Local Language / Global Network: Designing Mobile Technology for Indigenous and Minority Language Users · Global Voices
Derek Lackaff

The photo by Derek Lackaff taken in rural County Galway, Ireland in January 2016
A version of this post was originally published at r12n on Medium.
Last summer, an Irish woman named Caoimhe Ní Chathail sent her mobile phone company a tweet to let them know that she was having some trouble using their website. For months, the site had been rejecting her name as “invalid” because it contained an accented Irish letter (í). The mobile company’s response was to ask if Caoimhe could just use the “English version” of her name. This very public exchange caused a minor uproar on Irish Twitter: an Irish company seemed to be questioning the acceptability of using the Irish language online in Ireland. Within a few weeks, the company had actually updated its site so it would not reject Irish names, but this small incident illustrates the type of interactions that speakers of Irish frequently face in their interactions with digital media.
Irish (Gaeilge) is the first official language of Ireland, a recognized minority language in Northern Ireland, and an official language of the European Union. Instruction in the language is mandated throughout primary and secondary school. The language has a rich literary heritage, and is used in radio, television, and across the internet. Second language learners outside Ireland can find formal opportunities to study the language at universities like Notre Dame or University of Sydney, or more casually at a MeetUp or using free apps like Duolingo. Despite this support and interest, Irish is considered to be “definitely endangered.” Current estimates suggest that there are only about 40-70,000 daily speakers out of the Irish population of 4.6 million. Irish is a community language spoken outside of school in very few areas, most prominently in specially-designated rural regions called Gaeltachtai. Irish is thus a minority language in Ireland where most people use the majority language — English — in their daily life.
Percentage of people who speak Irish daily outside the education system, 2011 Census. By SkateTier CC BY-SA 3.0 via Wikimedia Commons
A recent report suggests that Irish could actually die out as a community language within a decade or two. This puts Irish in the company of thousands of other human languages that could disappear in the coming years and decades without active intervention. While the technical classification of languages and their endangerment levels is somewhat contested, the global trend is toward less linguistic diversity as languages like English, Spanish, or Chinese displace local and indigenous languages. Although languages have come and gone throughout human history, colonialism, assimilation programs, and globalization have put massive pressures on languages of fewer speakers over the past century. In the United States over 150 indigenous languages still have living speakers, but decades of cultural suppression have left these languages in a very precarious position. There are still 2,000 speakers of Lakota, for example, but most first-language Lakota speakers are now in their 60s and 70s. Intergenerational transmission of a language — grandparents and parents to children — is the core process in a language’s survival, so language preservation and revitalization is often a race against time.
There are many reasons why users should be empowered to use their own language, and for preserving linguistic diversity. A language represents an unbroken connection to a particular culture and perspective, and encodes unique information about humans and the world. Access to a heritage language provides profound benefits to peoples who are resisting and recovering from colonialism. With my colleague William J. Moner, I recently started a research project that examines the use of a minority language — Irish — across mobile media. We suggest that lessons from the Irish experience might be applied to the other thousands of endangered human languages, and that interaction designers have an important role to play in the language revitalization process.
Communication technologies, and social media more specifically, present both opportunities and challenges for minority language preservation and revitalization efforts. Benefits include the ability to widely distribute cultural and news media, as well as learning materials; and the opportunity to improve the prestige of the language, and promote its use among younger generations. This last point becomes increasingly salient as global youth cultures move towards always-connected communication contexts exemplified by mobile messaging platforms like Snapchat, Facebook Messenger, and WhatsApp.
Digital communication platforms potentially enable members of language communities to remain in contact with one another whether they are local or distant. However, as the personal computer loses primacy to mobile communication devices, interaction and input rely more on advanced processes like speech recognition and gesture typing than on traditional keyboard typing. While accepting basic textual input from a keyboard is a trivial computational exercise, providing more advanced functionality such as spelling correction or speech processing is more expensive in terms of time and resources. Some observers have started sounding alarms, such the poet and musician Sjón regarding his native Icelandic language (330,000 speakers):
The broader and more serious implications are for the language as it is used in daily life. Technology is moving towards AI and speech-controlled applications, and the companies developing it do not see preserving languages spoken by few as their responsibility. When the day comes that we have to speak to our refrigerators in English (which I believe is not far in the future), Icelandic will retreat very fast.
And Icelandic isn’t even technically endangered.
The computerization of minority languages provides several new opportunities for language preservation and revitalization. These opportunities include easier and less intrusive language documentation, for example by using mobile apps like Aikuma. They include new opportunities for pedagogy and learning, as new text, audio, video, and computer-assisted learning software becomes easier to produce and disseminate. Among my favorite examples are the game Never Alone, which was developed in close collaboration with Alaska Native storytellers.
Another fun example is the Lakota-dubbed version of Berenstain Bears now available on Youtube. Finally, software localization or translation, particularly of open-source software, is an area of particular promise for minority languages. Kevin Scannell, a computer scientist at Saint Louis University, lists many software applications — including desktop apps like Mozilla Firefox and LibreOffice, and webapps like Gmail and Twitter — that are now available in Irish thanks to his and others’ contributions.
Globally there has been a massive shift towards mobile, and it’s clear mobile devices shape the use of a language in different ways than personal computers do. The first generation of mobile text platforms was primarily composed of mobile phones with small screens and numeric input pads. Because the standard numeric keypad only had 12 buttons, most characters were entered using multiple button presses to cycle through available options. For example, a user would press the button labeled “2” six times to produce the Á character, first cycling through the characters A, B, C, 2, and Ä. Predictive text is an input technology designed to ameliorate some of these problems, allowing the software to “predict” which word a user intended to produce based on a reduced number of button taps. Predictive text increases text input speed by as much as 30%, and was broadly implemented for global languages, but support for minority languages with smaller commercial markets languished.
Sure beats button mashing on a feature phone.
Touchscreen devices such as the Apple iPhone and iPad characterize the second generation of mobile platforms. Such devices typically employ a “soft keyboard” that appears onscreen when text entry is necessary. For a minority language with a script closely related to a majority language, basic input of text is likely not a problem. Modern Irish, for example, uses the majority of letters from the Latin alphabet with five accented vowel characters (síneadh fada) that can be accessed from an English keyboard with just an extra key tap. Typing characters one at a time, however, is probably a foreign concept to most touchscreen device users. For users of global languages, software keyboards provide advanced features such as gesture typing, spellchecking, and automatic corrections. These types of interface technologies greatly increase the convenience, accuracy, and speed of text input on mobile devices, and this may introduces a bias toward the majority language even for fluent minority language users.
To better understand how minority language users interact with mobile technologies, we developed and conducted an online survey of young Irish speakers and learners in Ireland. The survey was derived from one used in a previous study of teenage speakers of Frisian, a minority language spoken in the Netherlands. As might be expected, these participants tended to be very heavy users of social media, particularly of mobile social media apps. They reported divergent levels of language proficiency: understanding of Irish was fairly high, speaking proficient was a bit lower, and writing was significantly lower.
These users had a lot to say about their experience using Irish in social and mobile media, and the various obstacles they faced, and three broad themes emerged from their responses to open-ended questions. The first theme was that the audience for Irish was much smaller than the audience for English. Most of the participants have many friends who speak Irish, but many hesitated to use the language socially in digital media. In fact, many suggested that it would be “unusual” or “abnormal” to see Irish in their feeds. No users reported that they attempt to segment their audience by language, for example by using Facebook lists.
The second, related theme was that users note their networks are linguistically pluralistic, and they were concerned that posting in Irish would exclude or offend those who didn’t speak the language. This finding was in many ways paradoxical, as participants also generally feel that the language is an important part of their culture and personal identity.
The third theme was that the mobile media context presents specific technical challenges. Even fluent users are faced with additional challenges in producing written Irish on a mobile device — rather than using advanced gesture typing with autocorrect, they are forced to laboriously input text letter by letter. In many cases, the keyboard actively works against them, as the English autocorrect marks all their Irish as misspelled, or “corrects” their Irish words into English. The challenge of producing “correct” Irish on a mobile keyboard is exacerbated for many potential Irish users who are not completely fluent. Several mentioned that they had been publicly critiqued or shamed by “grammar snobs” or prescriptivists for writing “incorrect” Irish online.
These findings suggest some specific issues that interaction designers might consider in designing for minority language users.
Most of our respondents find that current mobile interfaces inhibit the use of Irish. Whether it is accessing accented characters or battling the autocorrect in another language, the interface itself can push users toward the majority language. In some smaller language communities, users have no problem with switching keyboard layouts, disabling assistive technologies like gesture typing and autocorrect, and typing words out a letter at a time. However, in a minority language context such as the Irish example, users make the valid assumption that everyone in their network is proficient in English, and thus elect to take the path of least linguistic and technical resistance. Many of the fluent Irish users in our sample send English messages to their Irish speaking friends and family, even if they speak Irish with them in person and on the phone.
We also find that social media popularity metrics push Irish users towards using the majority language. Such metrics tend to be simplistic but are prominently featured on every post or update: think “likes” or “comment counts” or “retweets.” In the attention economy in which social media users are encouraged to perform, anything that might limit engagement — like the use of a minority language — will be avoided.
Second, all minority language users live and communicate in a complex cultural context. Irish has a deep and complex set of associations for Irish people, ranging from a sense of heritage and pride, to concerns about political conflict or painful experiences learning the language in school. Further, it is safe for Irish users to assume that English will be understood by their entire local network, while Irish will only be understood by some of their Irish network and none of their global network. This complicates the sustained use of Irish, particularly for younger generations. If the goal is to encourage casual, creative use of the language, there may be value in designing to promote new language norms, for example with monolingual platforms that remove the social ambiguity of bilingualism. Other minority languages will have different relationships with the majority language, warranting other types of designs.
Finally, languages have specific and unique linguistic requirements related to their orthography, morphology, and computational resources. Irish is somewhat unique as a minority language in that it is comparatively well-resourced. For example, several inexpensive or free functional Irish keyboard input technologies are already available for mobile devices, but adoption seems to be low among the potential userbase we surveyed. None of our participants mentioned using the Adaptxt keyboard, a free iOS and Android keyboard that provides high-quality Irish predictive and autocorrect functionality and easy language switching. A few users noted that they used the Swype keyboard, which is a separate download, but has full support for Irish. There are active centers of research in Irish computational linguistics, such as the ADAPT Centre collaborative in Dublin. Kevin Scannell suggests that for Irish, the challenge is not a dearth of technical resources or support, but actually connecting users with these resources: effectively marketing Irish-language software and interfaces to potential users, and convincing operating system developers and device manufacturers to integrate Irish language technologies into their products. There are indications that this may be happening for some languages — the latest version of Google's GBoard keyboard app for iOS and Android supports Irish, and allows for autocorrect and prediction in multiple languages simultaneously.
Teresa Lynn, a researcher with the ADAPT Centre, gives a TEDx talk on Irish and social media.
Language vitality requires practice in spaces where people are active in a mutual exchange of ideas and where conversation may occur. As global technology giants like Facebook and Google, for example, move aggressively into developing regions of the world, questions of linguistic self-determination and colonial resistance are becoming increasingly important. Encouraging the everyday use of endangered languages –in both offline and online contexts — warrants a strong push for participation and engagement by minority language activists, designers, and developers in these spaces.
As communication technologies move into increasingly intimate and developing cultural contexts, and as social media platforms facilitate communication within these contexts, the tensions between local languages and the global technologies will continue to vex both local communities and interaction designers. While Irish is a unique minority language, its examination provides insights into how global and local practices are mediated through mobile interfaces.