polcnv is designed to convert files between different encoding methods used for Polish texts. It can be also used to covert plain text documents in any language using supported character encoding methods. The program uses ISO-10646 UCS-4 (equivalent to Unicode UTF-32) as internal representation.
Linguaphile is a simple command line language translator. It is open source, platform independent, and programmed in Perl. Linguaphile currently supports the following languages: Afrikaans, Alawa, Albanian, Arrernte, Basque, Belarusian, Bulgarian, Catalan, Croatian, Czech, Danish, Dutch, English, Esperanto, Estonian, Finnish, French, Galician, German, Greek, Hawaiian, Hungarian, Icelandic, Indonesian, Interlingua, Irish, Italian, Kala Lagaw Ya, Korean, Kriol, Latvian, Lithuanian, Malay, Maltese, Maori, Norwegian, Pitjantjatjara, Polish, Portuguese, Romanian, Russian, Samoan, Serbian, Slovak, Slovenian, Spanish, Swahili, Swedish, Thai, Tok Pisin, Turkish, Ukrainian, Warlpiri, and Welsh. The Spanish to English translation is the most useful at this stage.
Respell converts English text between the American, British, and Canadian spelling conventions. It prompts the user for cases where more than one target spelling could be chosen for a source word. It can also create a 'universal' spelling which can be automatically converted to any of the three without loss of information.
Marko is a simple toolset that allows you to create markov chain databases of a corpus (or two) of text and then allows you to compare unknown texts to these databases. For any two marko databases you can calculate the probability that the unknown body is related to one over the other. Possible applications include intelligent mail filtering, plagiarism detection, and historical research.
dbacl is a digramic Bayesian text classifier. Given some text, it calculates the posterior probabilities that the input resembles one of any number of previously learned document collections. It can be used to sort incoming email into arbitrary categories such as spam, work, and play, or simply to distinguish an English text from a French text. It fully supports international character sets, and uses sophisticated statistical models based on the Maximum Entropy Principle.
The GNU Talk Filters are filter programs that convert ordinary English text into text that mimics a stereotyped or otherwise humorous dialect. Some of these filters have been in the public domain for many years, but here they are provided as a single integrated package. The filters include austro, b1ff, brooklyn, chef, cockney, drawl, dubya, fudd, funetak, jethro, jive, kraut, pansy, pirate, postmodern, redneck, valspeak, and warez. This package provides the filters both as individual executables and collectively as a C library, so they can be easily embedded in other programs.