby granted without fee,
229 provided that the above copyright notice appears in all copies and
230 that both that copyright notice and this permission notice appear in
231 supporting documentation. Kevin Atkinson makes no representations
232 about the suitability of this array for any purpose. It is provided
233 "as is" without express or implied warranty.
235 The part-of-speech database is taken from Alan Beale 2of12id
236 and the WordNet database which is under the following copyright:
238 This software and database is being provided to you, the LICENSEE, by
239 Princeton University under the following license. By obtaining, using
240 and/or copying this software and database, you agree that you have
241 read, understood, and will comply with these terms and conditions.:
243 Permission to use, copy, modify and distribute this software and
244 database and its documentation for any purpose and without fee or
245 royalty is hereby granted, provided that you agree to comply with
246 the following copyright notice and statements, including the disclaimer,
247 and that the same appear on ALL copies of the software, database and
248 documentation, including modifications that you make for internal
249 use or for distribution.
251 WordNet 1.6 Copyright 1997 by Princeton University. All rights reserved.
253 THIS SOFTWARE AND DATABASE IS PROVIDED "AS IS" AND PRINCETON
254 UNIVERSITY MAKES NO REPRESENTATIONS OR WARRANTIES, EXPRESS OR
255 IMPLIED. BY WAY OF EXAMPLE, BUT NOT LIMITATION, PRINCETON
256 UNIVERSITY MAKES NO REPRESENTATIONS OR WARRANTIES OF MERCHANT-
257 ABILITY OR FITNESS FOR ANY PARTICULAR PURPOSE OR THAT THE USE
258 OF THE LICENSED SOFTWARE, DATABASE OR DOCUMENTATION WILL NOT
259 INFRINGE ANY THIRD PARTY PATENTS, COPYRIGHTS, TRADEMARKS OR
262 The name of Princeton University or Princeton may not be used in
263 advertising or publicity pertaining to distribution of the software
264 and/or database. Title to copyright in this software, database and
265 any associated documentation shall at all times remain with
266 Princeton University and LICENSEE agrees to preserve same.
268 Alan Beale 2of12id.txt is indirectly derived from the Moby part-of-speech
269 database and the WordNet database. The Moby part-of-speech is in the
272 The Moby lexicon project is complete and has
273 been place into the public domain. Use, sell,
274 rework, excerpt and use in any way on any platform.
276 Placing this material on internal or public servers is
277 also encouraged. The compiler is not aware of any
278 export restrictions so freely distribute world-wide.
280 You can verify the public domain status by contacting
284 Arcata, CA 95521-4884
290 The word list used is a combination of several word list:
292 1) The ENABLE2K word lists which is in the public domain:
294 The ENABLE master word list, WORD.LST, is herewith formally
295 released into the Public Domain. Anyone is free to use it or
296 distribute it in any manner they see fit. No fee or registration
297 is required for its use nor are "contributions" solicited (if you
298 feel you absolutely must contribute something for your own peace
299 of mind, the authors of the ENABLE list ask that you make a
300 donation on their behalf to your favorite charity). This word
301 list is our gift to the Scrabble community, as an alternate to
302 "official" word lists. Game designers may feel free to
303 incorporate the WORD.LST into their games. Please mention the
304 source and credit us as originators of the list. Note that if
305 you, as a game designer, use the WORD.LST in your product, you
306 may still copyright and protect your product, but you may *not*
307 legally copyright or in any way restrict redistribution of the
308 WORD.LST portion of your product. This *may* under law restrict
309 your rights to restrict your users' rights, but that is only
312 2) All of the word lists except ABLE.LST in the ENABLE2K Supplemnt
315 2DICTS.LST ALSO.LST LETTERS.LST OSPDADD.LST UCACR.LST
316 LCACR.LST NOPOS.LST PLURALS.LST UPPER.LST
318 All of these word lists are also in the public domain.
320 3) The list of signature words from the YAWL package which is in the
323 4) The UK Advanced Cryptics Dictionary which in under the following
326 Copyright (c) J Ross Beresford 1993-1999. All Rights Reserved.
328 The following restriction is placed on the use of this
329 publication: if The UK Advanced Cryptics Dictionary is used
330 in a software package or redistributed in any form, the
331 copyright notice must be prominently displayed and the text
332 of this document must be included verbatim.
334 There are no other restrictions: I would like to see the
335 list distributed as widely as possible.
337 5) Some extra words found in the Part-Of-Speech database that was not
338 found in any of the above word lists.
340 6) Words found in the Jargon File Word List package, available at
341 http://aspell.sourceforge.net/wl/, which is in the Public Domain.
343 7) Words in 2of12id.txt not in any of the word lists above. 2of12id is
344 indirectly derived from all the above sources and most of the word
345 lists from the Moby Words package:
347 10196pla.ces 113809of.fic 21986na.mes 256772co.mpo 354984si.ngl
348 3897male.nam 4160offi.cia 4946fema.len 6213acro.nym 74550com.mon
350 The Moby Word package, like the Part-Of-Speech database is in the
353 8) And finally some extra words that I added myself. These words can be
354 found in the file "extra-words"
356 The "dontuse", "irregular", and "variant" file was created by me
357 (Kevin Atkinson) from numerous sources.