P1. Impact Case ​

Draft Impact Case ​

Educational and research materials which are only available in English are inaccessible to non-English speakers. Overcoming this language barrier is paramount to the democratization of education and acceleration of research.

We will address this barrier by developing a platform where English speakers can crowdsource translations for educational and research materials requested by non-English speakers.

We believe this app is a good solution because there currently does not exist such a platform where users can accurately translate materials or make formal requests for a specific material to be translated. For example, some documents such as research papers contain specialized or technical jargon which Google Translate cannot accurately contextualize or transcribe. Crowdsourcing is a valuable mechanism in this case as experts fluent in both languages can contribute accurate translations. Additionally, once these materials are translated, they become part of a broader, online collection of educational resources available for speakers of that language. This application thus acts both as a crowdsourced translation service and designated location for translated educational materials.

We could measure the number of translations being provided, the breadth of languages being requested, and quality of translations to prove that this is a good solution to overcoming language barriers in education.

Potential Interview Roles and Interviewees ​

Research Findings ​

Industry Translation

• Multiple large translation companies for authors looking to publish (search: translation services for academics), all paid
• Ubuntu, Google, and Facebook have all used crowdsourcing strategies to translate their software
• American Language Experts: leading translation company for academics looking to publish their research
  • match author with expert translator in specific field of study (medical research, humanities, social sciences)
• Crowdsourcing using Transifex
  • Transifex: SaaS, management system and web-based translation platform
  • Workflow: A project owner creates a project in Transifex. The person creates a translation team or appoints maintainers to create the teams instead of them. The project owner or maintainer then uploads the translatable content to Transifex. Then, the translation teams begin the translation process. Once the content is translated, the project owner can download it manually or pull it using the Transifex command-line tool
  • Translation team structure: translator, reviewer, language coordinator
  • recognizes top contributors, can host translate-a-thons
  • Speed in community translation is a byproduct of a good moderation program

Public Education Translation Resources

• Cambridge Public Schools provides language and interpretation resources
  • state-contracted vendors
  • interpretation by CPS Bilingual Liaisons for Amharic, Haitian Creole, and Spanish
  • community engagement team
• Boston Public Schools
  • BPS Parent Language Access Services include offering language support to the families of the district by translating and/or interpreting from English (Source Language) to the Target Language (such as Arabic, Cabo Verdiano, Chinese, French, Haitian Creole, Portuguese, Somali, Spanish, Vietnamese, etc.) of the parent/guardian in question.
  • Interpretation requests should be submitted a minimum of 2 weeks prior to the date of service.

Language Gap in Scientific Research

• Universities have individual translation services for their faculty and students but there is no centralized database for translated scientific work
  • Ex: Caltech
• There exists tools that help researchers and students find translated versions of certain articles. However, it may be the case that a certain research paper does not have a version translated to the desired language.
• Solving the Language Gap in Scientific Research
  • “It’s estimated that non-English language research could expand the geographical coverage of biodiversity scientific evidence by 12% to 25% and the number of species covered by 5% to 32%.”
  • _There is a plethora of research in non-English-language papers that gets lost in translation, or is never translated, creating a gap in the global community’s scientific knowledge. As the amount of scientific research grows, so does the gap. _
• Key findings from the translatE group, a research group that conducts research regarding language barriers and its effects in environmental studies research
  • Non-native English speakers need up to twice as much time to read and write papers and prepare presentations in English (Amano et al. 2023. PLOS Biology).
• 10 Recommendations for Overcoming Language Barriers in Science
  • Disseminate research in multiple languages
  • Use scientific knowledge sourced from multiple languages
  • Increase the visibility of non-English-language science
  • Translate scientific terms
  • Provide genuine support to non-native speakers
  • Distinguish language skills from scientific quality
  • Consider language balance in scientific activities
  • Acknowledge efforts to overcome language barriers
  • Be considerate of non-native speakers
  • Make use of existing resources and opportunities
• Studies published in languages other than English are often neglected when research teams conduct systematic reviews.

Interviews ​

Interview Plan: ​

1. Ask for academic background

2. Ask whether they had trouble accessing educational content because it was only in English and not their native language

a) If yes, ask what was the cause, and how they fixed/tried to fix that issue.

b) If no, ask whether they would be willing to help people in that situation by translating English content to other languages.

4. Tell them about the app idea, and see how they could see it being applied to the previously mentioned problems

Interview 1, Berkin ​

The first interviewee was a student at MIT, who studied high school in Turkey. During his time in high school, he participated in physics training camps and competitions, eventually representing Turkey at the International Physics Olympiad.

In the interview, he was asked if he had any trouble finding resources in his native language to study. He said he did not, but he did not usually look for resources in Turkish. He actually preferred English textbooks to their translated counterparts, as his English was already very good even before he started high school.

However he said that he knew many people who found understanding educational content (physics-related or not) in languages other than their native languages very hard, and even though Turkish is a widely used language (80 million speakers), those people had a lot of trouble finding content past the basic introductory level. To combat this, he and his friends worked together to translate a few resources (Mechanics and Thermodynamics Study Guides by Jaan Kalda, each about 40 pages long) related to the physics competitions they were participating in into Turkish. In fact, they tried expanding this endeavor to translate some commonly used textbooks, but these endeavors never reached completion as they were too labor-intensive for the small group (4-5 people) they had.

When asked about whether the app we were developing could have been of use, he was very much on board. He thought, even without the crowdsourcing aspect, this was a great idea, as it completely lifted the pressure of organizing the process, and it also could allow the translation work to be transferred from one generation of students to later generations with ease. About the crowdsourcing part, he also thought that reduce the load greatly, but he thought for very niche translations like the ones they were doing (translating books related to a specific competition) it would be a good idea to allow users to control who gets to contribute translations.

Interview 2, Fatih ​

The second interviewee was a college student studying CS in Turkey. He was originally from a rural town in Central Anatolia, then started attending a boarding high school for science in Istanbul.

During his time in high school, he developed an interest in CS, competitive programming in particular, and even managed to win medals at international competitions. However, throughout this process, he had difficulties finding any resources for CS learning content. His school had no dedicated CS teacher, and almost all the content he could find online was in English. The latter was not a problem for some people, but with his background, he had zero prior exposure to English.

He mostly had to make do with copying text from English sources to Google Translate, but Google Translate consistently had issues: it would translate scientific terms like "complete graph" literally and wouldn't use the proper term, and it created very convoluted translations for compound sentences, and it was just not ideal. It worked fine in simpler contexts, like translating the text of a programming problem, but for other sources, like an article describing a specific algorithm, he usually had to ask for help from other people.

He was very excited about an app where you could see translations for content in your language, and especially one where you could post content you don't understand so that people can translate it, and it also gets used by other people as well. He was lucky enough to have people around him who could help with translations, but not everyone is that lucky.

Interview 3, Laura ​

In this interview, Laura and I discussed language, education, and her relationship with her home country, Brazil. Laura did Physics olympiads all her life, and, coming from a humble background, she explained the structural flaws of teaching Physics in Brazil. Before getting to high school, Laura did a lot of independent learning – she learned English on her own and was thus able to find all the online resources to study Physics. However, she admits that, after being accepted into the more prestigious Physics circles in Brazil, she thought for the first time about how that skill was crucial to her story. That one can learn advanced topics just to a certain extent, and, after that, it’s hard to make progress without English. She said sometimes she would even stumble upon Russian problems, the best out there, and Google Translate didn’t do the job for her, so she imagines how it is for those who don’t even know how to find Russian texts to try to translate. She said this is a societal problem that has to be addressed: it’s not that Brazil should try to translate every scientific breakthrough out there, but that learning is and will be shared in a universal language. This way, Brazil has to better prepare the English language teaching in the schools. While this doesn’t happen – she mentioned it might take at least 30 years –, putting some effort into translating that material and making it available for everyone can be a temporary solution for those who were already damaged by the system.

Interview 4, Yan ​

Yan discussed his language skills and their importance in competitive programming (CP). He learned English in 2019, which helped him understand CP materials better. He acknowledged that the CP community predominantly uses English. He works for an initiative to democratize access to Computer Science called Neps, and there he translated CP materials from Portuguese to English. However, he believes that very few people use these translations and that the translation effort may not make the Neps platform more international.

Regarding Chinese and Polish algorithm papers, Yan mentioned using online translators. He feels that language is a barrier and that the quality of the translation he and his research group create isn’t ideal. He speculates that some countries might want to keep certain algorithmic knowledge to themselves and believes that language barriers work in their favor to promote this secrecy. For him, in an ideal world, countries wouldn’t keep the knowledge to themselves, but that is, unfortunately, our current reality.

VSD Analysis ​

Stakeholders ​

• Direct stakeholders:

  People who submit content to the app so that it could be translated, and people who translate the content shared would be the direct stakeholders for our app.

• Indirect Stakeholders:

  People whose work would be translated would be indirect stakeholders for our app. It is important that the app respects the intellectual property of these indirect stakeholders.

Time ​

• Reappropriation:

  People might want to use this app to not just translate educational/academic content, but other types of useful content they might want to share with others (government forms, literature, etc.) or their own projects they might want to keep private.

• The Long Now:

  A wealth of translated knowledge would be accumulated over generations such that the app could become a living digital library with ‘historic’ scientific literature that people of the future may engage with.

• Work of the Future:

  Our app will change the nature of research and education. In particular, the workers of these sectors will have greater access to a collection of knowledge available in their language. They will also be able to volunteer their expertise and lingual fluency to translate scientific and technical work. This can accelerate the process of research and learning for workers of all language backgrounds. Additionally, the app increases the visibility of non-English-language science and allows researchers and students to use scientific knowledge sourced from multiple languages, fostering a cosmopolitan academic culture.

Pervasiveness ​

• Political realities:

  Our app defies the scientific exclusivity that some countries exercise in order to keep their competitive advantage. However, we believe in international scientific collaboration to accelerate human progression.

Values ​

• Choose Desired Values

  Our app aims to democratize education across language barriers. Our app will also embody cooperation through collaborative translations and inclusivity by welcoming users from diverse languages and academic backgrounds.

Revised Impact Case ​

Educational and research materials that are only available in certain languages are inaccessible to individuals who are not fluent in those languages. Overcoming this language barrier is paramount to the democratization of education and acceleration of research.

We will address this barrier by developing a platform where users can crowdsource translations for educational and research materials requested by other users. The app will have an interface similar to that of “Genius” to allow users to collaboratively and simultaneously annotate documents with translations.

We believe this app is a good solution because there currently does not exist such a platform where users can accurately and collaboratively translate materials or make formal requests for specific material to be translated. For example, some documents such as research papers contain specialized or technical jargon which Google Translate cannot accurately contextualize or transcribe. Crowdsourcing is a valuable mechanism in this case as experts fluent in both languages can contribute accurate translations. Additionally, once these materials are translated, they become part of a broader, online collection of educational resources available for speakers of that language. This application thus acts both as a crowdsourced translation service and a designated location for translated scientific and educational materials.

We could measure the number of translations being provided as a base measure of user interaction. In addition, we can compare the percentages of completed and uncompleted translations at a given time, and how long each translation takes to be completed. To measure reach, we can find the total number of languages being translated. Also, through a system of upvotes, we could measure the quality of translations and have expert approvers verify translations. We could have user surveys within the app, similar to Facebook and Twitter, asking users about how they are using the app and how it has impacted their academic studies.

Pitch ​

https://docs.google.com/presentation/d/1AAJpiqykGv44z53UpQt6GGZArv8FDNc-TJIvyI8pM7A/edit#slide=id.g29835f137ff_0_2