--- title: Kevin Scannell description: Interview with Kevin Scannell on love.irish. Develops software and resources to help Irish language learners and fluent speakers use Irish online canonical_url: md_url: last_updated: 2023-12-23T05:54:03Z --- # Kevin Scannell Develops software and resources to help Irish language learners and fluent speakers use Irish online Kevin is a professor of Computer Science at the University of Saint Louis and develops software and resources to help Irish language learners and fluent speakers use Irish online. ## Some of his projects - [GaelSpell](https://cadhan.com/gaelspell/): an open source spellchecker for Irish. ## Questions & answers **Q: What are some initiatives or upcoming events you’re excited about? / Aon rud spéisiúil beartaithe agat don teicneolaíocht Ghaelach i mbliana?** _Translation: "Do you have anything interesting planned for Irish technology this year?"_ I will be on sabbatical leave from my university next academic year, and I’m planning on spending that time working on some important foundational technologies for Irish that I hope will lay the groundwork for better machine translation, speech recognition, and so on. I’ve mostly been focused on user-facing products like our software translations for the last several years so I’m looking forward to having the time to tackle these new challenges. **Q: How can existing people who have ability in Irish but are living their lives through English be encouraged to start using what Irish they have, particularly online.** Thousands of Irish people have passed through school, Gaelscoileanna or even from the Gaeltacht with a good level of Irish and are living their lives through English, especially online. They may have lost touch with using their Irish through new job, new place etc and undertaking these big changes might not have thought of keeping their Irish Language an active part of their daily lives. I'd imagine in Dublin and other urban centres they are a not insignificant number and have good basic language skills. How can they be encouraged to use their language everyday on and offline, there are many resources that people are unaware of, Gmail, Facebook and WhatsApp that have an Irish Language interface. Going online makes it very easy! First, it doesn’t matter where you live or if there are any Irish speakers near you, it’s possible to find other Irish speakers on Facebook, Twitter, etc. 24/7. Check out the group Gaeilge Amháin on Facebook, or follow some of the top 500 tweeters in Irish. All of the Irish speakers I’ve encountered online have been friendly, supportive, and patient, even going back to when I was a true beginner. Try posting a couple of words or start a conversation. If you don’t want to post to a Facebook group with 11000 members, tweet at me! **Q: An féidir cabhrú le muintir na hÉireann le fada a chur ar litreacha? Fadhb acu faoi láthair. Scéal san nuacht. GRMA as gach a dhéanann tú don teanga, a chara! Laoch na Gaeilge thú.** _Translation: "Is it possible to help Irish people put fadas on letters? They have a problem at present. Story in the news. Thank you for everything you do for the language, a chara! You are a hero of the language."_ The good news is that all modern computers and mobile devices can “do” síntí fada — don’t let anyone tell you otherwise! There is great advice for typing the Irish language on the website scriobh.ie, including a section on síntí fada. You can even buy their special Irish language keyboard! The tricky part is when “legacy” systems that were not built to handle síntí fada come into play. This was the situation that Ciarán Ó Cofaigh ran into with the HSE in Ireland, where certain clinical systems were unable to handle the fada so they avoided using them entirely, in any of their systems. But here’s the rub: some of those so-called legacy systems are less than 10 years old, and had support for the Irish language been a priority, could very well have supported the fada. **Q: What is your favorite Irish word or phrase?** **A: _Dúchas_** I really like the word **“dúchas”** — it’s one of those quintessentially “untranslatable” words that make for good lists on Buzzfeed. The entry in Ó Dónaill’s dictionary is fantastic; it’s chock full of useful collocations _(teanga dhúchais)_, proverbs _(briseann an dúchas trí shúile an chait)_, idioms _(filleadh ar do dhúchas)_ and a variety of English translations, none of which fully captures the meaning of the Irish word. To me it represents the full Gaelic cultural package: language, music, seanchas, etc., all rooted in place in the Gaeltacht. These are all the things I was totally clueless about when I started learning the language, having grown up in a culture where there are no strong ties to places or communities. I’ve joked that I’ve spent the last 25 years trying to fully understand this one Irish word! **Q: What got you into programming and how did you end up working with minority languages?** I’ve been programming since I was young, probably about 11 or 12. I’m among that first generation of people (born in the late 60’s early 70’s) who might have had a computer at home growing up. At that time, if you wanted to do something fun on your computer you had to program it yourself! So I wrote a lot of games as a teenager, first in BASIC, later in C and C++. I now mostly use Python and Perl for my language processing work. GaelSpell was my first big project for Irish, and as I mentioned above, part of that involved writing a web crawler to search for and collect Irish language texts online. This is something that has turned out to be quite useful for many other minority language communities as well. I’m now running the crawler to collect material in more than 2200 languages, and collaborating with many communities on basic language technologies like the ones I’ve developed for Irish. **Q: How does Irish compare to the other Celtic languages with building programs?** Linguistically the Celtic languages are similar enough that things can be structured in much the same way. Indeed, I’ve been able to repurpose some code from Irish projects for Scottish and Manx Gaelic with minimal changes. The real challenge for those other languages is the lack of data. Modern language technology is driven by data — big collections of text, mostly, sometimes with linguistic annotations (part of speech labels, for example). Irish has a huge amount of data relative to other European minority language, probably somewhat less than Welsh but much more than the other Celtic languages. **Q: What are some of the challenges working with Irish in computer science?** When I started out in the late 1990’s there were some huge challenges. There were almost no resources for Irish at the time in electronic form, not even a good word list, so I had to develop a lot from scratch. Originally I built that work into my language learning process: as I learned new words I tried to be disciplined about entering them into a structured database on my computer. That turned out to be a good idea. I wrote a bit of code for recognizing and generating inflected forms of nouns, verbs, adjectives, and so on, and that was enough for a basic spell checker (I released the first version of GaelSpell in June of 2000). That same database (now greatly expanded) underlies all of my Irish language projects even today. There are some linguistic challenges as well, but I tend to focus on the positives! There are great dictionaries and a long history of grammatical analysis of the language which help inform the work I’m doing. The spelling system is relatively simple and has few exceptions. Even things like initial mutations (which seem to cause problems on the surface) can be an advantage in processing Irish. **Q: What was your favorite project to work on?** I created translation engines that translate from Scottish and Manx Gaelic to Irish a few years ago, along with bilingual dictionaries. That project was fun as a technical and linguistic challenge, but also because I learned a lot about the other two Gaelic languages and their relationship with Irish. I also deepened my knowledge of Old and Middle Irish. I really enjoy maintaining the GaelSpell spell checker, even though it’s probably the least interesting of my projects from a technical standpoint. I have a number of web crawlers that are running 24/7 and which collect Irish language blog posts, new articles, tweets, Facebook posts, and so on, basically everything written in Irish online. This gives me an up-to-the-second (literally) snapshot of how people are using the language. I use these texts to keep a close eye on new terminology, coinages, etc. and incorporate those into the spell checker on a regular basis, once a month or so. **Q: An chun aimhleas teangacha mionlaithe an teicneolaíocht sa bhfadtréimhse?** _Translation: "Is technology detrimental to minoritized languages in the long term?"_ For most of them, yes. The language technology landscape is moving extremely quickly. I teach a graduate-level course on AI and Machine Learning and very little of what I taught in the same course five years ago is important or relevant now. The move now is toward so-called deep learning approaches that risk leaving most small language groups behind. There are a few reasons for this: first, most of the models researchers and big tech companies are using are language-independent. On the surface this seems like a positive for under-resourced and minority languages since companies like Google can scale certain technologies up to 50 or 100 languages without much extra effort. But as we’ve seen with Google Translate, the results can be far from satisfactory since they don’t take into account complexities of particular languages. Secondly, deep learning models require linguistic data on a massive scale and tremendous computing horsepower, so much so that most minority language communities are unlikely to be able to keep up. This means that in a future of voice-controlled devices, appliances, and so on, minority languages will be pushed out, and this will become yet another factor causing a shift to English. Despite all this, I’m optimistic about Irish since we have some good computational resources already, support from Foras na Gaeilge and Roinn na Gaeltachta, and a community of young researchers working in this field, people like Teresa Lynn, Meghan Dowling, and Neasa Ní Chiaráin. **Q: Could you recommend any resources like open source data, corpora, or APIs for people interested in CL and NLP for Irish?** The Cadhan Aonair website is a good place to start since it has links to all of the code and datasets for my Irish language projects. Many of those links point to my GitHub page which is worth a look too. I’m really pleased that there is a culture of openness and sharing among the people working on Irish language technology. Elaine Uí Dhonnchadha from TCD has made her Irish analyser and tagger available under an open license, as has Teresa Lynn with her important work on Irish dependency parsing. Open source software and open data is the only realistic way to develop language technology in a minority language context. In a talk I gave in Conamara in 2016 I argued for a kind of “service-based architecture” as a way of accelerating the development of new language technologies for Irish. Since then, as first steps in this direction, I’ve released APIs for Irish spell checking, grammar checking, diacritic restoration, standardization, and automatic translation from Scottish and Manx Gaelic. As a proof of concept, Michal Boleslav Měchura has used these to integrate spellchecking and standardization into his Pota Focal site. These APIs are all documented on the Cadhan Aonair site, and I hope there will be more to come soon. If you’re a developer interested in Irish, get in touch and let’s build something! **Q: What are some programs that all beginners should know about? Chrome plugins? Apps as Gaeilge?** The resources for learners that are available now are incredible, especially in comparison with what was available 20 years ago. Instead of trying a complete survey here, I’ll point you to Audrey Nickel’s great Beyond Duolingo post, and, if you’re on Twitter, the hashtag #DIYGaeilge started by Osgur Ó Ciardha. They both have great advice on learning resources and small ways you can incorporate the language into your daily life. I always recommend that people use the Firefox web browser instead of Chrome since it’s available in Irish. It doesn’t take long to get used to the basic menus: Comhad = File, Eagar = Edit, and so on. And with Firefox you can install my GaelSpell add-on which enables Irish spell-checking whenever you type in the language through your browser. I’m a member of a team of volunteer translators that is working hard to create Irish versions of all your favorite software and web sites. We’ve translated GMail, Twitter, WhatApp, Firefox, and much more into Irish. The site nascanna.com is now a bit out of date but is still the best source of information on software in Irish. **Q: How did you learn Irish?** I started in earnest around 1994 but at the time I didn’t know anything about how to learn a language. In a lot of ways I’m the perfect example of what not to do! For the first 10 years or so I learned entirely on my own with books, dictionaries, and cassette tapes. I had a copy of Learning Irish by Mícheál Ó Siadhail that I worked through diligently, but I was trying to learn the language in much the same way as I was learning mathematics at the time, as a kind of analytical exercise. There was a “brute force” aspect to what I was doing, too; I got my hands on a copy of the Foclóir Póca and basically read it cover-to-cover thinking it was very important that I know “all of the words”. I did develop a pretty good vocabulary and an understanding of the grammar so I could read without much trouble, but I couldn’t hold more than a basic conversation. I subscribed to the mailing lists Gaelic-L and Gaeilge-A in the 90’s and those were my first “social” interactions through the medium of Irish. I’ve tried to spend at least some time in Ireland every year since about 2006, and have done a few week-long courses which have helped a lot. I now listen to Raidió na Gaeltachta every day (Adhmhaidin, Iris Aniar, and Ardtráthnóna) and feel a lot more comfortable understanding fast, natively-spoken Irish. I still need more practice speaking. **Q: Cad a mheall thú go Gaeilge?** _Translation: "What attracted you to Irish?"_ Family connections. Both sides of my family are Irish and I was always interested in Irish history, culture, and politics. The language was not part of my upbringing though; I didn’t start learning it until my early 20s. _Read the full interview at https://love.irish/interviews/kevin-scannell_ ## More information - [Search love.irish for Irish & Celtic records (codex, library, collections, events, people, topics)]() - [Full markdown index of the site, every page (handy to browse, share, or hand to AI)]() - [Love.irish home]() --- Source: https://love.irish/interviews/kevin-scannell Made with ❤️ from love.irish – learn Irish (Gaeilge) through community, resources and a little adventure.