Fòram na Gàidhlig

I have just received an alert from DuoLingo that the new Scottish Gaelic Course is now in live Beta:

https://www.duolingo.com/learn

Have people found it to be pretty accurate so far?

Purple wrote: ↑Thu Nov 28, 2019 9:13 pm Have people found it to be pretty accurate so far?

It's difficult for me to comment on that as I'm an early learner (in my Gaelic journey - not my age ). However, there was nothing that disagreed with anything else I have learned previously and I believe the developers are native speakers (I think I read that somewhere).

I have been following Speaking our Language, Can Seo and the LearnGaelic resources for a few months and I grade myself as "A1 transitioning to A2" at the time of writing. When I took the test at the start of the DuoLingo course, I unlocked 76% of the entire course instantly. Within that there were a few words I didn't know (but could guess easily by context). So, based on this I would say that, so far (in Beta) the materials will take you up to A2 (given that I have 24% of the course to go and a few gaps in the 76% that I unlocked). This seems reasonable for a new Beta course and I expect they intend to add more in future. I don't know if that is helpful to you.

Ionatan wrote: ↑Fri Nov 29, 2019 9:28 amIt's difficult for me to comment on that as I'm an early learner (in my Gaelic journey - not my age ). However, there was nothing that disagreed with anything else I have learned previously and I believe the developers are native speakers (I think I read that somewhere).

Not sure that all of the developers are, but they only used native speakers for the voice files.

I am ambivalent about Duolingo as such, as it is basically earning money via ads for the owners, but all the considerable work, especially the immense start-up work done so far, is done by unpaid volunteers. (All credit to them! )

As far as accuracy goes, I have tried out a fair chunk (till I got sick of rats and tortoises...) and there are some oddities, e.g. it doesn't accept "glinn" as a plural of "gleann", only "gleanntan". It also didn't accept "seann taighean" - it wanted "seann thaighean", despite blocked lenition N>T.

You also have to be really careful not to miss out an "a" and to get the words you move up in the correct order, or it's wrong. The distractors in multiple choices are also so obviously wrong in many cases that I assume the exercise is more just re-enforcement, not a real choice.
There's not a very high level available yet, as far as I can see.
Also, presumably as still in development, there's no Forum - you can't query anything via the Discuss button.

Having said all that, it's certainly a useful further source for learners, thanks to the creators, and I suspect quite addictive, and like most things, probably most useful if combined with some sort of taught or other course.

faoileag wrote: ↑Fri Nov 29, 2019 4:32 pm I am ambivalent about Duolingo as such, as it is basically earning money via ads for the owners, but all the considerable work, especially the immense start-up work done so far, is done by unpaid volunteers. (All credit to them! )

I think this is a fair point (though I have a "but" below). Having personally achieved success with the LearnGaelic resources in conjunction with Anki spaced repetition, I'll be sticking with this method and not transitioning. I am a real convert to Anki - a tool that is consistently recommended by polyglots and medical students. My progress in Gaelic is definitely faster since I combined transferring lessons I've done on LearnGaelic to Anki (where the act of making the cards is, in itself, revision!).

BUT (as promised ), I am sure the creators of the Scottish Gaelic DuoLingo course are hoping for the same boost in learner numbers and language recognition that it has given Irish (and Hawaiian):

...by the time Irish President Michael Higgins commended them [the creators] at a public ceremony in 2016... , millions of people — more than the population of Ireland itself — had tried the Irish course, according to company data. Today, the language has more than 940,000 active weekly users, around the same as Hindi and Greek

[https://edition.cnn.com/2019/10/04/tech ... index.html]

Even allowing for the fact that the article is based on DuoLingo's own figures, it is very impressive. If Scottish Gaelic achieved something similar then we could soon expect a bit under 20 times the number of active learners than currently speak Scottish Gaelic (based on just the Irish weekly active users figure)! Now that surely not only helps safe-guard the language but coincidentally opens up opportunities for a whole cottage industry for Native Gaelic speakers (Skype lessons, immersion holidays, face-to-face tuition etc etc)... just musing here, obviously!

faoileag wrote: ↑Fri Nov 29, 2019 4:32 pmtill I got sick of rats and tortoises...

Now this is where DuoLingo wins over my approach. Chan eil fios agam an fhacal airson 'rat' no 'tortoise'. I know luch, which I suppose I can substitute for 'rat'...

I'm ideologically opposed to Duolingo, for similar reasons to the first paragraph of Faoileag's message. I'm also annoyed at how they've squandered the possibility of making something awesome, and made something with flashes of brilliance in a sea of "meh" (the platform, that is; not any specific course -- I've tried it out in about half-a-dozen different languages at different times), and for completely misinterpreting some of the basic rules of both language teaching and gamification.

I spent a few minutes doing Gaelic exercises the day it launched; here's a couple of observations:

The audio is generally good. While a lot of it is very clearly amateur, at least Duolingo seem to have levelled the volume so that you're not getting jarring leaps between quiet and loud depending on who and when it was recorded.

The Duolingo platform does Gaelic no favours. The "fridge magnet" type activities drop out punctuation marks, which is one thing, but Duolingo considers apostrophes at the start and end of words to be punctuation marks. Getting your head round all the different meanings of a and a' is hard enough without Duolingo rendering every a' into an a....

Ionatan wrote: ↑Sat Nov 30, 2019 1:53 pm

...by the time Irish President Michael Higgins commended them [the creators] at a public ceremony in 2016... , millions of people — more than the population of Ireland itself — had tried the Irish course, according to company data. Today, the language has more than 940,000 active weekly users, around the same as Hindi and Greek

[https://edition.cnn.com/2019/10/04/tech ... index.html]

Even allowing for the fact that the article is based on DuoLingo's own figures, it is very impressive. If Scottish Gaelic achieved something similar then we could soon expect a bit under 20 times the number of active learners than currently speak Scottish Gaelic (based on just the Irish weekly active users figure)! Now that surely not only helps safe-guard the language but coincidentally opens up opportunities for a whole cottage industry for Native Gaelic speakers (Skype lessons, immersion holidays, face-to-face tuition etc etc)... just musing here, obviously!

I have a slightly different take on that.
Given the phenomenal numbers of learners reported by Duolingo for several of their smaller languages, if Duolingo was making any real difference, the difference would already be blatantly obvious. I haven't seen a single report or academic paper discussing any significant change in the sales of Irish learning materials or the sign-ups to Irish courses.

To me, that kind of says it's pretty much inconsequential.

Níall Beag wrote: ↑Sat Nov 30, 2019 8:15 pm I'm also annoyed at how they've squandered the possibility of making something awesome, and made something with flashes of brilliance in a sea of "meh" (the platform, that is; not any specific course -- I've tried it out in about half-a-dozen different languages at different times), and for completely misinterpreting some of the basic rules of both language teaching and gamification.

I must admit that I didn't instantly understand the process but had put that down to leaping in over half way through having done the test, so would tend to agree with you on that score.

Níall Beag wrote: ↑Sat Nov 30, 2019 8:15 pm The Duolingo platform does Gaelic no favours. The "fridge magnet" type activities drop out punctuation marks, which is one thing, but Duolingo considers apostrophes at the start and end of words to be punctuation marks. Getting your head round all the different meanings of a and a' is hard enough without Duolingo rendering every a' into an a....

Well, you are convincing me even more to stick with my plan of continuing with LearnGaelic + Anki. Neither of these makes money out of other people's work and LearnGaelic (so far) has not received such negative reviews. Having tried a few Gaelic series, I really didn't want to change again anyway, so this is a good reason not to.

Níall Beag wrote: ↑Sat Nov 30, 2019 8:15 pmI haven't seen a single report or academic paper discussing any significant change in the sales of Irish learning materials or the sign-ups to Irish courses. To me, that kind of says it's pretty much inconsequential.

That's not an unreasonable point, but I wouldn't totally write it off on those grounds alone because, apparently, only a small minority of users are actually in Ireland, so we might not see any significant report in the UK detailing such impact (quote from the Irish Times - Thu, Nov 27, 2014):

Approximately 64 per cent of the Irish language course users are based in the US, while 9 per cent are in Ireland, Duolingo said. An additional 9 per cent of users are in Britain, while 6.5 per cent are in Canada and 3 per cent in Australia.

The Irish Times continues:

An independent study conducted by the City University of New York has shown that 34 hours of Duolingo are equivalent to a full university semester of language education

So, while I shall not personally be continuing with it, there would seem to be a benefit for some users in the Duolingo system, though it clearly does not suit everybody (with the same Irish Times article saying that many children in schools don't like it).

There are already 20,000 users registered for Scottish Gaelic (I'm one of them so make that 19,999!) https://www.bbc.co.uk/news/uk-scotland-50579327. All we can do is wait and see...

Ionatan wrote: ↑Sun Dec 01, 2019 3:05 pm

Níall Beag wrote: ↑Sat Nov 30, 2019 8:15 pmI haven't seen a single report or academic paper discussing any significant change in the sales of Irish learning materials or the sign-ups to Irish courses. To me, that kind of says it's pretty much inconsequential.

That's not an unreasonable point, but I wouldn't totally write it off on those grounds alone because, apparently, only a small minority of users are actually in Ireland, so we might not see any significant report in the UK detailing such impact

Erm.... you know that Ireland is not the UK, right...?
Leaving that aside, I recently did a masters in Computer-Assisted Language Learning, so I'm not talking about reports in the mainstream press -- I'm talking about specialist publications. Even taking into account the slow pace of research and publishing, I personally believe that the academic community couldn't fail to notice any effects given Duolingo's sheer scale.

What do people finishing Duolingo Irish do? If even 1% of Duolingo's Irish students bought an intermediate-level book or CD course, there would be record sales. That we haven't heard of these record sales tells me it hasn't happened.

The Irish Times wrote:An independent study conducted by the City University of New York has shown that 34 hours of Duolingo are equivalent to a full university semester of language education

That's... questionable.

The paper in question was commissioned by Duolingo themselves, and was not published in an academic journal, so was not subject to peer review.

The two researchers named on the paper are lecturers in economics and statistics respectively -- neither has any specific qualification for assessing language-learning tools. My understanding was that it was done as a private consultation by the researchers in question (only one of whom worked at CUNY) and not by the university, although I may be wrong about that. They had previously conducted very similar studies for 3 other computer-based language learning products (Rosetta Stone, Auralog and Berlitz).

I find the conclusions dubious on a number of levels.

The study measured language level by the score on a test called WebCAPE, developed in the 1990s, which is used by a lot of US universities as an "advanced placement test" to give students language credit for language learned prior to starting study (the US university system can be baffling -- all "general ed" requirements and whathaveyou). The test isn't pedagogically brilliant as it doesn't directly measure the students' skills as taught, but its score works well enough as a proxy score because scores correspond reasonably well to scores in more rigorous tests.
...but...
The test mostly consists of short, isolated, individual questions... so Duolingo users are going to be far more comfortable with the test format than your average school leaver coming from classes with longer-form activities, and familiarity with the format and medium of the test usually results in better scores even without a higher base level of skill. Furthermore, the narrow set of skills tested is a lot closer to the skills practised and trained in Duolingo than standard classroom skills -- if you focus only on the skills needed for the test, well, of course you're going to get a better score in the test quicker.

Then of course there's the sample set.

Students entering university are usually under 20... the average age on this study was ~32, and half the participants already had a degree... and a quarter already spoke another language. They were all people who were actively interested in learning a language, when most students in a language beginner's class are reluctant learners -- just there to get their language credit and move on to their major. There's really no valid comparison in there at all.

There are already 20,000 users registered for Scottish Gaelic (I'm one of them so make that 19,999!) https://www.bbc.co.uk/news/uk-scotland-50579327. All we can do is wait and see...

I personally think we'll be waiting a long time and seeing precious little.

Oops... forgot to say...

The one piece of research I'm aware of by an actual linguist with some kind of reliable supervision was a master's dissertation that found university more effective than Duolingo.

While I wouldn't say it was the best dissertation ever, it is still useful as an insight into the software.

Here's something from the conclusion:

One item in particular that was not
realized until after it was too late, was that there were no recordings created of the
German 1A course participants’ speaking section of the test. This was a lesson
learned, and recordings were created for the DuoLingo group, despite the fact that
only one participant was capable of uttering a single sentence in German.

There were only 8 participants to begin with, and only 5 completed.

But still... only 1/5 "was capable of uttering a single sentence in German".

Ouch.

Not a conclusive study by a long chalk, but certainly gives good reason to doubt Duolingo's hype train.

Hate to necro a thread like this but there are a couple of additions I'd like to make.

Firstly I've never much liked how the site seems to de-emphasise overt explanation of grammar - the web version hides these as "tips", while the mobile app (which, let's face it, is definitely the main avenue for people using it) lacks them entirely.

Similarly, the preference for word-picking rather than typing in answers gets pretty laughable, and again the mobile app is massively deficient in not allowing you to type in.

The site also seems to be lacking in any ability to create new sentences for exercise, which becomes problematic when you're like me and you start basically speedrunning the entire thing (the flipside of all that 'gameification') so you barely even pay attention to what the sentences actually are.

The course also is decidedly short, entirely lacking anything on subordinate clauses and skipping any mention of relative and conditional inflections, most of the irregular verbs, and missing out impersonal forms entirely.

Also a much more middling complaint but it's inconsistent as to whether it allows British English or not, notably only accepting "sweater" as a tr*nsl*t**n for geansaidh not "jumper".

Iain Sasannach wrote: ↑Sat May 09, 2020 11:40 amThe site also seems to be lacking in any ability to create new sentences for exercise, which becomes problematic when you're like me and you start basically speedrunning the entire thing (the flipside of all that 'gameification') so you barely even pay attention to what the sentences actually are.

I recall reading Luis von Ahn commenting on this. Apparently whenever they tried to mix up the sentences, people made more mistakes, got frustrated and left the site.

So rather than taking that as a sign that the system wasn't actually teaching language right and revising the model, they just kept going back to repeating the same examples, pushing the learner to memorise individual phrases and sentences instead of generalising the patterns and learning the language.

This is why I get so angry about Duolingo -- they've got the biggest audience in language learning history and if their evaluations were based on educational best practice rather than web design principles, they could have really perfected it by now. Instead, the actual teaching has barely changed in the whole lifetime of the product.

It's a squandered opportunity.

Airson deagh leabhar mun Ghràmar feuch https://gaelicbooks.org/index.php?route ... arch=byrne

The a Out of Stock an-dràsta co-dhiù.