This document provides a high level overview of the Global Storybooks project as well as an outline of our multilingual translation and recording procedures. The procedures and protocols outlined here have been established over the years as we have learned from our experience translating and audio recording 1000+ stories in more than 40 languages.
At heart, the Global Storybooks project offers a streamlined platform, software tools, and a translation / audio recording workflow that allow for the creation of rich multilingual literacy websites with the absolute minimum amount of time and effort. This allows translators to focus on translating, without having to worry about any of the details of website development.
This is partly made possible due to a fixed corpus of stories and a unified website interface that can be translated / localized in parallel to suit a variety of different contexts. However, all of our tools (and content) are entirely open source and open-licensed, so it is also very possible to design a completely new paradigm using the same platform and fill it with different stories or a mix of existing stories and other content. (Please note however that while we may be able to provide some limited support for explaining how this could be done in principle, the latter case is more of a DIY option, and new tools would need to be created independently to support the different paradigm.)
The websites on the Global Storybooks Portal are dedicated to translations of the core corpus of 40 stories from the African Storybook into a variety of languages.
These 40 stories have been carefully selected from the hundreds of stories on the African Storybook website specifically for their universal appeal, pedagogical value, and relevance to children and learners across a wide range of contexts.
The tools, software, and workflows we have developed make it as easy as possible to translate and deploy all of these stories online wherever and whenever there is the linguistic capacity to do so, which means that even with only a few translations to start with, a fully functional multimedia website for any region or group of languages can be made available online in very short order (usually a matter of hours).
All of the translations that we work with have been produced by volunteers from around the world with an interest in promoting global literacy. In other words, none of the stories on any of the Global Storybooks websites have been translated automatically (i.e., via machine translation).
Machine translation, even for well-resourced languages, is still notoriously unreliable, and thus all of the 1000+ translations we have collected thus far have been individually completed by human translators using our online web application, which facilitates the process of formatting, tracking, and submitting the translated texts.
The easiest way to get a sense of what the translation process is like is to dive in and try out using the online app to complete a simple translation. The translator app can be found here.
The stories have been arranged in the application such that the shortest and easiest ones come first, and then get progressively longer and more complicated as one goes along. This makes it is very easy to get started!
Translations of stories for the Global Storybooks websites are accepted exclusively through the online translator app. This ensures that:
- Stories are correctly formatted
- Each translator is credited and attributed properly for their work
- The most recent and up-to-date text is translated
- A target language name has been provided
- All text in the story is translated
- Stories are sectioned and paginated properly
- Titles are correctly identified and translated
- Illustrations are provided beside the text so that translators can make sure that they match
- Completed translations are not duplicated
Instructions and detailed help are available within the app itself by clicking on Help or About in the menu at the top of the page.
- Note: Translators do not have to translate from English -- the translator app can be (and has been) translated into other languages, so that the interface and original story text are presented in e.g., Spanish or Norwegian.
After all translations are complete, a multi-stage quality checking process begins to proofread and then finally record audio of the stories.
Below is an outline of our translation, proofreading, and audio recording workflow.
Audio recording can be a bit more involved and a bit more complicated than one might expect, and we have learned over time that there are some procedures and preparatory measures that are necessary if one wants to produce usable (and reusable) audio at the highest quality.
Although in the past we have experimented with recording in low quality using personal devices such as laptops or mobile phones, we now normally record all audio in a recording studio with a professional audio engineer. The recording process is only undertaken after we have finished translating all 40 stories into a particular language. Audio recording is the culmination of a multiple step process that ensures that the resulting recording is of the best possible quality.
A detailed outline of our quality control process is below:
- Translation: Translation drafts are completed by translators and submitted using the online app.
- They are immediately made available on the website.
- This also facilitates the identification of errors as more people have access to the texts in context (cf. the open source maxim "many eyes make shallow bugs").
- Proofreading: Once all translations are completed, the first round of proofreading commences.
- A bilingual draft is produced containing all 40 stories with the original and translated text side by side.
- Proofreading is collaborative: at least one (and ideally more) proofreaders read through the full draft together with the translator(s) and our editors to find and correct as many errors and inconsistencies as possible.
- Audio Stage 1 - Preparation: Once all proofreaders and translators are satisfied with the final draft, preparation for audio recording begins.
- A final version of the translated text is compiled into an audio script to assist the reader with preparing.
- A date and time for audio recording is established, and the studio and audio engineer are booked.
- Any errors identified in the script during the preparation process are corrected in the final draft.
- Audio Stage 2 - Recording: In the studio, high quality audio is recorded, usually over the course of two sessions.
- This requires approximately 4-5 hours total to record all of the stories, though this is dependent both on the target language and preparedness of the reader.
- During the recording session, errors will be identified by the reader that were not noticed prior to reading the stories out loud. These are corrected on the spot in the audio script.
- Audio Stage 3 - Post-processing: After audio has been recorded, audio post-processing begins.
- The audio files are transferred to our audio editors, who select the best "takes", compare the audio with the script, remove any errors or repetitions, and split each story into multiple segments for the website.
- This rough edit then goes to our audio engineer for final mixing.
- Audio Stage 4 - Publication online: Once the professional mix of the audio is complete, it is then converted and uploaded to the website. The process is now complete!
As can be seen from the above workflow, by the time audio is available on the website, the text will have been proofread and corrected at least 5 times by different people. The goal here is to avoid having to record audio more than once by removing any major errors that would necessitate re-recording.
From our experience we know that, in the end, this process will save an enormous amount of effort and time as compared to translating a few stories, recording, and then repeating the process. Audio recording and editing is "expensive" (in terms of time/effort as well as financially) and ideally should be done only once at the highest quality possible.
For those projects or languages that are unable to record audio in our studio, we also offer support through our audio recording guidelines for obtaining the best possible quality under the circumstances.
This project is released under a Creative Commons Attribution 4.0 Licence.