SaySomethingin Irish (Beta)

Teach Yourself Complete Irish - pronunciation guide.pdf (1.3 MB)

Just in case :slight_smile:

It helps to return to the rules time to time. They totally do not assimilate in one go, but some experience in speaking makes the process a bit easier.

2 Likes

I recently revisited the Irish course after some months away and my brain is better able to understand the prepositions! Definitely took some effort but here we are :smiley:

Which grammar books did you get?

2 Likes

To start with, I got the Collins Irish Dictionary and then Teach Yourself Irish Grammar. Think I might shell out on the Routledge ā€œBasic Irish Grammarā€, because I found the Welsh one so useful. I’m hoping they’ll have one of their sales soon!

3 Likes

I’m starting to figure out subtle pronunciation patterns and I don’t think robot is doing a very good job of modeling those.

Irish has difference between broad and slender consonants, meaning if consonant is surrounded by e/i it sounds softer than when it’s surrounded by a/o/u. Broad and slender consonants are articulated differently. At this point they don’t sound differently in the app. Fada (Ć”) doesn’t seem to be working either. A in tĆ” should not sound like a in ag, with fada it’s a rich deep sound, more like aw.

Unwritten vowels are also missing. Some consonant combinations have subtle a (in broad positions) or subtle i (in slender ones), so Gaeilge is actually more like Gaeilige, orm more like oram, which gives softer, more melodic texture.

Similar issues are considered a serious problem for Duolingo, which mostly teaches to read and maybe write a bit. If the aim is to teach speaking, I’d say correct sound should probably be a priority. At this point model sounds closer to an English speaker that doesn’t quite get Irish phonetics than to actual native speaker. Most Irish speakers learned Irish as second language and have anglicized pronunciation too as a result of low exposure to native speakers, and it is considered a problem. Compare here, for example, Irish of the interviewer, learned as second language, and native speaker’s Irish:

2 Likes

One of the hopes SSi has moving forward is that when our income levels increase from the major languages, that DO have an excellent choice of AI voices available, we’ll be able to engage first language speakers of the lesser spoken languages to do the recording for us.
Using AI voices for the Irish is just an intermediate step so that we can offer a course for people to learn from, but supplementing that with as much listening to native speakers as possible is important for learning correct pronunciation.

2 Likes

This approach sounds problematic. Munster, Ulster and Connacht Irish sound all differently. To figure out that you are not listening to another dialect variation, but rather to mispronunciation, and to repeat not what you hear, but what it should actually sound like, you need to already have a pretty good grasp of Irish phonetics, which you are unlikely to have if you never studied Gaelic languages before. Broad and slender L are similar to how Š» and ль sound in Slavic languages, but English normally does all consonants as broad ones, so natural inertia of English as first language aligns here with what robot does. Seems to me that likely outcome will be reproducing the way robot sounds after all the repetitions. If that’s your entry point, you are also unlikely to know that actual native speakers of Irish Gaelic are a minority of up to 80 000 people, optimistically, others were mostly taught by school teachers who were not native speakers themselves, and as a result authentic pronunciation is rare (which means podcasts are generally unreliable as pronunciation models). And the app doesn’t explicitly tell anywhere ā€œplease, don’t trust me, listen to radio Gaeltachtā€.
Irish is endangered language, I’d say it requires more careful approach.

3 Likes

In an ideal world we’d have a variety of dialects available from native speakers giving learners a wide choice of genuine pronunciations to learn from, however when the actual choice is between offering something that is less than perfect but will give people the chance to start taking part in conversations and contributing to the revitalisation of the Irish language, OR, offering nothing because what we have isn’t perfect, I think the first option is better.

It will be obvious to any first language speaker that someone who has done the SSi Irish course is still learning, but the reaction of most first language Welsh speakers is delight that someone is making the effort to learn their language, and I imagine it will be similar in Ireland.

Don’t worry about subtleties of whether a consonant is ā€˜broad’ or ā€˜slender’ or you’ll never speak! Just dive in and do your best, and if you have the opportunity to talk to native Irish speakers, your pronunciation will gradually adapt to sound more like them.

3 Likes

Ireland has different situation with language than Wales, as I understand. Language revival strategy was built almost entirely around mandatory school lessons, which means that generally people acquire English first, Irish second, but as academical subject rather than language to actually talk. Thirteen years of learning Irish three hours a week. Sociological data shows that outside Gaeltacht areas barely anybody actually speaks Irish at home, despite having this amount of learning. Irish is not phonetic language, so grammar can be taught more or less effectively through textbooks, but pronunciation is a very different art. With intergenerational transmission broken almost everywhere except for Gaeltacht areas, which are few and relatively small, for language this means very real risk of eventually losing it’s authentic sound patterns. Gaeltacht areas have their own issues. There’s lack of housing, for example, which means if you want to start a family of your own, you might need to leave the area because there are no available places to live. Which again harms intergenerational transmission. Isolated clusters of native speakers mean there’s a paywall for getting speaking practice with native speakers even for those who live in Ireland. Just another faulty pronunciation model is probably unlikely to harm the language more than it was already harmed by centuries of occupation, but it doesn’t help either. Authentic recordings on the other hand could be a game changer, because they could scale contact with language in it’s uniqueness in a way that schools by far failed to achieve.

My own questionable pronunciation is certainly not an issue here, I’m more concerned about possible impact on population level. Robots affect human speech in a non-trivial manner, frequency of a number of English words in natural speech already changed significantly as a result of LLMs overusing them. And if there are any plans for partnerships with Irish schools, we are talking about population level.


Just to get the idea of how rapid is the language shift, that’s some data on Irish-speakers over the past century. Gaeltacht areas where Irish is language of everyday communication have been shrinking in size. Specifically in Kerry which is still considered a Gaeltacht area only 7% of the population was speaking Irish daily outside of educational system (2016 data) and they now have Russian linguist as an Irish language officer who tries to improve the situation (which probably says something about the state of decline, when you need a Russian guy to talk Irish people into talking Irish).

I’m finding the Teanglann.ie app very good for pronunciation (and it’s possible to choose between the three dialects on that), along with videos/podcasts from Learn Irish Online (although I haven’t worked out what his dialect is yet…) Off to Galway/Connemara in April, so hoping to thrash out further detail while I’m out in the Gaeltacht :slight_smile:

6 Likes

I’m really looking forward to hearing how you get on in April! I’m sure you’ll make a great impression over there. :shamrock:

2 Likes

The spoken Irish of pupils in Irish-medium schools in Northern Ireland.pdf (1.0 MB)

Fascinating. Apparently there was some research done on language acquisition in Irish-medium schools, so we have some hard data on experiential route of learning Irish. That’s 130 pages of interesting read, in cĆŗpla focail:

  • Passive skills end up on near-native level
  • Active skills do not though, and typically English syntax persists even after several thousand (!) hours of instruction in Irish. What’s especially interesting, it’s not about mentally translating from English, sentences get formed directly in Irish with English syntax. Somehow without at least some analytical learning Gaeilge tends to consolidate like that when acquired outside Gaeltacht. Even with total immersion at school!
  • Motivation to improve accuracy drops sharply after becoming fluent enough to communicate effectively, so typically kids end up with errors in something like 30% of their sentences, which they mostly don’t notice, and they get surprised and disappointed when they discover their actual level of accuracy. Apparently fluency is a mixed blessing and it’s better to get grammar right before you are fluent enough to express whatever you want.

Practical considerations: yep, it’s still good to read some grammar, even for talking, and structures more complicated than verb-subject-object apparently require some targeted attention.

1 Like

There are a lot of interesting studies on the effectiveness (or lack of) of immersion learning from around the world comparing methodologies etc. I haven’t seen one from Northern Ireland though, so go raibh maith agat. I’ll take a look!

1 Like

I’d say best part is list of typical errors. Practically a plan which patterns to drill religiously.
Have you seen anything interesting on phonetical aspects of immersion learning?

I can’t remember the specifics now - it was a few years ago when I was doing an MA in Bilingualism and Multlingualism with a Welsh university - but I do remember that there is some evidence to say that a new dialect may be developing in the south east, based on the Welsh spoken by young adults who have gone through the Welsh medium education system, but come from non Welsh speaking homes. It’s something for sociolinguists to study :slight_smile:

2 Likes

This is very interesting! Blend of Irish spoken by kids who went through Irish medium schools is tentatively considered an emergent dialect too (either that, or it’s a seriously corrupted form of Irish). Family transmission of Irish is in worse shape than Welsh, so Gaelscoil Irish has a good chance to become a dominant variant of Irish when Gaeltacht tradition dies out. For methodology this poses a question of what does it mean to preserve Irish: try and maintain what’s left of Gaeltacht dialects or embrace natural life cycle and let them die out with all their complicated stuff.

I’m watching ā€œNo BĆ©arlaā€ now, it’s an old show and the situation is a bit better now (TG4 did more for speaking skills than schools, young generation is more likely to have Gaeilge), but still, the frustration is real. For all the passion ManchĆ”n Magan had for Irish (the man died this autumn), the difference is immediately obvious when he speaks with a man from Gaeltacht: ManchĆ”n’s Irish sounds very much like English with different set of words, Gaeltacht Irish is melodically unique. And ManchĆ”n was L1 Irish-speaker, he had no contact with English for the first five years of his life, but still, he was primarily a city-dweller.

Gaelscoil kids are fluent when speaking to each other, but they have trouble understanding Raidió na Gaeltachta, and folk from Gaeltacht don’t quite recognize their Irish as the language they speak themselves. So, if the strategy is supporting city Irish, broad/slender consonants and other intricacies don’t matter much, because speakers of this variant of Irish are unlikely to properly differentiate those themselves and probably won’t even notice something is missing.
Something very beautiful will eventually be lost on this road though, which is sad.

2 Likes

IIRC they found similar results with French immersion schools in Canada.

I don’t really think that we can blame SSi for not (currently) being part of the solution to what is being lost as Irish changes, as it’s really not SSi’s responsibility; but I can definitely understand your point of view. Like you, I also think something beautiful will have been lost. You can compare Cornish to Welsh in that respect: Cornish is still beautiful, but the way most of its speakers speak it (there are rare exceptions), it is clear that much of its beauty and character has been lost.

What I would say is that if you are as passionate about this as you seem to be, present some of this stuff to relevant Irish language bodies and try to secure some funding so that SSi can become the solution you want it to be. I imagine everyone at SSi would probably be thrilled to have the chance to do that. If you succeed, you could literally save the language you love.

4 Likes

Another lovely dream would be to have a Slack platform and online chat groups like we have for Welsh, but led by first language speakers from the Gaeltacht - different groups for the different dialects. Can you imagine how amazing that would be? Unlikely in the foreseeable future, but you never know!

4 Likes

I watched No BĆ©arla in 2014 or so and it’s well worth a viewing. I should watch it again.

SSi internship opportunities? :eyes: