funds for joint online translation
Disclamer: This is an unfinished article on collaboration tools. You can help the community by correcting and adding to it :)
(the article is published for the aggregation of opinions and online translation experience)
In general, the following points are relevant for joint translation:
On these points, IMHO, and it is worth considering the available means.
In our project, we, without any doubt, used google spreadsheet - with sharing the work file between the participants. (GoogleTalk for translation appeared simultaneously with the beginning of our project). From the experience of this work and write a review of opportunities.
As we translated the subtitles - they were obviously divided into separate lines, and put into a table with columns: timing, source text, translation.
For translation of other text formats, timing, of course, is not relevant.
In this format, sentences are naturally distributed to the table lines.
Perhaps such a presentation would not be very convenient for texts with long sentences, especially for the German language, famous for the possibility of making complex sentences, sometimes stretching over several pages.
')
In the Google table, naturally, all sorts of table formats are imported - csv, ods, xls
The problem point is that importing is possible only in a new, separate file,
it is necessary to transfer text from it manually, through the clipboard, in which no more than one hundred lines fit.
For controversial points, alternative versions of the translation were recorded in the additional column, with the indication of the author and sometimes comments. Alternatives were later discussed in the mailing list.
Also, in the additional column, we left brief comments and explanations for other participants, or explanations for the reasons for choosing one of the options.
It is clear that there are no built-in means of access to dictionaries. Everyone enjoys the fact that it is more convenient.
To create glossary, an additional table sheet was used.
There were decisions on controversial issues (such as how to translate the word Wave and whether to translate at all).
Online coordination in google speradsheet is done at its best.
Only because of this feature alone would it be worthwhile to make a choice in favor of guglotablits.
For all participants who simultaneously opened the file in edit mode, there is a built-in chat.
The participants are painted in different colors, and poke into the name - you can go to the cell of the table where the cursor of the selected participant stands.
It would be superfluous to say that the cursors of all the editors and their movements are visible in real time, they are also painted in different colors, and are signed with the name when the mouse is hovered over it.
For offline coordination, a google mailing group was organized.
Google docs provides the ability to provide access to the file to the whole group, which solves the issue of synchronization of participants. But this opportunity seems to be in beta.
The Google links are convenient because in the comments to the translated fragments you can insert links to the discussion of this fragment in the group.
In our case, for each block there are four states:
In guglotables, the state of each line is easily and simply indicated by the background color of the cell.
We denoted controversial moments with a yellow background to draw attention to them.
Phrases verified in spelling and grammar are green so that the others do not duplicate the work.
To form the final version, one of the participants was chosen as an editor and style specialist.
It was suggested to paint the final version in blue, but this is broken :)
In fact, two people were engaged in the reconciliation of style - and the built-in chat was enough for them to synchronize their actions.
Assess how complete the process, using tables is possible only visually.
But in our case, this moment was not particularly important.
The history of changes in guglottables when sharing is conducted automatically, creating a new version for each change of a particular cell (well, or a block). The author of this change is recorded in the list of versions.
Comparison of versions has been done very poorly - changes are highlighted though, but only the data from the latest version is displayed, and besides, when switching versions, the file is displayed from the beginning, and to see which version the particular cell was changed in, you have to skip each one.
The list of versions itself is displayed as a drop-down menu. This makes it difficult to assess the contribution of each participant.
Two levels of access are possible: reading and editing, and two ways of authentication: without authentication, or by Google account. There are some problems with using an account with a name other than gmail.com.
Additionally, the document may be "published" for viewing in the shared access, in the format of a web page.
(In our case, the working paper was publicly accessible all the time.) The published document can be configured to reflect the current state, or to publish a separate command.
The text is an obvious way, in the form of two columns (or two horizontal frames).
Translation is carried out according to the sentences into which the text is divided automatically.
When you select an offer, the corresponding translation line opens in an inline edit form.
It is impossible to select / select a block by yourself.
For zarzazki and unloading supported document formats: rtf, doc, html, odt.
In addition, an arbitrary url page, or an article from Wikipedia or knol, can be loaded into the system.
Tables are not loaded into the system. Just as the document is not transferred from GoogleDocs.
Adding and editing original text is impossible - before you begin to translate, you must fully prepare the source text.
(so if during the translation process it is necessary to do a transcript by ear - the toolkit is not an assistant in this)
From additional baubles there is an opportunity to attach a translatable block and scroll through the entire text.
You can leave comments for each block of text (but no more than one line).
Several translation options are not provided.
Immediately after loading the text and selecting languages, the machine translates the entire text.
In the additional panel, a dictionary search is available (which is very poor compared to, say, multitran.ru - there are no indications of categories of use, slang, or references to phrases)
In addition to the dictionary, there are tools for organizing glossaries for storing translations of words and “translation memory” for storing translations of fragments.
Both can be shared independently of the working paper itself.
Both can be downloaded from external sources and cannot be edited online.
(not tested)
Apparently, a piece of text can be in two states - the original (automatic translation) and the edited.
Edited is considered translated and the amount of such text is displayed in% in the list of working documents.
No intermediate states are provided, except as an opportunity to designate them in the comments.
There are no change histories either.
There are three levels: viewing, commenting, editing. Authentication, of course, by Google Account.
There are no hints about using the group address.
So, you have to synchronize the list of participants manually.
To be continued:
(the article is published for the aggregation of opinions and online translation experience)
In general, the following points are relevant for joint translation:
- presentation format: breakdown of text into separate blocks - predlozhenya / paragraphs
- editing and importing / exporting tools, alternative translations
- access to the dictionary, automatic translation
- coordination of joint activities
- process status monitoring:
- designation of the status of each translated block (subscript translation, literary, controversial moment, final version, etc.)
- completion indicator
- change history
- provision of access
On these points, IMHO, and it is worth considering the available means.
Google docs
In our project, we, without any doubt, used google spreadsheet - with sharing the work file between the participants. (GoogleTalk for translation appeared simultaneously with the beginning of our project). From the experience of this work and write a review of opportunities.
format
As we translated the subtitles - they were obviously divided into separate lines, and put into a table with columns: timing, source text, translation.
For translation of other text formats, timing, of course, is not relevant.
In this format, sentences are naturally distributed to the table lines.
Perhaps such a presentation would not be very convenient for texts with long sentences, especially for the German language, famous for the possibility of making complex sentences, sometimes stretching over several pages.
')
facilities
In the Google table, naturally, all sorts of table formats are imported - csv, ods, xls
The problem point is that importing is possible only in a new, separate file,
it is necessary to transfer text from it manually, through the clipboard, in which no more than one hundred lines fit.
For controversial points, alternative versions of the translation were recorded in the additional column, with the indication of the author and sometimes comments. Alternatives were later discussed in the mailing list.
Also, in the additional column, we left brief comments and explanations for other participants, or explanations for the reasons for choosing one of the options.
dictionaries
It is clear that there are no built-in means of access to dictionaries. Everyone enjoys the fact that it is more convenient.
To create glossary, an additional table sheet was used.
There were decisions on controversial issues (such as how to translate the word Wave and whether to translate at all).
coordination
Online coordination in google speradsheet is done at its best.
Only because of this feature alone would it be worthwhile to make a choice in favor of guglotablits.
For all participants who simultaneously opened the file in edit mode, there is a built-in chat.
The participants are painted in different colors, and poke into the name - you can go to the cell of the table where the cursor of the selected participant stands.
It would be superfluous to say that the cursors of all the editors and their movements are visible in real time, they are also painted in different colors, and are signed with the name when the mouse is hovered over it.
For offline coordination, a google mailing group was organized.
Google docs provides the ability to provide access to the file to the whole group, which solves the issue of synchronization of participants. But this opportunity seems to be in beta.
The Google links are convenient because in the comments to the translated fragments you can insert links to the discussion of this fragment in the group.
process
In our case, for each block there are four states:
- draft translation (it was suggested to use machine translation, but because of the peculiarities of the speakers' speech, they turned out to be absolutely terrible)
- points of contention
- literary translation
- final version
In guglotables, the state of each line is easily and simply indicated by the background color of the cell.
We denoted controversial moments with a yellow background to draw attention to them.
Phrases verified in spelling and grammar are green so that the others do not duplicate the work.
To form the final version, one of the participants was chosen as an editor and style specialist.
It was suggested to paint the final version in blue, but this is broken :)
In fact, two people were engaged in the reconciliation of style - and the built-in chat was enough for them to synchronize their actions.
Assess how complete the process, using tables is possible only visually.
But in our case, this moment was not particularly important.
The history of changes in guglottables when sharing is conducted automatically, creating a new version for each change of a particular cell (well, or a block). The author of this change is recorded in the list of versions.
Comparison of versions has been done very poorly - changes are highlighted though, but only the data from the latest version is displayed, and besides, when switching versions, the file is displayed from the beginning, and to see which version the particular cell was changed in, you have to skip each one.
The list of versions itself is displayed as a drop-down menu. This makes it difficult to assess the contribution of each participant.
access
Two levels of access are possible: reading and editing, and two ways of authentication: without authentication, or by Google account. There are some problems with using an account with a name other than gmail.com.
Additionally, the document may be "published" for viewing in the shared access, in the format of a web page.
(In our case, the working paper was publicly accessible all the time.) The published document can be configured to reflect the current state, or to publish a separate command.
Google Translation Toolkit
format
The text is an obvious way, in the form of two columns (or two horizontal frames).
Translation is carried out according to the sentences into which the text is divided automatically.
When you select an offer, the corresponding translation line opens in an inline edit form.
It is impossible to select / select a block by yourself.
facilities
For zarzazki and unloading supported document formats: rtf, doc, html, odt.
In addition, an arbitrary url page, or an article from Wikipedia or knol, can be loaded into the system.
Tables are not loaded into the system. Just as the document is not transferred from GoogleDocs.
Adding and editing original text is impossible - before you begin to translate, you must fully prepare the source text.
(so if during the translation process it is necessary to do a transcript by ear - the toolkit is not an assistant in this)
From additional baubles there is an opportunity to attach a translatable block and scroll through the entire text.
You can leave comments for each block of text (but no more than one line).
Several translation options are not provided.
dictionaries
Immediately after loading the text and selecting languages, the machine translates the entire text.
In the additional panel, a dictionary search is available (which is very poor compared to, say, multitran.ru - there are no indications of categories of use, slang, or references to phrases)
In addition to the dictionary, there are tools for organizing glossaries for storing translations of words and “translation memory” for storing translations of fragments.
Both can be shared independently of the working paper itself.
Both can be downloaded from external sources and cannot be edited online.
coordination
(not tested)
process
Apparently, a piece of text can be in two states - the original (automatic translation) and the edited.
Edited is considered translated and the amount of such text is displayed in% in the list of working documents.
No intermediate states are provided, except as an opportunity to designate them in the comments.
There are no change histories either.
access
There are three levels: viewing, commenting, editing. Authentication, of course, by Google Account.
There are no hints about using the group address.
So, you have to synchronize the list of participants manually.
To be continued:
notabenoid
translated.by
dotsub
Source: https://habr.com/ru/post/61755/
All Articles