Fetching contributors…
Cannot retrieve contributors at this time
2849 lines (2146 sloc) 109 KB
LanguageTool Change Log
*** See for all changes after LT 2.9: ***
*** ***
2.9 (2015-03-30)
-updated POS tag dictionary
-added new rules
-fixed false alarms
-Added a few rules and fixed a few false alarms
-Added many new style rules contributed by Heikki Lehvaslaiho. As these
may cause false alarms, they are not activated by default. You can
activate them by turning on all rules in the new 'Plain English' category.
-added a few new rules
-updated POS tag dictionary and Hunspell dictionary to Dicollecte-5.3
-added a few new rules and fixed false alarms
-Added a new rule that checks for subject verb agreement. For now, only cases
with 'ist', 'sind', 'war', and 'waren' are supported. Example for errors that
are detected: 'Der Hund sind schön.', 'Die Autos ist schnell.'
To make this rule work, phrases are now unified in disambiguation.xml: for
example, 'Mann' in the phrase 'ein Mann' will only retain its nominative
reading (SUB:NOM:SIN:MAS), whereas it used to have also accusative and
dative readings (SUB:AKK:SIN:MAS, SUB:DAT:SIN:MAS).
-improved a few rules
-added several new rules
-added/improved several rules
-3695 compound words (pre-reform) - the largest free database
-added and improved rules
-big dictionary update
-new grammar rules
-new simple replace rule for soft suggestions
-disambiguator improvements
-compound tagging and spelling improvements
-initials tagging improvements
-sentence and word tokenizing improvements
-improved handling of stres symbol and soft hyphen
-Bitext rules:
-added a simple rule for checking whether translations end with the same punctuation
mark as the original (this includes only .?! characters).
-it is now possible to add external bitext rule files on the command line, by using
--bitextrule <FILE> option. The file path has to be absolute. Note: this allows
using bitext rules also for languages that have no bitext rules included by
-The new files at <languageCode>/hunspell/spelling.txt can be used to add
accepted words to the spell checker that are also considered when creating
suggestions for misspelled words.
This is similar to the <languageCode>/hunspell/ignore.txt files, which list
accepted words which are *not* used when creating suggestions for
misspelled words.
-JLanguageTool.activateDefaultPatternRules() and
JLanguageTool.activateDefaultFalseFriendRules() have been removed - all
pattern rules and false friend rules (if a second language is specified)
are now activated automatically when the constructor of JLanguageTool is called.
Should you need a checker without the XML-based pattern rules, extend your
language class (e.g. 'English') with one that overwrites the getPatternRules()
method and returns an empty list there.
-ManualTagger.lookup() has been replaced by ManualTagger.tag() after being
deprecated since the latest release
-All static methods and fields from class 'Language' have been moved to the new
class 'Languages'. For now, the methods/fields in class Language still exist
but have been deprecated.
-LanguageIdentifierTools has been removed. Use LanguageIdentifier instead.
-Removed (Default)ResourceDataBroker.setResourceDir() and setRulesDir()
as these can be set with the constructor
-Cleaned up up class Contributor, e.g. removing getRemark()
-Category.setDefaultOff() has been removed, this can be set via constructor now
-Renamed classes: => =>
-Other small API cleanups that shouldn't affect the common use cases,
e.g. IncorrectExample.getCorrections() returns and unmodifiable list now,
removal of deprecated methods.
-Embedded server:
-XML escaping has been fixed, this could cause invalid XML documents
to be returned
-new config file option 'maxWorkQueueSize' that lets you set the maximum
size of the request queue - if it gets larger than this, requests will
be rejected (503 Service unavailable)
-The server now responds with more specific HTTP status codes to these
error conditions:
413 Request Entity Too Large - if text exceeds maximum text size
503 Service Unavailable - if check exceeds maximum check time
-The stand-alone GUI can now take a plain text file as an argument, this
file will then be loaded on startup (Github issue #232).
-It is now possible to add an external rule file when calling LanguageTool from the
command line. Use --rulefile <file> to add a file. If the file name has a format
that contains a language name, it will be used alongside other rules; otherwise,
it will replace the rules. You can also load an external file with false friends by
using the option --falsefriends <file>. The file name should be an absolute file
path, and false friend files are always added to the ones that are loaded for the
language. (Github issue #192)
-Rule syntax:
-A rule may now have a single example sentence as long as it has a 'correction'
attribute - this can save some redundancy if the only correct sentence is the
same as the incorrect sentence with the correction applied. Before, a rule
needed at least two example sentences.
-'example' element: type="incorrect" is now optional if there's a 'correction'
attribute. The 'correction' attribute implies that the sentence is incorrect.
-'example' element: type="correct" is now optional. No 'type' attribute and
no 'correction' attribute implies that the sentence is correct.
-We have switched from Apache Tika to language-detector
( for automatically
identifying the text language. It should be faster and results
should be more reliable.
Detection of Asturian and Galician had to be disabled because the
detection quality was too low and also affected detection of Spanish.
-Fixed a regression that made it impossible to load external rule files in the GUI.
2.8 (2014-12-30)
-removed dependency on Hunspell, now uses Morfologik for spell checking
-added and improved a few rules
-updated dictionary
-added and improved rules
-fixed false alarms
-added and improved many rules
-some new rules (thanks to Nick Hough)
-updated the tagger and synthesizer dictionaries, fixing issue #202
-new filter to be used for matching the part-of-speech of parts of words, e.g.:
<token regexp="yes">in.*</token>
<filter class="org.languagetool.rules.en.EnglishPartialPosTagFilter"
args="no:1 regexp:in(.*) postag_regexp:JJ"/>
This will only keep matches for words that start with 'in' and where the
part after the 'in' is an adjective (POS tag 'JJ'). The 'no:1' is the token
position, i.e. here the first (and only) matching <token> is referred to.
-added and improved a few rules
-added and improved a few rules
-added and improved several rules
-added and improved false friends with English
-added/improved several rules
-removed dependency on Hunspell, now uses Morfologik for spell checking
-Reformatted rules file
-Added more rules
-removed dependency on Hunspell, now uses Morfologik for spell checking
-the dash character ("-") is a delimiter now when tokenizing the text
-added and improved rules
-added a few false friend rules (Russian/English)
-many new rules (including agreement with nouns, time expressions etc)
-rule coverage improvement
-dictionary update (big improvements for proper nouns and vocative case)
-new tag and rule to warn about alternative spelling
-added word frequency information to improve spelling suggestions
-some new disambiguator rules
-Rule Syntax:
-<short>...</short> can now be added to a rulegroup to affect all the rules
of that group
-If you develop your own rules that are not part of LT you can now add external="yes"
to your categories to prevent the rule link to from
appearing in our stand-alone GUI (the link would not work for rules that are not part
of the main distribution of LT). (Github issue #223)
-If a rule group specifies default="off", the rules inside the rule group
may not also specify default="on"/"off".
-Removed classes and methods that had been deprecated since 2.7 or longer
-Embedded server:
-The config file options 'requestLimit' and 'requestLimitPeriodInSeconds' can
now also be used for the HTTP server (not just for the HTTPS server)
-New config file option 'trustXForwardForHeader': set this to 'true' if you
run the server behind a reverse proxy and want the request limit to
work on the original IP addresses provided by the 'X-forwarded-for' HTTP header,
usually set by the proxy. If you run behind a proxy but don't set this
property to true, one user can use all the requests so other users will
also get an error message because of the request limit.
-Fix response of After the Deadline mode: <description>...</description> was
sometimes empty, confusing the text check in WordPress
-Bitext rules were not disabled properly, even if they were specified with a
proper parameter for the server; now it's fixed
-Fixed problem with improper positions for some bitext rules (issue #218)
-A new 'errorColors' setting has been added to the languagetool.cfg configuration
file. It can be used to set the background color of errors. For example,
errorColors=typographical:#b8b8ff, style:#ffb8b8
will show 'typographical' errors with a blue background and 'style' errors
with a red background in the upper part of the LT window. 'typographical' and
'style' are the types that are set in grammar.xml as "type=...".
There's no user interface yet to configure these colors. Note that you should
only edit the languagetool.cfg file when LT is not running.
-Bugfix: rules inside a rule group had not been activated if a previous
rule from the same rulegroup used default="off"
-Words are not ignored anymore by the spell checker just because they occur in
a rule's suggestion. If you want the spell checker to ignore words globally, add
them to hunspell/ignore.txt. To ignore them depending on the context, add a
'ignore_spelling' rule to disambiguation.xml.
-A file 'hunspell/prohibit.txt' can now be used to mark words as spelling errors
even if the spell checker would normally accept them. This is useful to improve
the LanguageTool spell checker without waiting for the upstream checker to be
The 'prohibit.txt' file is the opposite of 'ignore.txt', which causes the
spell checker to ignore words.
-The part-of-speech tagger for most languages can now be extended by adding entries
to the file org/languagetool/resource/XX/added.txt (XX being the language code).
The format is "fullform baseform postag", three columns separated by tabs.
This makes it easier for users (and developers) to extend the POS tagger, as they
don't need to export, modify, and re-create the binary dictionary for every change.
2.7 (2014-09-29)
-added and improved rules
-New rule that checks if a weekday matches a date, e.g. detects
"Gwener 28 a viz Eost 2014", as that date isn't a Friday.
-added and improved rules
-fixed false alarms
-updated dictionary
-added and improved many rules
-switched to Morfologik-based spell checker
-Do you want to be part of the team that develops the world's most powerful
Open Source proofreading tool? We're looking for a maintainer for the English
rules in LanguageTool. See
for details.
-All English dictionaries have been extended to contain word frequency classes
to improve the spell checker suggestions (the frequency data is taken from,
as for other languages that already use this feature).
-Better suggestions for English learners: irregular verbs, nouns, and adjectives now
usually have a better suggestion. For example, 'thinked' suggests 'thought',
'womans' suggests 'women'.
-More misspellings provide suggestions now, e.g. 'garentee' (guarantee),
'greatful' (grateful). This may cause a performance decrease of ~ 10% (more for
texts with a lot of unknown words).
-New rule that checks if a weekday matches a date, e.g. detects
"Monday, 7 October 2014", as that date isn't a Monday. This rule will only
work if it detects the date format in use. So far, these formats are supported:
* "Monday, 7 October 2014"
* "Monday, 7 Oct 2014"
* "Monday, October 7, 2014"
* "Monday, Oct 7, 2014"
* (this also works with abbreviated week days like Mo or Mon for Monday)
-New rule that checks if a weekday matches a date, e.g. detects
"Vendredon la 28-an de Aŭgusto 2014", as that date isn't a Friday.
-updated POS tag dictionary and Hunspell dictionary to Dicollecte-5.2
-added a synthesizer - the agreement rule can now make suggestions
for some errors
-added/improved several rules
-New rule that checks if a weekday matches a date, e.g. detects
"vendredi 28/08/2014", as that date isn't a Friday.
-Fixed a rare NullPointerException and an ArrayIndexOutOfBoundsException
-Fixed several false alarms
-Added and improved rules
-New rule that checks for sentences without a verb (turned off by default due to
the risk of false alarms)
-New rule that checks if a weekday matches a date, e.g. detects
"Dienstag, 29.9.2014", as that date isn't a Tuesday.
-Performance improvements for spell check suggestions
-added initial support for Persian (Farsi)
-added and improved some rules
-new rule that checks if a weekday matches a date
-added/improved several rules
-added many dozens of compound words
-added new rules
-fix SourceForge feature request #38 (check for different quotation marks)
-added a few false friend rules (Russian/English)
-new rule that checks if a weekday matches a date, e.g. detects
"понедельник, 30 сентября 2014 г.", as that date isn't a Monday.
-expanded Russian compound rule with new words from postag dictionary
-Added new POS category Z (for spelled numbers, e.g. 'uno', 'dos', ...)
-Spelled numbers can now be detected and managed both in disambiguation and rules.
-Fixed some incorrect lemmas in POS dictionary.
-Added Hybrid chunker-disambiguator.
-Added initial support for Tamil. If the font for Tamil is not properly
displayed on your computer and you're using Windows, you might need
to apply the work around described here:
-big update for POS dictionary (fixes and new words)
-some POS tag renamed for consistency; new tags for abbreviations and rare words
-many new rules and fixes for existing rules
-new rule that checks if a weekday matches a date, e.g. detects
"понеділок, 7 жов 2014", as that date isn't a Monday
-token normalization performance improvement
-LibreOffice integration:
-Don't get confused by footnotes in LibreOffice 4.3 and later (it now provides us with
the footnote positions as meta data, so we can ignore them).
-Major performance improvements for the multi-thread use case, where JLanguageTool
gets created per thread, but the language object (e.g. 'German') gets
created only once. Overhead for creating JLanguageTool should now be much lower.
-Removed several classes and methods that had been deprecated since version 2.6
-Removed DutchSpellerRule - use MorfologikDutchSpellerRule instead
-The signature of Language.getRelevantRules() has changed
-The JLanguageTool and MultiThreadedJLanguageTool constructors don't declare to
throw an IOException anymore
-WhitespaceRule has been renamed to MultipleWhitespaceRule (WhitespaceRule
still exists but has been deprecated)
-Deprecated some methods whose visibility will be decreased (e.g. from public
to protected)
-MorfologikSpellerRule.getRuleMatch(String, int) has been renamed to
MorfologikSpellerRule.getRuleMatches(String, int)
-The RuleMatch constructor now throws an exception if toPosition is not
larger than fromPosition
-Introduced a new abstract class TextLevelRule that extends Rule and that
can be used for rules that cover more than single sentences.
-Command line:
-Enabling and disabling specific rules at the same time is now allowed.
In order to test only some rules (disabling all the rest), which previously was done
with "--enable LIST_OF_RULES", now use "--enabledonly --enable LIST_OF_RULES"
(or "-eo -e LIST_OF_RULES").
-Embedded server:
-Two new options can be set in the properties file to make LanguageTool
return the same XML format as After the Deadline (AtD). This way it can be used
as a drop-in replacement for AtD:
* mode - 'LanguageTool' or 'AfterTheDeadline'
* afterTheDeadlineLanguage - code of default language if mode is set to 'AfterTheDeadline'
NOTE: the 'AfterTheDeadline' mode should be considered experimental for now.
-The new option 'maxCheckThreads' allows setting the maximum number of threads working
on requests in parallel. The default is 10, as it used to be.
-New abstract rule AbstractDateCheckFilter that allows to check if a week day and date
match. For example "Tuesday, September 29, 2014" could be detected, as September 29, 2014 is
not actually a Tuesday. This uses the new experimental RuleFilter interface that can be called
from XML with the new 'filter' element. 'filter' takes these attributes:
'class': the fully-qualified name of a Java class that implements RuleFilter, e.g.
'args': a string like "year:\1 month:\2 day:\3 weekDay:\4", i.e. a space-separated list of
key/value pairs, where \x gets resolved to the pattern's token value (as in
the 'message' element)
-The compound rule now ignores tokens that have been immunized in the disambiguation.xml
-The "filter" action in the disambiguator is now applied only to POS tags that match the POS
tag given. If they don't match, the rule is not applied.
-If you're extending the XML rules as described at,
the external rule and disambiguation files can now be hosted on a password-protected server
by specifying an URL like this:
-The em dash ("—") is now a tokenizing character for all languages
-New feature: Use of language models
LanguageTool can now make use of ngram data. ngram data is information about
how often phrases occur in a language. Currently, this uses phrases of length 3.
The data is used by an English rule to find homophone errors, like mixing up
coarse/course or flair/flare. LanguageTool had some rules of this kind before, but
the new rule now supports about 900 of such word pairs/sets.
The data needed for this is huge (7GB for English) and thus not part or LanguageTool.
The data (English only for now) and more documentation is available at
Using ngrams makes LanguageTool slightly slower when the data is stored on an SSD.
If not stored on an SSD, the performance might drastically decrease.
Use the new --languagemodel option with the command line client to activate the rule
that uses the data. That option is not yet available for the stand-alone GUI.
2.6 (2014-06-30)
-updated FSA spelling dictionary from An Drouizig Breton Spellchecker 0.12
-updated POS dictionary from Apertium (svn r53329)
-improved several rules
-fixed false alarms
-added and improved rules
-updated tagger dictionary
-Morfologik spellchecking rule is enabled for use in LibreOffice/OpenOffice
extension. The Hunspell spellchecker should be manually disabled in
LibreOffice/OpenOffice for the results to be visible.
-The spelling rule now accepts words with hyphens if the parts are valid words.
For example, "web-based" is accepted now. This avoids a lot of false alarms.
-fixed a thread-safety problem in the synthesizer
-added/improved several rules
-added several rules
-updated POS tag dictionary and Hunspell dictionary to Dicollecte-5.1
-added/improved several rules
-fixed false alarm for words like "Stil-" in phrases like "Stil- und
Grammatikprüfung" (issue #93)
-added/improved several rules
-detect wrong uppercase spelling in sentences like "Die Blaue Tür"
-added two new punctuation rules
-added some rules, thanks to github user Shugyousha
-added/improved many rules
-improved disambiguation
-improved spell-checking of foreign words with apostrophes
-added/improved several rules
-a total of 3534 compound words
-added big wordlist to spellcheck dictionary. Thanks for wordlist to Dmitri Gabinski
-fixed wrong tag in tagger dictionary: ADJ_Comp --> ADJ_Sup, ADJ_S --> ADJ_Com
-added/improved several rules
-added and improved several rules
-Asturian, Italian, Lithuanian, Malayalam, Swedish, and Tagalog have been switched
to an SRX-based sentence tokenizer implementation.
-The deprecated options check-dump and wiki-index have been removed
from Please use check-data
and index-data instead.
-Almost all deprecated methods and classes have been removed. If you're upgrading
from a version earlier than 2.5, we recommend to upgrade to 2.5 first, fix all
deprecation warnings, and then upgrade to 2.6.
-If you extend the 'Language' class and don't implement the getSentenceTokenizer()
method, your language now uses SimpleSentenceTokenizer. This is a very simple
tokenizer and you probably want to implement getSentenceTokenizer() to return
a LocalSRXSentenceTokenizer instead that's adapted to your needs.
-Rule.supportsLanguage() now also works for PatternRules (i.e. rules loaded from
XML files). It used to always return 'false' for those rules.
-The public field Language.DEMO has been removed. It is now only available
internally for tests, together with the demo language itself and DemoChunker.
-JLanguageTool.printIfVerbose() has been changed from protected to private,
as it was not used anywhere and it's not really useful to extend JLanguageTool.
-JLanguageTool.getAllActiveRules() has been fixed to not return rules that have
the default="off" attribute (unless they have been enabled explicitly)
-StringTools.isAlphabetic() has been removed as it was a workaround for Java 6
which is not supported anymore
-ContractionSpellingRule.isDictionaryBasedSpellingRule() now returns false
-Specific rules can be enabled for use in LibreOffice/OpenOffice extension with
the new useInOffice() method. This is done, for example, for enabling Catalan
Morfologik spelling rule.
-Embedded server:
-The HTTP server now also accepts a --config option so that the maximum request size
can be limited with the 'maxTextLength' parameter (this used to be working only
for the HTTPS server)
-The HTTP and HTTPS servers now accept a new option 'maxCheckTimeMillis' in the
property configuration file to specify a maximum duration of a single check.
Checks that take longer (e.g. because generating spell corrections is slow for
some languages) will stop with an exception.
-Improved configuration dialog
-Allow the user to change the font of the main editing area
-Rule syntax:
-It is now possible to specify case sensitivity of individual exceptions
and tokens in a pattern (both in grammar and disambiguation XML rules).
Simply use case_sensitive="yes" or case_sensitive="no".
-The 'url' element can now be added to the 'rulegroup' so that all rules
of a rule group share the same URL.
-Fixed bug with longer complex strings of tokens to be unified
-Fixed bug in a disambiguator where tokens with min="0" were mislocated
2.5 (2014-03-31)
-a new rule checks that there is a space character between sentences
-added/improved many rules
-fixed false alarms
-added hundreds of suggestions for barbarisms in a simple replace rule
-remove a hack from the word tokenizer code
-several new false friends (by Sander van Geloven)
-added a new rule to handle corrections of mistakes in standard
English contractions (wasnt, didnt)
-changed word tokenization so that the hyphen at the end of a
word is no longer its part
-removed spelling replacement pairs that made LT work very slowly
-fixed a large number of false alarms
-added new rules to handle common contextual misspellings and
redundant phrases as well as common grammatical mistakes. Now
the number of rules for English exceeds 1000.
-made it possible to use synthesizer to both add a determiner and
make different replacement operations when creating suggestions;
simply add \\+INDT or \\+DT as special keywords to the regex that
creates the POS tag in the match element
-a new rule checks that there is a space character between sentences
-added/improved several rules
-a new rule checks that there is a space character between sentences
-updated POS tag dictionary and Hunspell dictionary to Dicollecte-5.0.2
-added/improved several rules
-a new rule checks that there is a space character between sentences
-fixed several false alarms
-added/improved several rules
-a new rule checks that there is a space character between sentences
-added some rules (thanks to Silvan Jegen)
-updated POS tag dictionary to PoliMorfologik 2.1
-cleaned up spelling rules and hyphenation rules, added some frequent
misspellings to generate proper suggestions
-added simple compounding support for compound words containing prefixes
such as "anty" or "mini", and adjectives with numerals such as "trzynasto"
-removed annoying false alarms
-added a number of new rules
-changed word tokenization so that the hyphen at the end of a
word is no longer its part; multi-word expressions with a hyphen
are also split into their component parts to avoid many false
alarms in spell-checking
-a new rule checks that there is a space character between sentences
-added/improved several rules
-added dozens of compounds: it now has 3400+ compound words
-a new rule checks that there is a space character between sentences
-added a few rules
-added some new rules, thanks to Julia Semenenko from
-added many false friends rules (Russian/English)
-updated POS tag dictionary and synthesizer dictionary from rev. 242
-added frequency information for spell-checking dictionary from
-fixed sentence detection so that sentences that begin with a lowercase
character can now be detected
-added common punctuation and mathematical characters (=#*∗×·+÷) to the
set of tokenizing characters in most languages
-added new element <unify-ignore/> in XML rules to mark the parts of the unified
sequence that are not checked (useful to ignore punctuation, connectives, or not
inflected words)
-added new element <antipattern> (which can include <token>, <and>, <unify>, and <marker>),
useful for marking complex multiword exceptions in rules and rulegroups (a single
antipattern can be shared by multiple rules in the same group)
-spell checking:
-Immunized tokens are now ignored and don't cause a spelling error anymore;
one can also add particular contextual expressions to ignored words by using
a new action of the disambiguator: "ignore_spelling".
-The words in ignore.txt are checked using case conversions allowed in the
speller dictionary (in MorfologikSpeller-based dictionaries).
This usually means that the word added in lowercase will be accepted
also when it is found at the beginning of the sentence; but if it is added in
uppercase, it won't be accepted when written in lowercase.
-updated Morfologik libraries to 1.9.0 to speed up suggestion generation in spell
checks, not only when using replacement pairs but also for ignoring diacritics.
-stand-alone GUI:
-The "More..." dialog now contains a link to
with details about the matching rule
-incompatible change: changed the return type of Rule.getLocQualityIssueType()
from String to ITSIssueType
-incompatible change: changed the parameter of Rule.setLocQualityIssueType()
from String to ITSIssueType
-deprecated Rule.isSpellingRule(), please use Rule.isDictionaryBasedSpellingRule()
2.4.1 (2014-01-08)
-updated Morfologik libraries to 1.8.3 to fix slow suggestions
in the spell checker, which affected at least en-US
2.4 (2013-12-30)
-SRX sentence tokenization
-added/improved a few rules
-fixed some false alarms
-fixed incorrect suggestions thanks to added tests on corrections
-added/improved several rules
-fixed false alarms
-made additions and fixes to the tagger dictionary
-removed some words from synthesis dictionary (see filter-archaic.txt)
-added frequency data to the tagger dictionary; frequency wordlist comes from the Gaia
project, with Apache License, version 2.0
-added/improved a few rules
-fixed some false alarms
-added/improved several rules
-fixed some false alarms
-added/improved several rules
-added a synthesizer - the agreement rule can now make suggestions
for some errors (not all suggestions are correct, though)
-added/improved several rules, especially for hyphen and dash usage
-added frequency information for spell-checking dictionary;
frequency wordlist comes from the Gaia project, with Apache License,
version 2.0
-fixed some false alarms
-added/improved several rules (it now includes gender rules "a"/"o")
-it now has 3400+ compound words
-the JAR file has been renamed to languagetool.jar, from formerly
languagetool-standalone.jar to avoid confusion about what 'standalone'
means in this context (github issue #29)
-for languages with many rules (like French or German) performance
on long texts has been increased by about 20-30%
-fix for thread-safety (could cause hang in MultiWordChunker)
-fixed a bug where chunk annotations were not tested in <and> groups
-fix: \1 and <match no="..."/> had not been evaluated in <short>...</short>
-fixed a bug in the unification mechanism that discarded some of the matching
interpretations prematurely
-added support for chunk annotations in the disambiguator, and fixed one bug
in filtering tokens with chunk annotations
-updated Morfologik libraries to 1.8.2 (bug fixes, stricter input sanity checking,
add frequency data to dictionaries)
-added the option of including frequency data to tagging or spelling dictionaries.
The expected format of the frequency wordlists is the one in the Gaia
project, with Apache License, version 2.0
-new command line tools to export and create binary dictionaries:
-LibreOffice/OpenOffice integration:
-added a workaround for incorrect sentence detection for the case that a
footnote appeared after a sentence full stop (Sourceforge bug #191)
-stand-alone GUI:
-The dialog opened by the "More..." item in the context menu of an error
will now also display correct and incorrect example sentences
-SentenceTokenizer is now an interface, the implementation has been moved to
RegexSentenceTokenizer, but this is deprecated and SRXSentenceTokenizer
should be used instead
-Some methods from have been moved to
the org.languagetool.gui.Tools class in the languagetool-gui-commons project
-LanguageToolListener.languageToolEventOccured() has been renamed to
LanguageToolListener.languageToolEventOccurred() isn't public anymore (shouldn't affect anybody)
-removed DanishSentenceTokenizer which had been deprecated for three years
-Rule.getCorrectExamples() and Rule.getIncorrectExamples() don't return null anymore
but an empty list if there are no examples. Consequently, setCorrectExamples() and
setIncorrectExamples() don't accept null anymore.
-Rule.getId() may return any string now, not just ASCII-only strings (actually this
has been the case before, as the ASCII-only restriction was never enforced and only
mentioned in the javadoc)
-languagetool-wikipedia: the command line options for checking a Wikipedia dump
have been simplified. The command can now be called like this:
java -jar languagetool-wikipedia.jar check-data -l en -f enwiki-20130621-pages-articles.xml
Call just "java -jar languagetool-wikipedia.jar check-data" to get a usage message.
More than one file can be specified with the -f option. Additionally to Wikipedia
XML dumps, CSV files from Tatoeba ( are now also supported,
they need to be filtered first to contain only the relevant language.
2.3.1 (2013-10-07, released on Maven Central only)
-fixes for thread-safety
2.3 (2013-09-30)
-added/improved a few rules
-fixed false alarms
-updated POS dictionary from Apertium (svn r47282)
-added support for language code ca-ES-valencia (Catalan Valencian),
to be used in LibreOffice 4.2.0
-added a simple replace rule with hundreds of replacement suggestions
-added/improved several rules
-fixed false alarms
-added a workaround for a StringIndexOutOfBoundsException
-added replacement patterns for the spelling checker to make suggestions
better (now offers 'taught' for 'teached')
-added/improved a few rules
-added/improved a few rules
-fixed false alarms
-updated POS tag dictionary and Hunspell dictionary to Dicollecte-4.12
-added/improved several rules
-added/improved a few rules
-it now has 3300+ compound words
-added/improved several rules
-the source code has been moved to github:
-LanguageTool requires Java 7 now
-LanguageTool makes use of multiple threads now for text checking on modern
hardware, improving performance (this affects the stand-alone version, the
command line version and the LibreOffice/OpenOffice extension)
-Rule syntax:
-preliminary support for new min/max attributes that allow to match an
element that appears the given number of times. For example:
<token min="0">foo</token> will match nothing or "foo", i.e. "foo" is optional
<token max="2">foo</token> will match "foo" or "foo foo"
<token min="0" max="2">foo</token> will match nothing, "foo", or "foo foo"
Use max="-1" to allow unlimited occurrences.
For min, only 0 or 1 is supported (1 is the default).
-support for OR-statements. For example:
<token postag="V"/>
Internally and in run-time, a rule containing OR-statements is converted into
several rules without OR-statements.
-English now has a chunker to detect, amongst others, singular and plural noun chunks.
This is documented at
-standalone version:
-The standalone version now underlines errors with a red (spelling errors) or
blue (other errors) line (Panagiotis Minos)
-Remember the language selection for the next start
-Improved window and dialog placement in a multi-monitor setup
-embedded server: uses default port (8081) again if started without arguments
-updated the morfologik-stemming library to version 1.7.1 to enable better suggestions,
including proper handling of diacritics and replacement patterns (equivalents of MAP
and REP features in hunspell dictionaries)
-OpenOffice/LibreOffice integration:
-fix: the "About" dialog didn't work in Apache OpenOffice 4.0
-fix: country specific rules (like for British English) didn't work
-In class Language, getCountryVariants() has been renamed to getCountries(), and a new method
getVariant has been added.
-Some methods have been deprecated
-Some methods have been moved from the Tools class (languagetool-core) to the
new CommandLineTools class (languagetool-commandline)
-AbstractRuleDisambiguator has been renamed XmlRuleDisambiguator and is not abstract anymore.
The <Language>RuleDisambiguator classes have been removed, XmlRuleDisambiguator can be
used directly instead.
-A new method JLanguageTool.check(AnnotatedText) has been introduced that allows
you to check text with markup. Use AnnotatedTextBuilder to build up the input.
-Thread-safety has been improved. The recommended use case is now to
create a new JLanguageTool object for each thread, but to create the
language only once (e.g. new English()) and use that for all JLanguageTool
instances. This changed the API of some public classes, but for the standard
use case of checking texts with the JLanguageTool object it shouldn't make a
difference. (patch by Stefan Lotties)
-JLanguageTool.loadFalseFriendRules() now behaves like JLanguageTool.loadPatternRules():
it looks in the class path first, and then, if the given file is not found there, in
the filesystem
-Introduced the Chunker interface that can assign chunks (also known as phrases)
to tokens. For example, for noun phrases like "a fast computer" the chunker could assign
an 'NP-singular' (noun phrase, singular) chunk to each of the tokens in that phrase.
In the grammar.xml, such a token can then be matched with this syntax:
<token chunk="NP-singular"></token>
-The new class MultiThreadedJLanguageTool makes use of as many threads
as the computer has processors. In our tests this has improved text checking
time by about 70% on an Intel i7 processor when used on 30KB text.
-AnalyzedTokenReadings now implements Iterable so it can be used in foreach loops
-AnalyzedGermanTokenReadings has been removed, AnalyzedTokenReadings can be used instead
-Embedded HTTP server: the server now uses 10 threads instead of 1 (thanks to
Panagiotis Minos)
-text extraction from Wikipedia dumps has been improved
2.2 (2013-06-30)
-added/improved several rules
-fixed some false alarms
-updated POS dictionary from Apertium (svn r45122)
-added/improved many rules
-fixed false alarms
-rules have been categorized according to the upcoming Internationalization Tag Set (ITS)
Version 2.0 standard from W3C.
-updated rules to fix false alarms, thanks to Ruud Baars
-The Dutch spell checking has been switched back to Hunspell for now to avoid too
many false alarms because of unknown compounds. Unfortunately, Dutch spell
checking does not provide suggestions anymore, for performance reasons.
The dictionary used is the one at
-added/improved a few rules
-added/improved several rules
-fixed some false alarms
-updated <url> links to PMEG.
-added/improved several rules
-fixed some false alarms
-updated POS tag dictionary and Hunspell dictionary to Dicollecte-4.10
-added/improved several rules
-added a few rules (by Panagiotis Minos)
-small rule improvement
-avoid an ArrayOutOfBoundsException in the POS tagger
-added some rules (by Nathan Wells)
-added a few new rules
-added/improved a few rules
-it now has around 2000 compound words
-added some new rules (thanks for these rules to Julia Semenenko)
-fixed some false alarms
-added new segmentation rules
-added false-friend rule
-added bitext rule
-added new style rules
-new POS dictionary
-new synthesizer dictionary
-new spelling dictionary
-new grammar rules
-updated sentence tokenizer rules
-disambiguator implemented
-word tokenizer updated to ignore accent and soft hyphen and understand different apostrophes
-HTTP server:
-enabling and disabling rules at the same time (keeping the rest of the default options) is now allowed.
To disable all the rules except those explicitly enabled, you can use the parameter enabledOnly=yes. Ex.:
-Fix bug "java.lang.StringIndexOutOfBoundsException" in DifferentLengthRule
(Ex.: http://localhost:8081/?language=ru&text=№&srctext=No.&motherTongue=en )
-Worked around the "There is an incompatible JNA native library installed on this system" error
-Updated Tika (used for language detection) from 0.9 to 1.3
-The "--version" parameter of languagetool-commandline.jar now also prints
the build date
-Several small bug fixes, code cleanups, and Javadoc improvements
2.1 (2013-03-31)
-added/improved several rules
-fixed several incorrect suggestions thanks to added tests on corrections
-spelling checker now ignores words with a POS tag, hence accepting words
which are in Apertium or in An Drouizig spelling checker
-updated FSA spelling dictionary from An Drouizig Breton Spellchecker 0.11
-updated POS dictionary from Apertium (svn r43173)
-added/improved several rules
-fixed multiple false alarms
-improved sentence and word tokenization
-the tagger dictionary has been fixed and expanded (added 9000 nameplaces;
115000 tagged forms from Softcatalà dictionary)
-verbal forms have been tagged according to regional variants (with script
-Hunspell dictionary (Softcatalà) has been replaced with LT tagger dictionary for spellchecking
-spelling checker now ignores words with a POS tag (from multiwords.txt or disambiguation.xml)
-fixed some false alarms
-added a few new rules
-updated the tagger and synthesizer dictionaries to recognize more words, including
plural forms of nouns, new verbs, and missing cardinal numerals (fixes Sourceforge
bug #3560624)
-updated the dictionary building script
-added/improved several rules
-fixed several incorrect suggestions thanks to added tests on corrections
-single quotes at beginning/end of word no longer cause false spelling error
-added/improved several rules
-updated POS tag dictionary and Hunspell dictionary to Dicollecte-4.9
-fixed several incorrect suggestions thanks to added tests on corrections
-SRX sentence tokenization
-single quotes at beginning/end of word no longer cause false spelling error
-added/improved several rules
-The German spell checker now returns suggestions. We combine Hunspell and our own
morfologik-based spell checker to create the suggestions for good performance,
but they are far from perfect yet. First, some misspelled compound words fail to have
a suggestion. Second, sometimes the best suggestion is not at the front of the list
of suggestions.
org/languagetool/resource/de/hunspell/ can be used to create
the three morfologik spell checker dictionaries (German, Austrian, and Swiss).
-New language variant "Simple German", with rules contributed by Annika Nietzio.
This variant uses 'de-DE-x-simple-language' as a language code.
-added/improved a few rules
-now uses the default sentence tokenizer, not the English one (because languages
are now modules that don't depend on each other)
-false alarm fixes
-added and improved some rules
-major tagger and synthesis dictionary update (to Morfologik 2.0 PoliMorf),
which also fixes bug #3554018
-added/improved several rules
-it now has over 1500 compound words
-several new rules
-bugfix: the uppercase sentence rule sometimes was not triggered
-spell checking: URLs, if specified like "", are now ignored and
don't cause a spelling error anymore
-API: Languagetool has been split up into several Maven modules. This causes
some API changes:
-No more language constants: "Language.GERMAN" now needs to be "new German()",
similarly for all other languages
-org.languagetool.gui.ResourceBundleWithFallback is now org.languagetool.ResourceBundleWithFallback
-org.languagetool.gui.ContextTools is now
-LanguageIdentifierTools.ADDITIONAL_LANGUAGES has been removed, languages are
detected at runtime now
-rule syntax: Suggestions are now also allowed outside the <message> element. Example:
<message>Did you mean <suggestion>A</suggestion>?<message>
<example type="incorrect" correction="A|B|C"><marker>asdf</marker></example>
-bugfix: suggestions for compounds parts were missing sometimes
-bugfix: Portuguese translation was not used
-bugfix: fix false alarm of unpaired bracket rule on smileys :-) and ;-)
(Sourceforge bug #3604367)
-bugfix: testrules now report all cases when untouched examples were touched
(Sourceforge bug #3600995)
-bugfix: the whitespace rule now reports also combinations of non-breaking space and white space
characters (Sourceforge bug #3608410)
-stand-alone GUI: the very first check for languages with a lot of rules
(e.g. German, French) should now be faster
-embedded HTTPS server:
-two new properties, to be set from the property configuration file,
allow limiting the maximum number of requests:
requestLimit - the maximum number of requests
requestLimitPeriodInSeconds - the time period in which the requests are considered, in seconds
-bugfix: only the first line of a POST request's body was considered
-OpenOffice/LibreOffice integration:
-Another try to fix the ConcurrentModificationException (Sourceforge bug #3572536)
-Command line:
-In verbose mode, the subId of disambiguator matched rules is displayed.
-Testing rules:
-Now regular expressions in disambiguation rules are heuristically tested
(Sourceforge bug #3599002)
2.0 (2012-12-30)
-added several new rules and fixed false alarms
-updated POS dictionary (fix wrong POS tags) from Apertium (svn r42124)
-updated FSA spelling dictionary from An Drouizig Breton Spellchecker 0.10
-updated classification of words tagged as plural nouns of persons
(these have different mutation than other nouns)
-many new rules
-fixed false alarms
-multiple fixes and additions in the tagger dictionary
-Hyphen is used as a word separator, as this is very common in web pages. This way some
ambiguities cannot be solved if the typography rule substituting hyphen for dash is disabled.
(For example: Dir-li -ho és important.)
-uses the same word tokenizer again as other European languages
-fix Sourceforge bug #3479817 (capitalization of the first item of a list not required)
-a few rule updates
-few new rules
-updated POS tag dictionary and Hunspell dictionary to Dicollecte-4.8
-fixed false alarms
-many new rules and rule updates
-Added hundreds of compound words
-Added/improved several rules
-Replaced the current dictionary with PT-pt (pre-language agreement).
To use the post-language agreement, use the PT-br one.
-improved some rules
-several new rules
-updated POS dictionary (fix wrong POS Tags)
-several new rules (thanks to Alexander Socop Arango)
-updated a few rules
-OpenOffice/LibreOffice integration:
-Fixed ConcurrentModificationException (Sourceforge bug #3572536)
-stand-alone GUI:
-the tray icon menu (reachable with the right mouse button on the tray
icon) now has a checkbox to enable the embedded HTTP server
-the tray icon will show a small "S" symbol when the server is running
-fixed bug "Tray icon too big sometimes" (Sourceforge #3573078)
-Language.getLanguageForShortName() now consistently throws an exception
if the given language code is not known
-Tools.median() is now private (it was accidentally made public)
-the Java API of HTTPServer has been modified in incompatible ways. You might get
compile errors if you have used this class from your Java code.
-support for auto-detecting text language (parameter autodetect=1)
-added HTTPSServer, a lightweight embedded HTTPS server which works like HTTPServer
but supports SSL encryption. This server supports *only* https, not http.
-the XML we return now contains a new attribute "locqualityissuetype", which
is the "Localization Quality Issue Type" in the upcoming Internationalization Tag Set (ITS)
Version 2.0 standard from W3C. This means errors are now categorized according to
a standard, additionally to LanguageTool's own categories. Useful values are only
returned for English for now.
*** Please consider this to be a prototypical implementation for now ***
For the values and their meanings, please see
For rule developers: specify this using the new 'type' attribute. It is
inherited from category to rulegroup, and from rulegroup to rule. If a rule also
has the 'type' it overwrites the rulegroup's and category's 'type'.
1.9 (2012-09-30)
-several new rules and fixed false alarms
-the internal spelling engine can now additionally tokenize words if the
source dictionary did not include compound versions (for example,
hyphenated words).
-updated classification of words tagged as plural nouns of persons
(these have different mutation than other nouns).
-many new rules
-fixed false alarms
-hundreds of new Chinese rules (Tao Lin)
-unit tests for the Chinese tokenizer and tagger
-major update to the dictionary and the rules (Henrik Bendt)
-fixed false alarm (sf bug #3543914)
-some new rules and rule updates
-several new rules and fixed false alarms
-updated list of transitive/intransitive verbs
-added <url> links to REVO (Reta Vortaro) with permission from Wolfram Diestel.
-updated POS tag dictionary to use Dicollecte-4.6
-several new rules and fixed false alarms
-opening guillemet (U+00AB) followed by thin no-break
space (U+202F) wrongly detected as error (sf bug #3545050)
-added simple verb/subject agreement checker
-several new rules and rule updates
-fixed some false alarms by upgrading to jwordsplitter 3.4 (sf bug #3475553)
-initial support with about 20 rules (Takahiro Shinkai, Maxim Mozgovoy's student)
-fixed bug with the word repetition rule (sf bug #3560925)
-several rule updates (Marco A.G.Pinto)
-several new rules and rule updates
-improve user interface translation
-Support for Swedish has been re-enabled after it had been disabled in LanguageTool 1.7.
-bugfix for command line: We removed XML from even plain text input. Now XML/HTML elements
are only filtered out if the new --xmlfilter option is specified. Note that there's still
a bug that can screw up position information with that option, thus it is deprecated.
-command line: new option "--list" to list all available languages
-introduced a file resources/<lang>/hunspell/ignore.txt with words that the spell checker
will ignore
-development: new ant target generate-ignore-files: call "ant generate-ignore-files" if you
want to re-generate the ignore.txt files, populated with those suggestions from the grammar
files that the spell checker would complain about.
In other words, this makes sure "simple" suggestions (i.e. those without back references) made
by the LanguageTool rules are never considered spelling mistakes.
If you have downloaded the stand-alone ZIP of LanguageTool, call this command for the
same effect:
java -cp LanguageTool.jar org.languagetool.rules.spelling.SuggestionExtractor
-stand-alone GUI: rules can now be disabled and re-enabled with a single click
-stand-alone GUI: copy and paste from the result area now keeps line breaks (Sourceforge
bug #3554842)
-HTTP API: "+" was incorrectly removed from input (Sourceforge bug #3543914), patch by
Sourceforge user hiddenspirit
-HTTP API: the XML returned by the "/Languages" URL (which returns a list of all supported
languages) has been extended to contain an 'abbrWithVariant' attribute which lists the
language with its variant, e.g.:
<language name="Catalan" abbr="ca" abbrWithVariant="ca-ES"/>
-HTTP API: small internal cleanup for better exception and charset handling (always
expects UTF-8)
-HTTP API: the embedded HTTP server can now be started with a "--public" parameter so
it can be accessed from anywhere (not just localhost). Please be careful with this
option, it is not recommended for production use:
java -cp LanguageTool.jar org.languagetool.server.HTTPServer --public
-HTTP API and XML output: extended XML to include the version and build date
of LanguageTool and the category and offset of each match
-The word tokenizer now considers the following characters as a word separator: | (pipe)
and ` (backtick).
-bugfix: the column count was sometimes totally wrong (Sourceforge bug #3557738)
1.8 (2012-06-30)
-initial support for Greek with few rules for now (Panagiotis Minos)
-many new rules and rule updates
-added spellcheck dictionary for the standalone version; the dictionary used is the
one by Franz Michael Baumann,
LanguageTool can now detect spelling errors, but it does not offer suggestions
for performance reasons
-added variants for Austria, Germany, and Switzerland
-added several thousand words to the internal dictionary for better
error detection and less false alarms
-several rule improvements and new rules (Scott Kuhl)
-although a matter of debate, "an historic", "an historical", and "an habitual"
is not considered incorrect anymore,
also see
-fixes in the tagger dictionary and disambiguation rules to reduce false alarms
-added British and American English spelling dictionaries
-added some rules to detect typical American expressions in British English
-major update, including many new rules and new tokenization (Jaume Ortolà i Font)
-new synthesizer (given a part-of-speech, this generates the inflected word
forms) (Jaume Ortolà i Font)
-added spellcheck dictionary for the standalone version
using Hunspell dictionary ca-valencia (avl) 2.3.0 by Softcatalà
-fixed a few false alarms
-added URL element for some rules with URL with more information about this rule
-added spellcheck dictionary for standalone version.
-several rule improvements (adding links, improving messages)
-new Java rule "repeat word in sentence" (disabled by default)
-several new rules (Riccardo Murri and Agnese Dal Borgo)
-added spellcheck dictionary for the standalone version
-updated dictionary to use Dicollecte-4.5
-several new rules and fixed false alarms
-added spellcheck dictionary for the standalone version
using Dicollecte-4.5 «classique & réforme 1990».
-updated dictionary to use Apertium svn r38896.
-several new rules and fixed false alarms
-added some references to the Breton grammar:
"La grammaire bretonne pour tous (Eugène CHALM)".
-added spellcheck dictionary for the standalone version
-added spellcheck dictionary for the standalone version
-some new rules and fixed false alarms
-major update of the tagger dictionary
-initial support by Marco A.G. Pinto
-several new rules and fixed false alarms
-added <url> for further information about errors with links to
PMEG (Plena Manlibro de Esperanto) with permission of its
author Bertilo Wennergren, and to also with
permission of the lernu team.
-added spellcheck dictionary for the standalone version
-LibreOffice / OpenOffice integration:
-If a LanguageTool rule has an URL with more information, the
grammar checking dialog in LibreOffice will now offer a "More..." link
to that URL. That URL is also displayed on the command line and stand-alone
user interface as "More info:", and as "url" attribute of the <error>
element in the XML API format. This makes it possible to have external
documentation about rules.
-Bugfix #3526635: SingletonFactory now implements XServiceInfo (Stephan Bergmann)
-Bugfix #3534637: Fixed false alarm about word being uppercase when the previous
sentence ended with a footnote
-made the result of "Tag Text" a bit more readable
-Bugfix: when LT was hidden in the tray, two very fast mouse clicks could activate
the check twice, and that caused errors. Right now only one checking thread is active.
-Command line:
-In verbose mode, the log of the rule-based disambiguator actions is displayed.
-In the profiling rules' mode on the command-line, you can now enable
and disable rules.
-Deprecated some methods and the SentenceTokenizer class (SRXSentenceTokenizer
should be used instead)
-LanguageTool in the standalone version now supports spell-checking. Some of the languages
use hunspell (, some use LanguageTool-specific code.
There are now two distribution files: a .zip file for standalone use,
and an .oxt extension for LibreOffice/ApacheOpenOffice. All languages supported by
LanguageTool, except for Chinese, have spelling dictionaries bundled with the standalone
version. Spell-checking errors appear in red in the LanguageTool standalone
version, other errors appear in blue.
-LanguageTool now supports separate rules for different local variants of a language,
for example American English and British English. To use them from the command line,
simply use "en-US" or "en-GB" as the language code. This implements feature
suggestion #3287388.
-The configuration file now stores settings individually for all languages. This means
that you can enable spellcheck for American English, disable it for Polish, and British
English, and all these settings will be saved separately.
-If you start the HTTP Server from the GUI, it now reads the configuration files
that are configured in the GUI (if the appropriate checkbox is set).
This way the user can control the behavior of the server easily.
-Two new options for the HTTP Server added: "disabled" and "enabled", which is used
to disable or enable rules in the same way as on the command-line.
-The XML format for rules has been changed to use <marker>...</marker> tags instead
of mark_from and mark_to attributes.
-Overlapping rule matches are filtered now so that only the first match per <rulegroup> is kept
-It is now possible to suppress misspelled suggestions altogether in XML rules by applying
an attribute suppress_misspelled="yes" on the <suggestion> element, AND on the <match>
element. If only <match> element has this attribute set to "yes", then the suggestion is
displayed, but no content of <match> is contained within (this might be a conditional part
of the suggestion). Note: for this to work, the tagger dictionary needs to be fairly complete;
words without lemmas and POS tags are considered to be misspelled.
-Improved startup speed (Jarek Lipski)
-Some internal bug fixing in disambiguation and pattern rules, including problems
with unification.
-Update of morfologik-stemming library to 1.5.3 (bug-fix release).
-Larger tagger dictionaries were encoded in cfsa2 binary format to save space.
-Bugfix #3054895: fixed incorrect column reported by LanguageTool in command
line mode or with --api option when error spanned new line.
-Bugfix #3431788: When the message contained the reference to a previous token as the first
character, the corrections generated via regular expression replacements on match elements
were wrong.
1.7 (2012-03-25)
-a few new rules
-updated dictionary to use Apertium svn r36822
-several new rules
-several new rules
-updated dictionary to use Dicollecte-4.4.1
-several new rules
-fixed a few false alarms
-added a few new rules
-fixed some false alarms
-added several new rules
-fixed a bug in the POS tagger incorrectly returning lowercase word, breaking
the rule that checks for uppercase sentence starts (Maksym Davydov)
-Support for Swedish has been disabled because it was not actively maintained
for a longer time and never reached a level that the old maintainer was happy
with. It will be activated again as soon as a new maintainer is found, any
help is welcome!
-bugfix: in LibreOffice/, the wrong error message was displayed
for some rules (those that did not have a "short" message)
-bugfix: the "fromx" position in the XML output was wrong sometimes
-bugfix: the "contextoffset" position in the XML output was wrong
when there were XML-escaped characters in the context
-consecutive tab characters are not considered an error anymore
-Ellipsis ("…") is no longer considered as a sentence separator (only relevant
for languages not yet using SRX tokenizer).
-new optional rule element "url" (optional): an URL to a page that explains
the rule leading to the error in more detail. Not used yet.
-GUI: errors are now marked with red background instead of red font
so whitespace problems can easily be spotted
-Server mode: by default, the server only binds to localhost now. Although requests from
non-localhost had been denied before by default, this should be more secure.
-Internal changes:
- adapted to a change in Unicode handling in Java 7 that broke support for Khmer
- added sanity check to detect duplicate exceptions in grammar rules.
- added sanity check to detect exceptions with scope="next" yet
without skip="..." in grammar rules.
1.6 (2011-12-31)
-updated dictionary from Apertium
-several new rules
-corrected and improved style of messages, translated all messages
thanks to Fulup Jakez
-several new rules (Tao Lin)
-new Java rule to check if three successive sentences begin with the same word
-fixed false positives
-updated list of transitive verbs
-updated dictionary to use Dicollecte-4.3
-fixed false positives
-several new rules
-new Java rule to check if three successive sentences begin with the same word
-updated compound splitter component to fix some false alarms on agreement (jwordsplitter 3.3)
-several new rules (Paolo Bianchini)
-configuration of Khmer rules should now work in
-small updates for coherency rules and error rules (Nathan Wells)
-added multiwords disambiguation rules
-several new rules
-several new rules (Martin Srebotnjak)
-new trigram profile
-new rules
-GUI: small cleanup of the user interface
-XML output encoding was wrong in server mode on platforms not using UTF-8
-Internal changes:
-Update to Morfologik 1.5.2, fixes a rare crash for at least German
-moved the packages from de.danielnaber.languagetool.* to org.languagetool.*
-added new public field JLanguageTool.BUILD_DATE
1.5 (2011-09-25)
-Added support for Chinese (Tao Lin)
-Added support for Asturian (Xesús González Rato)
-Added support for Tagalog (Nathaniel Oco)
-Added support for Breton (Dominique Pelle)
-Fixed some of the problems in the Sourceforge bug #3396850 (false alarms)
-Extended the internal part-of-speech dictionary with
about 2000 word forms. This should improve coverage of the
agreement rule. It mostly affects words that have a new spelling
after the spelling reform. See
for the source of the new data.
-Several new rules
-Updated dictionary to use Dicollecte-4.2
-Added and updated a few rules. A few rules taken from
Grammalecte-0.0.12 written by Olivier R.
-Added disambiguation for Esperanto
-Improved SRX sentence tokenization rules
-Added and updated a few rules
-Updated translation
-Initial SRX sentence tokenization rules
-Updated translation
-Initial SRX sentence tokenization rules
-Fix SourceForge bug #3334498 (False warning about unpaired bracket)
-Added and updated a few rules
-Updated and improved part-of-speech dictionary
-Fixes and updates to rules (R. J. Baars)
-Updated the dictionaries
-Added SRX sentence tokenization rules
-Added new rules
-Added more false friends
-Updated dictionary to support enclitics
-Added disambiguation support and rules
-Added synthesizer
-Updated coherency rules
-Rule development:
-Fixed to not throw an AssertionFailedError error if called
with a language parameter
-Fast Rule Evaluation is supported in "dev.index" packag, see
-LanguageToolGUI and LanguageTool now have support for automatic language
detection (Michael Bryant). On the command-line, you can use -adl or
--autoDetect to enable automatic language detection.
Also see
-Internal changes:
-The disambiguator has a new action "immunize" that protects
token from being matched by error-detecting rules (especially
in XML rules).
-The disambiguator allows removing interpretations by specifying just some
attributes with the <wd> tag.
-Allowed three character codes for languages, e.g. 'ast' for Asturian. This
is the ISO 639-2 code, as Asturian doesn't seem to have an ISO 639-1
two-character code.
-Fixed the bug with the special tag UNKNOWN: now it deals properly with
the words at the end of sentence / paragraph.
-Languages now initialize their classes (tokenizers, part-of-speech taggers
etc.) lazily, i.e. only when the classes are actually needed. Improves
startup performance a bit.
-Switched to for translation. Also changed our code to use the
English string if no translation exists.
1.4 (2011-06-26)
-LanguageTool now requires Java 6.0 or later
-English: added a few new rules
-German: added a few new rules
-Updated dictionary to use Dicollecte-4.1
-Added a few new rules
-Updated list of transitive verbs
-SRX sentence tokenization rules
-Word tokenizer now properly handles words with apostrophe
-Added a few new rules and fixed false positives
-Khmer: Added support for Khmer (thanks to Nathan Wells)
-Russian: added a few new rules
-Pressing Ctrl-Return will check the current text
-Fixed pre-selection of user's language
-Made screen messages and buttons in Language Module Manager translatable,
thanks to Ilona Kuzmickaja.
-enabled bilingual mode for HTTP API; if you use srctext parameter, LT
will automatically check in bilingual mode, assuming that mothertongue
specifies the source language, and lang - target language
-renamed RuSimpleReplaceRule to RussianSimpleReplaceRule
-renamed SlovakVes to SlovakVesRule
-renamed JLanguageTool.paragraphHandling to JLanguageTool.ParagraphHandling
-removed deprecated GermanSentenceTokenizer
-Internal changes:
-Java rules are not loaded dynamically from the classpath anymore,
instead every language needs to implement the getRelevantRules()
method that returns the rule classes relevant for that language.
-Removed jaminid dependency for HTTP Server, thanks to Ankit.
-The HTTP server doesn't block anymore when checking a long text. It
now has a longer startup time per call though (50-150ms).
-Developers using LanguageTool as an API can now configure from which
addresses the embedded HTTP server will accept requests
-Rule development:
-Extended and testrules.bat so that they can take a
language code as argument to only check that language.
Thanks to Michael Bryant.
For example: sh en
1.3.1 (2011-03-28)
-Fixed a NullPointerException in UnpairedBracketsRule
1.3 (2011-03-27)
-Updated rules for Dutch, thanks to Ruud Baars
-Updated rules and disambiguator for French
-Updated the French dictionary to use Dicollecte-4.0.1
-Updated rules for Russian
-Added disambiguation Russian rules
-Added disambiguation support and disambiguation rules for Spanish, reducing
false alarms in a significant way.
-Forked Spanish word tokenizer fixing word recognition when it is after
a dash.
-More and updated Spanish rules, adding multiple suggestions when applicable.
-Updated Spanish rule descriptions preventing apparently duplicated rules to
appear in configuration windows.
-Revamped short descriptions making them more informative improving Spanish
OpenOffice experience.
-New Spanish rule grouping with new categories improving configuration
-Updated rules and tagger for Esperanto
-Updated rules and dictionaries for Polish (to Morfologik 1.8)
-Small cleanup of rules for English, some new rules
-Added one exception to Dutch SRX rules to avoid false alarms
-Bugfix: made loading of external rule files in the GUI work
-Bugfix: fixed the bug in UNPAIRED_BRACKETS rule that caused false alarms
in languages that use the same characters to end different pairs of
quotation marks (especially Dutch)
-Bugfix: fixed the sf bug #3006662 for English (false alarm for
"...those ideas came.")
-Bugfix: fixed the sf bug #3200352 for English (the rule for "can not" is now
switched off by default)
-Bugfix: fixed the sf bug #3163988 for Polish ("PatternSyntaxException:
Unmatched closing ')' near index 33 ...")
-Updated help pages for English and Russian
1.2 (2011-01-02)
-New and updated rules for Romanian, Dutch, Polish, German, Russian,
Spanish, French and Danish.
-Fixed false positives in French rules.
-Updated the French dictionary to use Dicollecte-3.9.1
provided by Olivier R.
-Updated the Polish dictionary.
-Added new scripts and testwikipedia.bat to the distribution. These
let you check a local Wikipedia XML dump. This helps to test new rules, especially
to avoid false alarms. Get the Wikipedia dumps from
The file you probably want to use is pages-articles.xml.bz2.
-SRX sentence tokenization rules for German, thanks to Jarek Lipski.
-SRX sentence tokenization rules for Danish, thanks to Esben Aaberg.
-Added support for Esperanto.
-Now it's possible to enable and disable rules in bitext mode
from the command line.
-In bitext mode, the suggestions can now be automatically applied
to target text.
-Fixed API output for bitext mode.
-Fixed a bug with mark_to and mark_from attributes being ignored
for bitext XML rules.
-Fixed a glitch in API output: now it's always correctly encoded as UTF-8,
just what the XML header declares.
-Fixed sf.bug #3076989: incorrect line numbers for files larger than 64000 bytes.
-Minor performance fixes.
1.1 (2010-09-26)
-Added 'subId' attribute to API output - identifies the sub id of a rulegroup
-Updated French rules, thanks to Dominique Pelle.
-Updated Dutch tagger and synthesizer dictionary, as well as
fixed some false alarms and updated grammar rules,
thanks to Ruud Baars.
-Fixed a problem with Dutch apostrophes in the "unpaired brackets and
apostrophes" rule.
-Enabled disambiguation rule file testing from the command line. Now
the testrules.bat/ runs the disambiguator tests as well.
1.1beta (released 2010-09-18, internally marked as 1.0.1-dev)
-Update of the decompounding component for German (jWordSplitter 3.1)
-Added preliminary support for checking bilingual texts (bitexts),
including use of false friends rules, bitext pattern rules and
generic Java rules. LanguageTool may be therefore used via API
to check translation quality in computer-aided translation tools.
-Added a lot of new "false friend" rules (mostly Italian), thanks
to Fabio Rainone
-Fixed line number when checking texts from stdin (but there are still
some problems with the column number)
-Fixed LT menu integration for 3.3.0. (OOo Issue 114078)
-New method JLanguageTool.getDataBroker() that provides a
ResourceDataBroker which can be used to load LanguageTool's
resources (grammar.xml etc) from any path, not just from the
class path.
Patch provided by Andreas "PAX" Lück.
-Add initial support for help system.
Help available for English and Russian.
-Fixed the "LanguageTool server could not be started..." error that
occurred when closing the options dialog and the HTTP server
had been activated before
-Updated a tagger and synthesizer dictionary for Russian.
Fixed POS tag for Russian.
-Added a synthesizer for Spanish, thanks to Juan Martorell.
-Fixed Spanish dictionary separator issue, thanks to Juan Martorell.
-Fixed several Spanish dictionary POS tag flaws, thanks to Juan Martorell.
-Added a tagger dictionary and synthesizer for Catalan, thanks to
Jimmy O'Regan.
-Added an SRX-based sentence tokenizer for Catalan, thanks to Jimmy
O'Regan, and for Spanish, thanks to Juan Martorell.
-Major update of the French dictionary, thanks to Olivier R. The
dictionary is now based on data from the Dicollecte project.
-Updated the morfologik-stemming library used for tagging dictionaries
to version 1.4.0 (performance improvements and ability to build the
binary automata directly from Java)
-Fixed a problem with salutation ending with comma and false alarm
for a missing uppercase letter ( bug #2830181 for German;
the problem appeared also for Italian, Polish, and Russian)
-Improved the rule that detects wrong space around brackets; it
detects errors concerning curly brackets, parentheses, and
square brackets.
-Added exceptions for the rule that checks the whitespace after the
comma so that two consecutive commas are detected only by the double
punctuation rule.
-Initial support for Malayalam (thanks to Jitesh V. S.)
-Initial support for Belarusian (thanks to Alex Buloichik)
-Added some German rules (thanks to Jan Schreiber)
-Added several German and a few English rules (thanks to Felix Sasaki and
his students)
-Added much more Spanish Rules (thanks to Juan Martorell)
-Improved German sentence detection (no sentence split at
"Friedrich II. ist auch...", "13. und 14. Jahrhundert")
-Fixed sentence tokenizer to better deal with Windows line
breaks (\r\n instead of just \n)
-Small fix for English rule ("rather then" -> "rather than"), some
new English rules thanks to Jimmy O'Regan
-New Catalan rule for castellanisms, thanks to Jimmy O'Regan
-Fixes and additions for Catalan, Dutch, Italian, Polish, Russian,
Slovak and Slovenian rules
-Fixes for Polish tagging and synthesis dictionaries
-Fixes and additions to the Polish disambiguator
-Polish chunker now uses a separate text file for its multiword tokens
(see resource/pl/multiwords.txt)
-Fixed a #2957133 bug in Italian Rules (thanks to Paolo
-Added a rule for correct whitespaces around punctuation for French to
the default set of rules working for French
-Updated SRX-based sentence tokenizer rules for Polish, Russian and Slovenian
-Fixed bug #2998138 concerning an incorrect ICU regex in
SRX file
-Updated Slovenian and Russian translation
-Added some disambiguation rules to simplify punctuation rules in Polish
-It is now possible to add just a new POS tag using <wd> element in the
disambiguator. If you omit the lemma attribute and/or the token, the token and
the lemma will be set to the matched token.
-It is now possible to include skipped tokens from the sentence when referring
to parts of the pattern in suggestions. Use include_skipped as attribute
of the <match> element. To include all the tokens, including the one with the
skip attribute set, use "all". To include only the skipped sequence, use "following"
as the value of include_skipped. To include nothing, use "none".
-The syntax used for unification was slightly modified to make it more XML-friendly.
The details are to be found on the wiki at
-DTD for disambiguation rules is now replaced by an XML Schema.
-Fixed a bug that could raise a false alarm for some unchecked exceptions
in XML rules.
-Fixed a bug with the rule disambiguator: sentence end and paragraph end
tags no longer get filtered out during disambiguation.
-Fixed a small bug with unification: now unifier used for rules and
disambiguation rules are kept separate, so that you have different sets
of unification equivalences, as defined in their files.
-Fixed a bug with footnote numbers before a comma that were reported
as whitespace.
-Fixed bug #2972055 (LanguageTool menu no longer appears in
Calc where it didn't work anyway), (Yakov Reztsov).
-Fix to make HTTP server mode work on more Mac OS X versions
1.0.0 (2009-11-01)
-translation fixes, especially for Danish
-small GUI fix: making port field larger
-fixed false alarm for Polish
-small rule fix for Russian
1.0.0-rc1 (2009-10-25)
-Support for Danish, thanks for Esben Aaberg
-Support for Catalan, thanks to Ricard Roca
-Support for Galician, thanks to Susana Sotelo Docio
-Corrections to Spanish dictionary (conversion from Freeling)
and a new rule for Spanish that checks if the determiner
(if any) preceding a feminine noun is "el", thanks to Susana Sotelo
-Small corrections to Swedish dictionary, thanks to Niklas Johansson and
corrections to Dutch dictionaries, thanks to Ruud Bars; small corrections
to the French dictionary
-Added more features for the server-side LanguageTool, including setting
some parameters.
-Added missing support for New Zealand and South African English.
-More Dutch, Polish, Russian, Spanish, Slovenian, and English rules
-Enabled general rules in the Slovenian module
-Fixes for the Italian, Russian, French, and Polish rules
-Fixed sourceforge bugs: #2819925 for English, #2879819, #2883277, #2882789,
#2884642, #2880448, and #2885488 for French, #2848079 for Polish, and
#2886208 for Spanish.
For details, see<ID>
-The simple replacement rule (designed originally only for Ukrainian)
can be used now with any language to find simple, non-context dependent
typos. Implemented now for Ukrainian, Russian, Romanian, and Polish.
-Added SRX-based sentence tokenizers for: Slovenian (thanks to Petra
Galuscakova), Icelandic (thanks to Anton Karl Ingason), Slovenian
(thanks to Martin Srebotnjak), and Russian (thanks to Yakov Reztsov).
-Added new Slovak tagger and synthesizer dictionaries (thanks
to Zdenko Podobný and Slovak National Corpus)
-Added Russian synthesizer dictionary and new rules based using it.
-Added Russian disambiguator.
-Change from DTD to XML schema for rule files.
-Slight enhancement for testing XML files on the command line.
Slight improvement of word tokenization.
-Update of the Italian tagger dictionary (Morph-it! 0.48)
-Update of the decompounding component for German (jWordSplitter 3.0)
-Update to segment library, removed loomchild-util library that
is no longer needed.
-Update to morfologik-stemming library, resulting in some
tagging speedup. Tagger & synthesizer code cleanup.
-Fixed a small bug that could result in unpredictable behavior
in rules using match elements inside tokens.
-Fixed a bug with counting column positions (sf bug #2878819).
-Fixed a bug in unification.
-Fixed a bug with matching elements that have POS tag information
and whitespace attributes. Fixed a bug with losing whitespace info
from AnalyzedTokenReadings when copying them.
0.9.9 (2009-05-22)
-Added preliminary support for Icelandic thanks to Anton
Karl Ingason
-More English, Dutch, Polish and Romanian grammar rules.
More English/Polish false friends. Improvements to Russian
-Fixed some bugs with pairing brackets. Now the rule also
complains about balanced parentheses put in the wrong
order (like ")abc("), or overlapped with other symbols (like
'his "mother'") and in all places the symbols weren't
found to be balanced.
-Added sentence tokenizer that uses SRX format for
specifying end-of-sentence breaks (for specification, see This
enhanced the quality and speed of sentence tokenization
(for example, tokenizing the whole Brown corpus in the old
tokenizer took 572613ms, while it takes 125971ms right now).
Currently, English, Polish, Romanian and Dutch are using
SRX segmentation rules.
-Added Dutch (Belgium) as a country variant to support it
-Updated Russian translation (Yakov Reztsov)
-Updated Slovenian translation and added more rules
for this language (Martin Srebotnjak)
-Updated Slovak translation and added Slovak sentence
tokenizer (Zdenko Podobný).
-Fixed a NullPointerException that could appear when using
paragraph-level rules (SF bug # 2787814), removed bugs that lead to false positives
in some English rules. Some code cleanup.
-Fixed a bug that truncated the suggested replacements for false
-Fixed encoding problem for some systems (SF bug #2794857), thanks
to Petr Mladek.
0.9.8 (2009-04-25)
-Much more new rules, rule-based disambiguator and synthesiser for
Romanian (thanks to Ionuț Păduraru)
-New rules for Italian and translation of messages into Italian
by Paolo Bianchini
-Initial support for Slovak, including new rules by Zdenko Podobný
-Translation of messages into Slovak by Zdenko Podobný
-Small fixes and additions for Polish and English. Small dictionary
update for English.
-Now sentences written in different languages in
are recognized as such and checked properly (previously it was
assumed that the whole paragraph has always the same language).
-New mode of operation on command-line: you can apply suggestions
automatically by supplying "--apply" switch (or "-a" for short).
LanguageTool will then output a text with suggestions applied,
while suppressing all other text (you can, however, still get
verbose output on STDERR). This way LanguageTool may be used to
automatically edit a text (to clean a corpus before some other
operations, for example). Note: only the first suggestion is
applied in the automatic mode, so make sure that the first
suggestion is the best one.
-The command-line tool works now with STDIN. STDIN is activated
anytime a filename is not specified or if you specified "-"
-Update of morfologik-stemming library to version 1.1.4
fixes a NullPointerException (sf bug #2613574) and some
problems with UTF-8 dictionaries.
-Now it's possible to test disambiguation XML rules in a similar
manner that error-matching rules are tested. Look in English
disambiguation rules for examples (at the end of the file).
-Taggers recognize now mistyped lowercase words that should be
-Fixed a bug that disallowed using rules that were turned off by
-Added more tests and fixed some position errors in the XML rules and
disambiguation rules.
-Fixed some glitches with line numbers, time statistics and
match numbers, and unknown words when checking large files
on the command line. Fixed a bug with case conversion in rules
that contained element <match> in suggestions and numerical
references such as \1 in the message.
0.9.7 (2009-03-03)
-Small updates to Polish disambiguation rules, German
error rules, and English dictionary
-Updates for Dutch error rules
-Fixed bugs in Russian rules (Sourceforge bug #2545548 and #2545525),
thanks to Yakov.
-Preliminary support for Romanian (thanks to Ionuţ Păduraru)
-Fixed a ClassCastException crash that occurred with non-English
texts that contain soft-hyphens.
-Fixed bug with no errors visible in context menu for some languages
in, for example French.
0.9.6 (2009-01-21)
-Support for new Proofreading API in OpenOffice 3.0.1. The
non-native dialog box used for checking the text in LT was
removed as it duplicated the native function and was buggy
(see for example sf bug #1943695).
-The toolbar button became redundant and was removed. A new
menu command "Recheck document" was added to start a new check
and reset the rules set as ignored during the previous check
(the rules disabled in the configuration dialog are not
enabled on recheck).
-Added possibility to specify feature unification (introduced
in XML files but the implemented mechanism is general).
-Added new features to the rule disambiguator, such as unification,
filtering, deleting and adding interpretations. Fixed some
small bugs.
-Added a tagger dictionary for Russian (thanks to Yakov).
-Added compounds rule for English and Polish, and made the
existing compound rule easily extensible for new languages.
-Added possibility to add shorter comments for rules to be
displayed in 3.x context menus.
-New command-line option -t or --taggeronly can be used to
tag a text file with POS tags and lemmas as seen by LanguageTool.
This means that you can use LanguageTool as a standalone
tagger for testing or for building a corpus.
-The rule that checks for whitespace surrounding punctuation
marks now detects an orphaned full stop as requested by OOo
users (
-Added a possibility to check if there is a whitespace before a
token in XML rules (simply specify attribute spacebefore="yes";
possible values are also "no" and "ignore". "ignore" is the
-Added new English, Dutch, Polish and Russian rules. Fixed false
alarms for many English rules (including 'a vs. an' rule) based
on the user votes and
tracker reports (#2319686, #2319670, #2381522). Fixed false
alarms for many Dutch (thanks to R.J. Baars) and Polish rules.
-Massive corrections and additions to the English tagger lexicon. Added
a lot of new disambiguation rules for English.
-Massive corrections to the Dutch tagger and synthesizer lexicon (thanks
to R.J. Baars).
-Corrections to the Polish tagger and synthesizer lexicon. Added
rule-based disambiguator for Polish.
-The scripts testrules.bat and are now also testing
the validity of XML files, including false friends, rules and
disambiguator rules.
-Fixed problems with uppercase sentence start rule with lists for Russian
and Polish (thanks to Yakov Reztsov).
-Fixed a bug in synthesizer suggestions generated for indeterminate
articles before compound words (such as 'all-inclusive').
-Fixed a bug that could cause errors in XML error/disambiguation rule
-Fixed a bug that could cripple case conversion of tokens matched in
suggestions formatted with regular expressions.
-Fixed a bug that resulted from missing synchronization of code used
for doing actual grammar-check in (see sf bugs #2416655,
#2411768, #2379365).
-Fixed other small bugs (rule numbering in text output; archaic words
offered in some English suggestions; ignoring some exceptions that
were set for preceding tokens; incorrect positions of errors on some
tagged punctuation tokens; incorrect formatting of some synthesized
suggestions; ignoring manual hyphenation marks during checking in; not printing unknown words on command-line).
0.9.5 (2008-11-02)
-Fixed a bug in whitespace rule that caused LT to crash with English text.
-Fixed a bug that caused some languages to be wrongly
tagged in Turkish Unicode locale (reported by Serkan Kaba).
-Updated, fixed and added some Polish rules.
-Updated and added some English rules.
-Fixes to the German tagger and its compounds splitting
algorithm which could cause false alarms.
-Added full support for various Unicode whitespace
characters and tabs as word separators for English.
-Small fixes with a missing icon in the Extensions Manager
in and formatting of comments visible in the
context menu.
-Fixes in Polish sentence tokenizer that allow some abbreviations
and symbols at the end of sentences.
-Fixed sentence tokenizer: the footnote at the end of the sentence
caused LT not to break the sentence at all.
0.9.4 (2008-09-30)
-Fixed freezing on startup with 3.0.
-Updated Swedish rules, added false friends, and
fixed the compounds rule (thanks to Niklas Johansson).
-Updated French dictionary, disambiguation rules
and grammar rules (thanks to Camille Moulin).
-Added abbreviations for Russian sentence tokenizer and
Russian compound rule. Updated Russian rules (thanks
to Yakov Reztsov).
-LanguageTool button in is now placed next to
the "Spelling and Grammar" button.
-It is now possible to turn the whole category of rules
by default off (in the XML file).
0.9.3 (2008-08-06)
-Works with 3.0 beta m26 or later.
LanguageTool.jar needed to be removed from classpath.
Added support for the new grammar checking API (displays
blue wavy lines for errors automatically, a context
menu with suggestions, rechecks document when options
changed...). With the new API, it is possible to check
documents with multiple languages. When using the blue
wavy lines, all known previous bugs with
were solved; note that 3.0 "Spell and Grammar
Check" dialog is *not* completely implemented yet.
Sometimes you need to use our slightly buggy dialog instead
(progress dialog is always on top to solve focus problems).
-Added full support for various Unicode whitespace
characters as word separators (including typographic
space characters).
-Added support for Russian (thanks to Yakov Reztsov).
-Fixed the bug that caused that "&" was shown in buttons,
and not interpreted as shortcut.
-Fixed menu and toolbar icon which were broken
during compilation (black square or empty).
-Fixed a small bug with formatting some case-sensitive
-Added a generic rule that detects redundant whitespace.
-Added Dutch synthesizer for making suggestions based
on part-of-speech changes.
-Added a rule-based disambiguator for English to minimize
false alarms.
-New Dutch, English (dealing with agreement and redundancy),
Polish and Swedish rules. Added more false friends. Removed
some false alarms.
-The German tagger now splits compounds words correctly which
avoids several mistaggings and thus false alarms.
-Added new Swedish tagger for more complex rules, thanks to
Niklas Johansson.
-The command line client now prints a number after the
rule id so you can see which of the rules in a rule group
it refers to, for example:
"2.) Line 1, column 43, Rule ID: MUEHLHEIM_A_D_RUHR[1]"
0.9.2 (2008-02-17)
-LanguageTool is now part of the "Tools" menu in and its toolbar button is docked to
standard toolbar by default.
-LanuageTool's configuration is now saved
in the user's home directory.
-Added preliminary support for Swedish (thanks to Niklas
-More false friends, especially Polish-English, and
generalization of existing false friend rules for English
and German.
-Added possibility to display a false friend list as
a dictionary formatted in HTML (rules/print-ff.xsl).
-Added an attribute to the DTD so XML rules and groups
of rules can be disabled by default by specifying
default="off". Java rules can be set as turned off by
default by using setDefaultOff() method.
Both settings can be overridden in the Options dialog
-Added possibility to set hot keys for UI items in
translation. Corrected French translation (thanks to
Hugo Voisard).
-Fixed bug in OOo extension that caused that false friend
rules did not display in the configuration dialog.
-Added the possibility to disable complete rule groups
(categories) in the GUI.
-Fixed some Polish rules. Added more Dutch rules and several
hundred more French rules.
-XML rules: added possibility to refer to matched tokens'
POS in <match> elements inside <token>. This is useful
for defining agreement without combinatorial explosion.
-OOo extension will be distributed in .oxt format that
allows automatic updates.
-Internal change: updated tagger library and streamlined
some tagger code.
0.9.1 (2007-09-17)
-Fixed a bug with the resource paths that was discovered
in 2.3 and that made it impossible to use
LanguageTool with 2.3.
-XML rules: enabled referring to whole phrases in suggestions, and
changing their POS tags appropriately. Added support for attribute
scope="previous" in exceptions.
-Fixed error marking problems for XML rules that use phrases
and skipping.
-Added and tweaked some Polish and English rules.
-Fixed problems with badly written XML rules that used skipping.
-LanguageTool now requires Java 6.0 to be compiled, but it
still runs with Java 5.0.
-The icon in the system tray now offers a menu when clicking
the right mouse button (Java 6.0 only).
-Several people had problems with the LanguageTool dialog
appearing in the background (when using LanguageTool inside A workaround has been added that tries
to fix this problem.
-The embedded webserver can now work with UTF-8. Input must
now always be in UTF-8 (or 7-Bit ASCII, as that is a subset of
-Fixed bug (#1726044 in tracking system) concerning the
Italian tagger dictionary that resulted in a badly formatted
dictionary file.
-The stand-alone LanguageTool can now load an XML rule file using
the new menu item "Add language...". This way people can write
rules and use without touching any Java code and without the need
for the rule file to be distributed with LanguageTool.
Currently, the file must be named in the format "rules-xx-name.xml",
where "xx" is the two-character language code and "name" is the full
name of the language in English. Example: rules-en-English.xml
TODO: needs more polishing: i18n, store and load configuration, add
the same to OOo
0.9 (2007-05-20)
-It's possible to edit rules using XML Mind XML Editor (xxe) in a
WYSYWIM (what you see is what you mean) mode. Simply open the file
in xxe, and it should display in a form input mode. It is recommended
that you uncheck Options > Preferences > Save > Add open lines. Set
a high value in Max line length.
-Now you can simply open rule files (rules/<language>/grammar.xml) with
Firefox to see examples of errors and descriptions of rules in a
human-readable format.
-A lot of new French rules by Agnes Souque (mostly grammar errors) and
Hugo Voisard (mostly stylistic errors).
-Everyone can now develop, modify, and test grammar rules without requiring
the LanguageTool source code. Just call this command, it will run all
tests in all grammar.xml files (tests are the sentences in the
<example>...</example> tags). If you work in Linux, type:
In Windows command line, type:
-German compound nouns are now split so that also compound words
which are not in the dictionary are recognized. This way more
agreement errors can be detected, e.g. "die Donaudampfschiff".
The dictionary for jWordSplitter was created using the new class
-New option -u or --list-unknown for the command line version: lists
all words that are not known, i.e., words without POS information.
This currently also lists words written with an uppercase first
letter unless they appear at the beginning of a sentence.
-Added support for variables in XML pattern rules and disambiguation
rules; it is now possible to refer to previously matched tokens using
<match> element.
-Added support for logical AND operation for combining tokens
in error pattern and disambiguation rules (XML-based). In other
words, all possible logical combinations are now available in
the rule declaration.
-Added support for XML-based rule disambiguation of POS tags
(based on an idea by Agnes Souque).
-Added support for regular expressions and POS tags to
create complex suggestions (for example, suggesting another
grammatical case or a word without an apostrophe). Suggestions
based on selecting POS tags require a special synthesis dictionary,
compiled for the language. Currently, two such dictionaries are
available: for English and Polish.
-Added initial support for Slovenian (Martin Srebotnjak)
-Fixed and refined some faulty suggestions for Polish, English, and
Spanish rules.
-Added special POS tag value "UNKNOWN" that matches any token without
any part of speech token. Useful for matching proper names
or misspelled words that are not in the dictionary.
-The tray icon should now also work on the Mac.
-Added option to test suggested replacements in internal quality
-Refined and added more Polish and English rules with new suggestion
-Fixed bug with saving preference for mother tongue (for false
friend checking) in the stand-alone GUI and OOo extension.
-Fixed bug for users that have Windows username with special
characters; extension can now read its files
-Internal change: changed source directory structure; now files
are accessed based on the classpath and not file structure
(more portable way).
0.8.9 (2007-04-07)
-Security: The embedded webserver didn't escape HTML when an error
message was returned which led to the possibility to inject
Javascript via cross site scripting (thanks to "sumit")
-A lot of new Polish rules
-Improved German rules to give less false alarms
-Improved English rules
-Added initial support for Czech (by Jozef Licko)
-Added initial support for Ukrainian (by Andriy Rysin)
-Make integration work with NeoOffice (by Ray Hazlip)
-Added disambiguator interface used to filter POS tags (by Jozef Licko)
-Enabled use of much more compact dictionaries for languages with
heavy use of infixes or prefixes (mostly Slavic languages).
-Fixed a bug with suggestions for some rules that match correlated
words in the sentence, and a bug with matching in some of them.
-New UnpairedQuotesBracketsRule (for pairing (), [], {}, „”, »«,
depending on the language); works on a paragraph-level. Additionally,
new mechanisms for paragraph- and text-level rules were introduced.
-New French punctuation rule for inserting a non-breakable
space before '!', '?', ':', and ';' and after or before
French quotation marks ('«' and '»').
-Fixed crash when started in an unsupported locale
-Added support for declaring phrases (or chunks) in rules,
which allows reusing same structures in various rules. This
could be thought of as simplified chunking.
-The command line version (LanguageTool.jar) can now be started
from any directory. In previous versions, you had to be in the
LanguageTool installation directory to start it.
-Internal change: rules are loaded dynamically to make adding
a new language easier in the future (patch by Andriy Rysin)
0.8.8 (2007-01-30)
-XML output of HTTP interface changed for compatibility with
An Gramadoir (, see
resource/api-output.dtd for the new format. Fixed some encoding
problems with the HTTP response.
-Added --api option to Main method so the XML output can be tested
on the command line.
-Improved English, Dutch and Polish rules
-Fixed several false alarms in German agreement rule (e.g., for "Dies liegt
aller Kommunikation zugrunde." and "Der Vorgang, der soziale Akteure
miteinander verknüpft.")
-Improved German coherency rule, e.g. using "Ketchups" and "Ketschups"
in one document will lead to an error
-German part-of-speech information can now be added to resource/de/added.txt
-Slightly improved suggestions of German compound rule, e.g., "x mal"
will suggest "x-mal", not "xmal"
-Fixed false alarm in German "wieder vs. wider" rule
-Command line version now stops with error message if unknown options
are used
0.8.7 (2006-12-31)
-Extended and improved English, Polish, and Dutch rules
-Added some Lithuanian rules (Thanks to Mantas Kriaučiūnas)
-Improved UppercaseSentenceStartRule for sentences that start with a
quote character
-Bugfix: Small fixes for German compound rule (e.g., "New Yorker" and
"New-Yorker" are both okay)
-Bugfix: Don't crash but show error if LanguageTool is used with Java <= 1.4
-Bugfix: LanguageTool stopped with a MissingResourceException when started
in a locale that's not supported (e.g., fr_CA = French, Canada)
-Added French translation of the GUI (Thanks to Hugo Voisard)
0.8.6 (2006-12-17)
-Added support for Dutch (Thanks to Ruud J. Baars)
-Added initial support for Lithuanian (no XML rules yet)
-Error rules slightly improved for English, German, and Polish
-Small improvements to the sentence tokenizer
-Command line version and stand-alone version now print suggested fixes
-New rule for German that complains about the spelling of compound words
that are written as separate words, e.g., for "System Administrator"
LanguageTool suggests "System-Administrator" and "Systemadministrator".
See resource/de/compounds.txt for a list of words that this rule detects. integration:
-fixed that some errors messages (e.g., Language not supported) could freeze
the GUI
-the "server mode" configuration is now only available in a stand-alone
mode, as it doesn't make sense inside
-configuration dialog is now mostly translated
-error marking corrected for documents that contain index markers
-improved icon
-Stand-alone GUI:
-improved tray icon
-added shortcuts for its menu items
-new menu item "Check Text in Clipboard"
-will now hide itself to the system tray when
clicking the window close button when started in system tray mode
(with -t or --tray). Use "Quit" from the menu to exit the program
-user interface is now mostly translated
-Debugging output (-v) of German texts has been improved, it now shows the
part-of-speech tags as they are used during rule matching
0.8.5 (2006-10-30)
-Fixed small bugs connected with special HTML characters used
internally and entities encoding integration: documents > 64K can now be checked
WARNING: There's still a problem with documents that contain tables,
the wrong part of the text will be selected. integration: progress dialog now really shows the progress
-Introduced new POS tags for uncountable nouns (NN:U) and for nouns
that can be used as uncountable (NN:UN). Added most common proper names
to the English tagger dictionary.
-LanguageTool can now run as a server that returns its results
in a simple XML format via HTTP. Activate the server using
"File -> Options..." or start it via command line using
java -cp jaminid.jar:LanguageTool.jar de.danielnaber.languagetool.server.HTTPServer
-By default, a single linebreak is not supposed to be a paragraph
delimiter anymore (this was introduced in 0.8.4). With the command
line application, you can use the new option -b to make LanguageTool
take single linebreaks as paragraphs delimiters again. Two consecutive
linebreaks are always considered a paragraph boundary (and thus a
sentence boundary).
-Added initial French, Spanish and Italian support
-Added new rules for English and Polish
0.8.4 (2006-08-20)
-Added several new rules for English, German, and Polish
-The false friend rules now offer a possible correction
-Stand-alone GUI: the program can now be docked to the system tray.
Selecting any text and clicking on the icon in the system tray pops up
the main window and checks the selected text. You need to unzip for this feature to work (this was necessary so, which works on the same ZIP, doesn't get confused).
-Stand-alone GUI: now stores its configuration to a file ".languagetool.cfg"
in the user's home directory. Still buggy, as it only stores the configuration
for one text language. integration: cursor doesn't jump to start anymore before it
selects the error integration: LanguageTool's suggestions can now be accepted
using a double click
-A single line break in texts will now be interpreted as the end of a
paragraph thus also as the end of a sentence.
-Reduced size of LanguageTool by using a more compact
part-of-speech tagger
-Changed DTD to support more XML-like rule encoding, added categories
-Extended exception matching to multiple tokens, added more flexible ways
to match variable word order languages (via skipping tokens in rules)
-Added lemmatization support, regular expression support
-Added AnalyzedTokenReadings class to support multiple readings
in all taggers, changed the code appropriately
-Some API changes
0.8.3 (2006-06-25)
-Requires Java 1.5 now
-Support for Polish (Marcin Milkowski)
-Fixed a NullPointerException in DoublePunctuationRule
-Added information to the README about how to compile LanguageTool for
.NET using IKVM The selection of the potentially incorrect text was wrong
after paragraphs (on Windows only)
0.8.2 (2006-02-26)
-New menu entry "Configuration..."
-If you're a native speaker of German and write a text in English, LanguageTool
can now warn you if you're using a "false friend", i.e., an English word that sounds
like a German word but has a different meaning (e.g., become <-> bekommen).
This also works vice versa (i.e., English native speakers will get a warning
when writing a German text using a false friend).
-Added a few new pattern rules
-Missing whitespace after a comma is now detected
-Added a new German rule that detects missing agreement in e.g.,
"meine Auto" (correct: "mein Auto"). This also introduces some new
false alarms.
-Errors for German text are now displayed in German
-Several other small fixes
-Only relevant for developers: changes to rules.dtd (affect "lang" attribute)
-Only relevant for developers: XML validation doesn't happen at runtime
anymore but only when "ant test" is called
0.8.1 (2005-12-22)
-Made it work for cases where the add-on installation path contains
spaces (often seems to be the case under Windows)
-Added a few rules for English
-Added more exceptions so the English "a vs. an" rule works better
-Other small fixes
0.8 (2005-12-17)
-Added integration. Install the LanguageTool ZIP via
Tools -> Package Manager -> Add..., re-start and you'll
have a new menu item "LanguageTool" which checks your text.
-The German data is now more compact (but still 40MB when uncompressed)
and it's included in the standard distribution
-Fixed morphology data for a few German words
-Added some new simple pattern rules for English and German
-Switched to new English part-of-speech data and new version of
part-of-speech tagger (OpenNLP-tools 1.3)
-Files are called LanguageTool instead of JLanguageTool again
as the Java version is the only maintained version anyway
0.7.2 (2005-11-21)
-several new pattern rules for typical German and English typos
-new rule that checks for ".." and ",,"
-new CaseRule for German text: complains if non-nouns are written
in uppercase and checks uppercase spelling of "substantivierte Verben",
e.g., "Das Laufen fällt mir leicht."
-extension of pattern synatx: DT^foo|bar will match words tagged
as "DT", except the word is "foo" or "bar"
-fixed several false alarms (i.e., rules matching text which is
actually correct)
-fixed error marking for matches of pattern rules
-fixed morphology data for these German words: Papagei, Virus, Stasi
0.7.1 (2005-09-08)
-Rules can now be enabled/disabled in GUI
-the error displayed to the user is now taken from the "message"
element, not from the "name" attribute anymore
-implemented mark_from and mark_to attributes, these can
be used to expand/shrink the area of text that is marked
as wrong. Both attributes default to 0 so that the text
marked as wrong is exactly the one that matches the pattern.
-back references like \1 can now be used in error messages
to refer to the parts of the original text that is matched
by the pattern
-a few German noun phrases like "der große Mann" are now checked
for agreement. For example, "des großer Mannes" is detected
as incorrect. This requires data in rules/de/categories which needs
to be downloaded separately.
-a new rule checks whether a sentence starts with an uppercase letter
0.7 (2005-08-15)
-first release of Java rewrite