It’s great that everyone can contribute to the library. But I think there should be some quality standards.
I appeal to all providers to be careful. Please, check content before you upload it. I think, providers should be able to check the content before they upload it.
Is it a native speaker?
Does the text fits the audio?
Is the spelling correct?
Am I able to set the level right?
Please do not upload content if you don’t speak the language properly enough to check this out.
It is not fair to expect other members doing this. At first the provider is liable for the quality in my opinion.
This is why I suggested that the uploads to the library should comply with the European Language Framework, so that setting the level of difficulty, is more accurate and consistent. A1… C2.
Beginners: A1 and A2. Intermediate: B1 and B2. Advanced: C1 and C2.
I agree with Vera that all providers should be careful. I would not call this the proiders’ liability, but rather their responsibility. It is in all our interest to have good, interesting content at different levels.
With regards to the issues raised here, my views are as follows.
native speaker only. This is very important.
audio matching text. This is ideal, but some deviation is quite OK and inevitable. I have found some excellent content with missing bits of audio, audio introductions that are not in the text, and deviations in the transcript. As long as these are minor, I do not see a problem.
Spelling and typos. It is not reasonable to expect the provider to check for spelling and typos on third party content, but spell checkers should be used on anything the provider transcribes or writes. I am not bothered by the odd typo, and we can correct this after the fact as we do now.
Levels: The biggest problem was caused by our automatic level system based on word frequency, which we used when we introduced Levels and had to grade all of our content at once. In general, as long as intermediate and better content is not graded beginner, we need not worry too much about the levels. The users can be guided by their interests and their “new words %”.
The Common European Framework is not an issue here. Our providers need concern themselves with it. They can refer to the 6 levels we have (which are largely based on the CEF) and make their best guess. Let us not complicate things.
In the future we hope to introduce a user evaluation system, with comments, which should help. Ultimately, what matters most is how much the learners enjoy the content. In my case, it is the interest of the content, and the quality of the narration that matter the most.
The Common European Framework is not an issue here. Our providers need NOT concern themselves with it. They can refer to the 6 levels we have (which are largely based on the CEF) and make their best guess. Let us not complicate things.
Thank you Steve.
Responsibility is the better expression. Sometimes I’m not sure which English expression is the best.
2) is for me a question of quality. If I add items to the library I make always sure that the transcript fits the audio, and if not I add the missing parts. If you want to be taken serious by institutions you should be more aware of quality.
3) Also a question of quality. I want to rely on what I’m reading. It is learning material. You should consider this. Users expect that learning material is proper. (That is what you get on other language learning websites.)
4) I agree.
What I dislike is the impression that providers do a quick upload don’t worrying about quality and do an upload in 2 minutes. Some members including me work hours on one lesson to offer quality learning.
I think LingQ should not only offer quantity. Now that we have so much quantity for a lot of languages LingQ should consider to focus on quality.
Noted Vera. Ultimately what matters the most is user satisfaction and introducing a system for users to evaluate content will so,ve a lot of problems. Eventually we would hope to eliminate content that got a low score. But that is in the future.
I know that many (maybe most) of you don’t bother about the accent (and pronunciation for that matter). But it’d be nice to have a library with the accent on each lesson (many many lessons don’t even have one set up).
I’m guessing many people upload content in a language that they aren’t even studying and have no knowledge of. So Vera’s requests are maybe too much to ask.
Good point Berta. Providers should add accents where possible and we are going in and adding accents where we can, as we adjust the difficulty levels. I hope other members can do the same.
Every incremental step in improving the library is good, and gradually it will all get better.
This is not a problem but a suggestion: I would like to filter for spontaneous material- conversations or otherwise that are transcribed after the fact, and not read into the microphone.
Ideally, this type of content will be in the conversations and interviews or podcasts types. We will see if this can be standardized across all languages.
Re. quality with text/audio, I don’t think it matters a great deal for higher-level content (certainly it does for beginner content, I agree they should be identical here). Indeed it’s just not practical when you’re importing very large collections. The Yvan Amar series I’m currently working on has over 1,000 lessons and the audio does not always match the text. However, if I corrected every single lesson so they were identical it would take me far too much time. I wouldn’t do it basically.
As an intermediate/advanced learner of French, these differences don’t matter to me. In fact I find that my brain notices them, is more alert to them and I am convinced this improves my overall learning experience.
I am fixing the level of the ItalianLingQ podcast and at the same time adding the accent. The problem is when people with different accents (e.g. an Italian from the North and one from the South) speak in the same podcast. This has been the case for an audiobook I started uploading some months ago. In such a case I leave the field empty. Maybe there could be an option to choose more than one accent in the future.
Michele, what I’ve been doing is either leaving it blank or selecting the accent that is less popular. For example, one conversation between a New Zealander and a Canadian had equal representation of each accent throughout the talk, but the New Zealand accent is more rare in the library.
Fully agreed Jamie, and I also find that when I discover mistakes or strange things in advanced content, this helps me to notice this in subsequent content.
On the other hand, beginner content should have fewer errors.