Version at: 12/09/2013, 03:32

The Tatoeba corpus is stored as a set of sentences and a set of links between them. Whenever a sentence is translated, a link is automatically created between the source and the translation. Sometimes, however, a translation already exists in the corpus. In this case, the ideal thing to do is to link to that translation.

Since linking indiscriminately can be destructive, the privilege to create and destroy links is reserved to advanced contributors, corpus maintainers, and admins. Once a contributor is promoted to an advanced contributor, he or she will begin seeing chain and scissor icons. Clicking on the chain icon links a pair of sentences, and clicking on the scissors icon unlinks it. This is quite straightforward. However, the chain icon is only displayed for indirect translations. *(For a description of indirect translations, see "Indirect Translations" below.)* If a sentence is a valid translation for a newer sentence, but there is no chain connecting them, you will not see them on the same page, so you cannot click on a chain icon to connect them. Since the standard Tatoeba interface does not provide a way to link "unconnected sentences" (sentences that are neither directly nor indirectly linked), you must turn to other means.

##Use a Bookmarklet to Link Unconnected Sentences
The recommended way to link unconnected sentences is to install a bookmarklet on your bookmarks toolbar and then use it. This works for all browsers. The [bookmarklet](bookmarklets) page gives instructions for installation, along with a sample workflow for using the bookmarklet to link unconnected sentences.

##Use a User Script to Link Unconnected Sentences
A more problematic way to link unconnected sentences is to install a [user script](user-scripts) on your bookmarks toolbar and then use it. User scripts do not work on all browsers, they require installation of a separate add-on (e.g., Greasemonkey for Firefox), and they can cause mysterious bugs.

##Manually Type a URL to Link Unconnected Sentences
Finally, two sentences can be linked by manually typing a URL in the address bar of the browser (namely, "http://tatoeba.org/eng/links/add/"+*first_sentence_id*+"/"+*last_sentence_id*). However, this is error-prone and laborious.
##Duplicating as an Alternative to Linking
An alternative to linking to an existing sentence is to create a duplicate version of the sentence and wait for the periodically executed deduplication script to unify the sentences. This is less desirable since merging two sentences, their existing links, and their existing comment threads may create technical problems and confusion, and because the script must be run manually. For this reason, regular contributors are encouraged to leave comments asking advanced contributors to create links, rather than to create duplicate sentences.

##Indirect Translations
An indirect translation is a pair of sentences that are not directly linked, but have an end-to-end series of direct translations between them. Often indirect translations can be made into direct translations, but there are many exceptions. Sometimes all the direct translations in a series are valid but the ends of the chain should not be linked. This may happen when some languages in the sentence chain make a distinction between, say, verb forms based on gender, while others collapse those differences. (It may also happen when there is a bad link somewhere in the chain.)

version at: 12/09/2013, 03:34

The Tatoeba corpus is stored as a set of sentences and a set of links between them. Whenever a sentence is translated, a link is automatically created between the source and the translation. Sometimes, however, a translation already exists in the corpus. In this case, the ideal thing to do is to link to that translation.

The privilege to create and destroy links is not granted to regular contributors. Once promoted to an advanced contributor, he or she will begin seeing chain and scissor icons. Clicking on the chain icon links a pair of sentences, and clicking on the scissors icon unlinks it. However, this is only possible for indirect translations. *(For a description of indirect translations, see "Indirect Translations" below.)* If a sentence is a valid translation for a newer sentence, but there is no chain connecting them, you will not see them on the same page, so you cannot click on a chain icon to connect them. Since the standard Tatoeba interface does not provide a way to link "unconnected sentences" (sentences that are neither directly nor indirectly linked), you must turn to other means.

##Use a Bookmarklet to Link Unconnected Sentences
The recommended way to link unconnected sentences is to install a bookmarklet on your bookmarks toolbar and then use it. This works for all browsers. The [bookmarklet](bookmarklets) page gives instructions for installation, along with a sample workflow for using the bookmarklet to link unconnected sentences.

##Use a User Script to Link Unconnected Sentences
A more problematic way to link unconnected sentences is to install a [user script](user-scripts) on your bookmarks toolbar and then use it. User scripts do not work on all browsers, they require installation of a separate add-on (e.g., Greasemonkey for Firefox), and they can cause mysterious bugs.

##Manually Type a URL to Link Unconnected Sentences
Finally, two sentences can be linked by manually typing a URL in the address bar of the browser (namely, "http://tatoeba.org/eng/links/add/"+*first_sentence_id*+"/"+*last_sentence_id*). However, this is error-prone and laborious.
##Duplicating as an Alternative to Linking
An alternative to linking to an existing sentence is to create a duplicate version of the sentence and wait for the periodically executed deduplication script to unify the sentences. This is less desirable since merging two sentences, their existing links, and their existing comment threads may create technical problems and confusion, and because the script must be run manually. For this reason, regular contributors are encouraged to leave comments asking advanced contributors to create links, rather than to create duplicate sentences.

##Indirect Translations
An indirect translation is a pair of sentences that are not directly linked, but have an end-to-end series of direct translations between them. Often indirect translations can be made into direct translations, but there are many exceptions. Sometimes all the direct translations in a series are valid but the ends of the chain should not be linked. This may happen when some languages in the sentence chain make a distinction between, say, verb forms based on gender, while others collapse those differences. (It may also happen when there is a bad link somewhere in the chain.)

Note

The lines in green are the lines that have been added in the new version. The lines in red are those that have been removed.