T2T: Language Access Texting
Case Study Lead(s)
Sahil Chopra, Judy Dunbar, Lily Liu, Belinda Mo, and William Penniman
T2T: Text to Translate
T2T is a translation technology project implemented via text messaging to facilitate communication between legal aid lawyers and their clients, across language barriers.
It was built for non-English-speaking parents of special needs children and the stakeholders in the medical, legal, and education fields who advocate for and represent these families, as independent education plans were developed for the children.
The project was developed by students enrolled in a Stanford University project-based class taught over 20 weeks. The class was cross-listed in the Sociology Department and the Hasso Plattner Institute of Design, and it was open to both graduate students and undergraduates interested in public interest tech work.
T2T moved beyond the class with a pilot in partnership with a local legal aid group, and it is continuing to expand to other partners with support of a legal design research lab at Stanford Law School.
Problem Identified: Language barriers exist between Spanish-speaking parents, and the English-speaking professionals who work with them to develop Individualized Education Programs for their children.
Students in Stanford University’s Justice + Poverty Innovation class were encouraged to work with partners to find problems relating to improving public services in the Bay Area, with an emphasis on improving language access. The T2T team comprised graduate and undergraduate students studying business, computer science, engineering, public health, and design. The team decided, based on team member interest, to focus on the intersection of the medical and legal fields, and thus was urged by the class instructors to collaborate with the Silicon Valley Medical-Legal Partnership to determine where needs existed most. The team conducted site observations at hospitals, interviews with stakeholders, and background research on potential initiatives.
From initial discussions with the Silicon Valley Medical-Legal Partnership, the team found that 80% of the referrals pediatricians made to the Medical-Legal Partnership were for children with special education requirements, where the majority of parents are monolingual Spanish speakers. Through the partnership, the group also learned about the difficulties parents of children with special education needs faced when attempting to secure educational resources for their children.
Notably, monolingual Spanish parents had low levels of participation in the development and maintenance of Individualized Education Programs (IEPs) for their children. IEPs are written plans that are required in the U.S. school system for children to receive special education services under the Individuals with Disabilities Education Act. The team’s user research indicated that the process of securing and maintaining IEPs was particularly cumbersome, lengthy, and frustrating for most parents, and even more so for parents with language access needs.
Research Question Identified: How might we help monolingual Spanish parents effectively advocate for their children with special education needs in the long term?
The class was taught according to a human-centered design process, which encouraged the team to build relationships with a variety of community partners and individuals. This involved interviews, site visits, observations, and other kinds of meetings. Through this design research, the T2T team was able to build relationships with key stakeholders and to glean insights to guide how to develop a public interest tech project to respond to the identified problem.
Legal Advocates for Children and Youth (LACY), as an affiliate of the San Jose-based legal aid nonprofit Law Foundation of Silicon Valley, was partnered with the T2T team and instrumental in fostering interviews with lawyers and doctors. The teaching team of the class had provisionally established this partnership before the class, and the T2T team formalized it around this project. LACY helped to provide a base of lawyers and clients to define necessary use cases, review the proposed solutions in pilots, and provide feedback on ease of use and value.
The Silicon Valley Medical-Legal Partnership was introduced as a partner to the team through faculty contacts, and it provided a valuable source of monolingual clients to interview, as well as translators.
Preliminary Research Insights
Within the first part of the class, the T2T team gathered key insights to guide its product development and theory of change.
Early design work of the T2T team
Two central insights concerned the power of informal advocacy and the promise of text messaging. The T2T team decided to focus on the informal IEP process that parents conducted along with the more formal meetings and communications. From their interviews, the team noticed that many of the English-speaking parents who had gone through the IEP process smoothly had good relationships with their teachers through informal channels, such as text messaging.
According to Sahil, a T2T team member, the team learned that an informal parent-teacher relationship provided a “bond, connection, something to translate into the actual meeting’s effectiveness” by instilling a sense of trust between parents and teachers. The technology of text messaging was particularly suited for relationship and trust-building.
If parents of children with special education needs could develop a regular text message communication with a teacher, the teacher became more inclined and better equipped to act as an informal advocate for the child. Parents could have the reassurance that teachers were updated with relevant information and vice versa, through an open and inexpensive channel of communication.
The team also realized that this bond may be necessary not just with teachers but also with representative lawyers in the IEP process: What if parents could communicate with their lawyers quickly and informally, to build a stronger relationship? The T2T team members worked with their stakeholders to understand whether texting can act as an informal channel of communication between monolingual non-English-speaking parents and monolingual English-speaking lawyers.
To better facilitate an informal parent-teacher relationship for monolingual parents, the team decided to create a text-based translation service, called T2T. The goal of the first prototype was to empower monolingual-English lawyers from the legal aid group LACY to be able to text with their monolingual-Spanish parent clients.
The hypothesis was that the lawyers would be trained to initiate these translation-supported text message conversations. The lawyer would text “Start” to a designated phone number and answer three brief on-boarding questions. The lawyer would receive a message from a new phone number and could reply to it by inputting a message in the language of the lawyer’s choice. This message would be sent to the parent client in both the original language and the desired translation language.
The technology used to accomplish this was familiar to the team, which used Google Translate for text translation, Google Cloud to host the server, and the Twilio API to facilitate automatic texting. To best serve the local population in Santa Clara County, the team began by prototyping Spanish-to-English translation.
Key Insights from the Prototype Testing
Strengths of the team and project became clear in the second half of the course. As team members tested their initial technical prototype with the partners, they learned more about what kind of service design and outreach would best make a technology that both the lawyers and the parents would use.
The T2T team chose to base the service over text because it was a platform already successfully leveraged by parents who spoke the same language as their child’s teacher in the IEP process. According to the team, based on interviews, texting was often used by parents “who had successful relationships with their children’s teacher” in the IEP process. It was also used by many of the monolingual Spanish parents who would be the target user group of the new initiative.
The team decided to implement its solution without requiring separate mobile application be downloaded. The team found that clients at the Silicon Valley Medical-Legal partnership often had limited access to data for internet access, as well as limited access to email and landline phones. This led the team to design a technical solution that worked with existing messaging platforms, rather than a separate app or website.
Despite the team’s best efforts, on reflection there were critical weaknesses that emerged, which the team had to work to address.
While efforts were made to ensure the tool was easy to use, the team realized that it needed to remove some key usability barriers be removed. For example, after testing extensively with lawyers at LACY and the Law Foundation of Silicon Valley, the team received feedback that the on-boarding flow — the messages an individual receives in the setup process of the T2T tool, could be simplified and reduced. However, there are constraints of the native texting interface that make it difficult to present on-boarding information in a clear and efficient fashion.
This challenge of on-boarding is one the team is still working to address, to get a balance of explaining the system sufficiently and not adding too much friction to the initial process. The team has seen that when lawyers and parents were able to overcome initial difficulties in understanding this new tech system, their comfort and usage grew substantially.
The T2T team mentioned that it would have benefited from having a consistent and involved partner in the education system, such as a school district, in order to get feedback and have a wider network which to pilot the project. While team members interviewed 20 teachers school administrators, they said it was “hard to get teacher involvement directly because [the teachers] often needed top-down approval” on any decisions. This limited the initial implementation to lawyer parent communications, but ideally this can expand to a fuller range of advocates, including school administrator/teacher-to-parent communications.
From Class to Pilot
As the team finished the second quarter of the class, two team members decided they wanted to keep working on the T2T implementation. With the support of their instructors and partners, they used the summer break to continue running focus groups and usability testing with their stakeholders. They also established the necessary hand-offs to the partner to allow LACY to own and host the project for a pilot period.
Within several months after the class, the team helped LACY begin a live pilot with its clients. They remained involved in gathering feedback, providing technical assistance, spotting bugs, and talking with the stakeholders about what could be improved in an expanded or replicated version. This pilot is continuing, and the team has been presenting the T2T project to wider groups of possible partners, including at legal aid and court conferences.
The T2T team took a critical approach to the possible outcomes or concerns of the new messaging system. Team members tried to explore and plan for as many of these ethical and values concerns upfront, while also remaining open to unexpected concerns that would arise during the pilot.
The T2T team was concerned over the privacy, and perceived security, of messages sent over the T2T system. Privacy popped up early on when talking with the lawyers at LACY, due to the professional responsibility of lawyers to preserve confidential information contained in communication with clients.
To alleviate such concerns, the T2T team invested in a “privacy by design” approach. Its tool does not store any messages in its database, meaning that even the team members themselves do not have access to the content of the messages being sent through the tool. Instead, a phone number intercepts the message, translates it, and sends it directly to the next person. The legal aid partner also is the owner and administrator of the phone accounts and technical system. The student team provided the legal aid partner the initial system design, guidance on how to use it, and strategies for fixing bugs and new features, and then handed over ownership to the partner.
Because of the sensitivity of information and the limited accuracy of present-day translation services, the T2T team was also concerned about liability if the tool provided inaccurate translations.
The T2T team decided to approach this issue in three ways. First, in order to alleviate liability on themselves as students if something went wrong, team members transferred ownership of the server to their legal aid partner — a broader and more resourced organization that could address issues it was experienced in dealing with.
Moreover, the team decided that to reduce chances of confusing translations, it would ensure that the tool sends the message in the original language as well as the language it was translated to, so recipients could attempt to translate the original message if necessary. This policy of sending both the English and Spanish text could ensure that users had an opportunity to spot errors, understand any confusing translations, and have more control in the messaging.
Lastly, the team briefed pilot participants about possible limitations, particularly in regard to the accuracy of the T2T tool and that it was currently in beta testing.
The ethical issues with potential inaccuracies of the T2T translation tool were minimized by leveraging common, industry-standard technologies. According to the interviews with the team, many lawyers were already using Google Translate to communicate with non-English-speaking clients, so using the same program in the back end eliminated the risk of providing significantly different results than what would normally be the case when the tool wasn’t being used.
In future iterations of T2T, there is also a possibility of working with experts in language access translations to supplement Google Translate’s proposed translations with glossaries of how best to translate legalese and other complicated language. There is interest from courts and legal aid groups to take informal offline translation glossaries and structure them into a T2T implementation to make up for Google Translate’s limitations.
Now that the product is through an initial pilot and moving toward scaled-up implementation, the team has some insights for future students and teams working in a similar type of public interest technology development class.
1. Team Commitments
When forming teams, lay out team member commitment expectations from the start, including duration of project commitment and amount of time commitment needed. Regularly check up on these expectations and whether they have changed to prevent unexpected drop-off in key project work times.
Because of the class-based nature of this project, too many students were interested in the topic initially to form an effective team. Also, because this 20-week class spanned two quarters of the academic year, several students who committed to the project early on became less invested as their interests and schedules changed during the year.
Conversely, several more students joined the project halfway through, as the second quarter began, leaving highly committed team members concerned as to which core members would complete necessary tasks. The nature of a class-based project is that there may be varying levels of commitment to working on the project beyond class requirements, especially after the class ends. A key factor in this team’s success was that a pair of students remained committed throughout the entire class — and beyond it — during the pilot.
2. Setting Expectations
When working with community partners that would benefit from your innovation, be realistic with your promises to interested stakeholders, particularly when developing a technological tool.
As a team member mentioned: “Don’t promise the world … be really upfront about what you can and can’t offer.”
T2T team members realized that although their technology was aimed at addressing a deep frustration of monolingual Spanish-speaking parents, the state-of-the-art translation technology was nowhere near perfect. They had to design a solution that recognized that “translation technology isn’t going to solve everything” related to the problem.
The team felt that when dealing with the legal language context in particular, accurate translation became especially difficult to provide. Nevertheless, testing showed that stakeholders still valued the translations, even without being able to rely on them as fully accurate.
Being transparent about the constraints of the technology offered the T2T team members a sense of shared integrity with their stakeholders and ensured that the impact of an ineffective final product on individuals in need was mitigated.
3. Diverse Research Support
The team also had recommendations for future teaching teams for public interest tech classes. When anticipating that students may be conducting interviews with a wide range of stakeholders, teaching teams should consider providing students with a comprehensive curriculum on how to interview a variety of populations and how to conduct ethnographic research.
While the class provided a foundational training for interviewing people through a mock interview workshop, T2T team members expressed a desire to learn how to effectively interview different stakeholders and how to personalize their interview style to suit relevant contexts in order to garner the most effective feedback while respecting interviewee time and values.
4. Don’t Jump to Product Development Too Soon
The team recommends that teachers and students ensure that during the initial stages of the project, the priority be on rapid ideation, prototyping, and feedback-garnering over high-fidelity product development.
The T2T team found that the design-thinking elements of the class that promoted guerrilla-style drafting of ideas, while being encouraged to independently reach out to relevant stakeholders of different perspectives for feedback, was critical to understanding the nuanced and less obvious issues at stake.
For example, if T2T team members had not reached out to stakeholders with a broad set of initial, rough ideas, they may not have realized the importance of the informal teacher-student relationship to the development of the IEP as well as longer-term educational stability. Spending the first part of the class exploring a variety of possible solutions let the team gradually find a promising challenge definition, as well as the type of technical solution that could work with the stakeholders’ preferences and capabilities.
The team seeks to continue the project in partnership with Stanford University’s Legal Design Lab, which has offered to provide funding and human resources to sustain the project. One member of the team is working directly with the lab to continue development of the project.
As for next steps, the team aims to remove bugs from the tool and produce an informative webinar to teach interested legal aid organizations across the country how to use the T2T tool. Anyone interested in working on an implementation or study of language access text messaging, whether in regard to IEPs, legal advocacy, or other channels, can contact with the Stanford Legal Design Lab.
Critical Learning Questions
- How might a project team secure a valuable partnership, when the most directly relevant stakeholders do not hold the authority to implement a proposed solution?
For example, the T2T team mentioned that a crucial stakeholder — local teachers who could use the T2T app to communicate with Spanish-speaking parents about their child’s Individualized Education Proposal — could not independently make the decision about whether T2T could be adopted at a school wide level.
- What could students do to explore a pilot partnership in this situation and offer value to their pilot partner?