Back Translation data

Question

Hello all,
Some time ago I asked whether you, or someone you know, has a Paratext project with a Back Translation also in Paratext. Thank you to everyone who has responded. (We’re still open to more examples if you know of any).

Do any of you know of a complete Bible that exists with a Back Translation? It isn’t important whether or not they are both in Paratext. There are so few that any format would be a great resource to have for our work.

If either the Bible or the Back Translation are incomplete that would still be very helpful.
Feel free to email me at [Email Removed] if you know of such data.

Very many thanks,
dcb .

[Email Removed]

Answer 1

Dear dcb,

Every recent translation project of which I am aware has done a back translation in Paratext. There must be hundreds in existence. I can only guess the reason why you have so few is that others, like me, were not completely clear on what exactly you are looking for and your purpose.

Are you looking for BTs in English only? Is it necessary that the translators used the Paratext interlinearizer, or is a back translation in a separate project sufficient? Are you looking for a particular quality of back translation? Do you need the vernacular translation too? Is a Paratext backup fine, or do you need to be added to the project? How will these be used? Will you be able to protect sensitive work?

If you need access to a large number of back translations, it would seem to me that you’d have better luck getting them directly from the Paratext servers or from SIL’s REAP.

Blessings,
LivingField

Comments

Hello LivingField,

Thank you for this, it’s really helpful advice, to make our reasons for asking clear and answer all the other questions that naturally arise.

We are a small team from SIL International’s Language Software development department. We are investigating whether advances in technologies could assist translators. We are working with other groups including UBS and it is possible that our work could be used to improve Paratext and Scripture Forge in the future. Those possibilites are a fair distance in the future. We will also need to find teams who are happy to work with us and test out some ideas to see whether they are useful.

• We are happy to receive projects with Back Translations in any language.

• We need the vernacular translation too.

• It is very helpful if we can chat with the team to have an idea of the quality of the Back Translation.

• It is possible to automatically filter out some data that the machine can see isn’t a good match - for example if there are missing verses, vernacular verses in the Back Translation, or English verses in the vernacular, or if there’s a large discrepancy in the length between them.

• It isn’t necessary for the team to have used the Paratext interlinearizer. If they have it would be helpful to know, which I imagine we can see if we can see the project.

• Being added as observers to the project is ideal please add [Email Removed] and [Email Removed].

• Being added as observers isn’t essential and a backup of the project is equally useful.

• We will use these as data for machine learning experiments. Currently we are storing the data on a Corporate Google Drive that is shared with the five team members and a manager.

• We are avoiding sensitive projects at present.

• There are very few Back Translations available as far as I can tell. I think we in SIL have treated them as a by-product of the translation process. So they haven’t been registered or archived very often and they aren’t always kept up-to-date with changes in the vernacular translation. Any help finding and accessing more data would be greatfully received. If I search for Back Translation in REAP there are a total of 27 items. Only two are actually back translations of scripture, one of John in Angave and the other of Philippians in Fasu. There is no way to download either from the site.

• We’re also interested in projects that have used a front translation.

• We’re interested in projects that are adapting a translation to a related language.

Thanks for your questions! I hope these help to clarify our aims and what I’m requesting.
All the best,
dcb.

Answer 2

Are you interested in daughter projects? I’m serving a cluster that has one source, three daughters, and two granddaughters. [Email Removed]

Answer 3

Hello anon042033,

YES! That would be a very helpful resource for us because of the rich web of relationships between the translations.
Very many thanks,
dcb