Reading tool to help you build vocabulary

[sgalant15 0]

Hi !

I'm on a team of people creating a new website called lingocracy.com

We are trying to change the way people learn languages, by promoting the idea that you can learn by reading things that actually interest you!

We have tons of short stories/news articles/full novels/fables at different levels, so you can find exactly what you want. 

We're still building the site and are constantly trying to improve it. As you all know, each language is different, so we're always working out the kinks to make sure it is the best tool for every language. 

 

I would really appreciate you checking out the website and letting us know what you think/if you have any suggestions!

 

(you have to create an account because the website tracks your progress/vocabulary, but it's free and only takes a few seconds)

Thanks!

[roddy 5,168]

Past experience shows that if you don't make the site accessible without an account, you won't get much feedback. If anyone likes it enough to want their vocab tracked, they can then sign up. Or have a guest account anyone can play with...

[Elizabeth_rb 82]

So far the site looks quite good.  I've signed up for an account in Chinese and German.

 

Some suggestions based on my initial impressions:

 

1) I agree with Roddy that people are far more likely to give you feedback if they can have even a limited 'Guest' look around the site.  You could have a publicly available taster with something from each of the three levels (beginner, intermediate and advanced) for popular languages for people to get an idea how it works before they sign up.

 

2) I recommend an 'upload a profile photo' option as opposed to required that someone have a Gravatar account in order to use a pic.  If I sign up for one service, I don't really want to have to sign up for another in order to fully use the first.  Do you see what I mean?  It might seem a small thing, but it's a pain when you don't have a choice.  Actually, this kind of networking everything together, which is appearing all over the shop, is driving me nuts!! There are a quantity of things I can't use as I can't 'sign in with facebook' (I don't have, or want, an fb account) and I can't comment on Youtube anymore unless I consent to everything being publicly visible in a list of activity and/or appearing on my Google+ page. Bleuch!   

 

3) For Chinese, the 'word count' function doesn't work as well as for other languages.  In this case, it's actually character count, which would be twice to three times as high as the actual word count would be.  Whilst it's not practical to make it a real word count for CJK languages (as the same will go for Korean and Japanese, albeit often the other way around as many of the characters they use are syllables and so the 'word count' could then be somewhat lower than it should be), maybe changing the signs to 'number of characters' might be more accurate?  Again, just a small thing, but might be helpful for some to get a clearer idea of what they're looking at.

 

Anyway, I plan to use the site, so you've got one new member at least.  

 

A question:  Who decides the level of the text?  The uploading user?  Are there some guidelines somewhere?  I noticed that one of the users uploading Chinese has put a 20 character recipe in beginners level.  I wouldn't call that beginners as it's abbreviated etc.

[jasoninchina 45]

I actually came across your website last week on duolingo. After looking at the German section, which I thought was very well done, I gave the Chinese section a gander. The only issue I had was that, because I'm an intermediate learner, I had to go through and tell it all the words I already knew. And if I can assume that my vocabulary is around 2k words, well...I think you can see my problem. I attempted to highlight entire sections as 'known', but it then treated the sentence as a single word. So, I don't know. I recall there being a button to mark all words as 'difficult' (am I remembering that correctly?), but if there was a button to do the opposite (known), then perhaps one could do that and just go back and pick out the words they didn't know. I think the larger issue is that you have to tell the program which characters constitute a word in any given situation.

 

Perhaps I missed something, but if not, then there you have it.

[Elizabeth_rb 82]

Hi Jason!

 

I'm in the same position and what I'm doing is just quickly reading through the 'beginner' level texts and, with the exception of a handful of words I wasn't sure of at the time, I just used the big tick function at the bottom of the page, which marks the whole lot left in blue as 'known'.

 

For the individual words, in the dialogue box that comes up for clicking on an unknown word, you click on the down arrow next to where it says 'Difficult' and select any more appropriate option from the list there.

 

Hope that helps!

[Elizabeth_rb 82]

Something else about the site I've noticed is that on the Chinese section, the words are not always split up correctly in the texts so as to give the right translation of the meanings.  For instance, on the text I've just read, 不知道 is split into 不知 and 道 for translation, instead of 不 then 知道。I noticed that with a lot of the character groupings as I began to realise how the site works. 

[imron 5,124]

I will also add that logging in silently fails when cookies are disabled.

[renzhe 1,675]

Perhaps using widely available HSK vocabulary lists would make sense so that you don't have to enter each word separately.

You could enter the approximate number of words you know, or your rought level (novice, elementary, beginner, lower intermediate, upper intermediate, advanced), and it would pick from the HSK lists.

It could make a good bootstrapping process for new users.

[_Yan 0]

First I'm grateful for all for those feedbacks, that's exactly what we need at this moment to improve things.  I also work on Lingocracy : ), here are few comments and answers.

 

@Roddy, Elizabeth_rb

Thank you Roddy and Elizabeth for the suggestion, I guess you are right about making the text readable without account. Let's put this on the todo. We are also making a video presentation, that way you can get a better idea of what's about before deciding to create an account or not.

 

@Elizabeth_rb

Sure, I also agree that we should allow uploading profile picture. We didn't had time yet, so in the meatime we decided to use Gravatar, but we are definitely going to allow picture upload soon.

 

As for the "word count", I'm not sure to completely understand your comment. Here is how it works at the moment. The "word count" is based on words, not on chracters. For example,  "你好" count for 1 word (when grouped). If you later also add "你" alone, it count for another word. At this moment, in your vocabulary list you will have two words 你好 and 你. For CJK, it may be possible to provide another useful information about how many characters you know. In the previous case, you know 2 characters 你 and 好. 

 

"Who decides the level of the text? The uploading user?"

Yes, the user who add the text sets the level. So, the accuracy depend on the that user's judgment. If a text has a completely wrong level, you can send us a report and a moderator will modify it to make it more accurate.

 

"Are there some guidelines somewhere?"

Not yet, but we are working on that : ) We truly lack of tutorials, we will put more priority on this.

As for the 20 characters text you are suggestion, I temporarily unpublished it since it has some quality content.

 

Yes, we know well this segmentation problem (不知道 beigne segmented as 不知 and 道), at the moment, the rule is very basic, it group from left to right. So that's not always accurate. We know this issue quite well and are working to solve that problem. This is a bit tricky to deal with since we need to set some contextual rules, like when 不知道 is met in a text, spilt them this way 不 / 知道.

In the meantime, if you delete the word 不知 (popup>more>delete) and refresh the page, you will get the right segmentation 不 / 知道, but the downside of this is that if you meet 不知 alone, they won't be grouped since you have deleted the word. I know, this is not perfect.

MDBG search for 不知道 seems to detect that kind of situation correctly, I'm wondering what kind of algorithm/rules they use to deal with that.

 

@jasoninchina

As Elizabeth_rb commented, the button at the end of the text will mark all new (blue) words as know. That should be especially useful in your case. You can mark many words as know fairly quickly by reading few texts and using this button.

 

 

 

 

 

 

[Elizabeth_rb 82]

Thanks for your comments, Yan!

 

Something else I wondered about:  I've just been using the 'Practice' function.  I notice that, if you get am exercise on your 'difficult' list right, it moves to the 'medium' and vice versa.  However, all my 'easy' ones have moved to 'medium' too, regardless of how many times I got them right and no 'mediums' have moved to 'easy'.  Not sure if I'm talking rot (eminently possible), but if my observations are correct, there may be a problem...

 

Enjoying the site anyway.=)

[Silent 374]

 

"Who decides the level of the text? The uploading user?"

Yes, the user who add the text sets the level. So, the accuracy depend on the that user's judgment. If a text has a completely wrong level, you can send us a report and a moderator will modify it to make it more accurate.

To me this rings alarm bells.

- user judgement (of 1 user) on level is imho pretty useless as it's quite subjective. A more accurate level might be obtained by multiple user judgements (e.g allow every user that read it judge it) or by a more objective measure like number of different words and/or a  weighted average rating of the vocabulary compared to a general frequency list. You might even personalize it by taking the users known vocabulary in consideration.

- If users can upload texts and you make them available to all users, how are you protecting yourself from all kinds of copyright claims?

[Elizabeth_rb 82]

Mmm, copyright's an interesting point as, in order to be decent texts, they need to be authentic....

 

Some further thoughts on the exercises:  You could have a method of selecting exercises rather than random either translations (which would be better going both ways instead of just one too) or gap filling.  Also, the gap filling could use some attention as some of the other vocab used in the sentences may not be known by the learner.

[_Yan 0]

Elisabeth_rb:

Something else I wondered about: I've just been using the 'Practice' function. I notice that, if you get am exercise on your 'difficult' list right, it moves to the 'medium' and vice versa. However, all my 'easy' ones have moved to 'medium' too, regardless of how many times I got them right and no 'mediums' have moved to 'easy'. Not sure if I'm talking rot (eminently possible), but if my observations are correct, there may be a problem...

Thank you Elizabeth for pointing out this problem, we will check this level issue.
We have also planned to to rework the practice section soon. It seems a bit too difficult for some some users, especially for languages like Chinese or Japanese where you may not understand half of the sentences displayed to you.

Silent:

To me this rings alarm bells.
- user judgement (of 1 user) on level is imho pretty useless as it's quite subjective. A more accurate level might be obtained by multiple user judgements (e.g allow every user that read it judge it) or by a more objective measure like number of different words and/or a weighted average rating of the vocabulary compared to a general frequency list. You might even personalize it by taking the users known vocabulary in consideration.
- If users can upload texts and you make them available to all users, how are you protecting yourself from all kinds of copyright claims?

Yes you are right, this has some flaws and may lead to some inacurate text level.
Indeed an automated level evaluation may be the solution, we have already planned to implement a such system soon : )

As for the copyright, at the moment, we suggest to add only free to use content (public domain, creative common, ...). Otherwise, the content may be removed any time (e.g. after a copyright infrigement report). At later time, we may provide copyrighted content by building partnerships with content providers. Hope this answers your question.

Elizabeth_rb:

Some further thoughts on the exercises: You could have a method of selecting exercises rather than random either translations (which would be better going both ways instead of just one too) or gap filling. Also, the gap filling could use some attention as some of the other vocab used in the sentences may not be known by the learner.

Actually, this is not random, the "translation as question" is a fallback when we don't have sentences enough sentences. But as mentioned above, we will definitely rework and make the practice section better : )

[Elizabeth_rb 82]

I like the translation questions.  Actually, I like them better and think they have a lot of value for me personally.  I understand that having the translation questions available in many language pairs may be a challenging project though.

[_Yan 0]

Do you mean, you prefer translation kind of questions rather than "fill in the blank sentences"?

Actually, this shouldn't be that difficult, especially for languages like chinese where we have a lot of good dictionaries.

[Elizabeth_rb 82]

Yes I do. Especially as one doesn't always know allow the characters in the sentence.

[roddy 5,168]

Any updates from lingocracy or our members? Anyone become a loyal and regular member?

[Ania 52]

I just signed up for the service last night, so I don't yet have that much constructive criticism, but it looks like it could become one of my favorite tools if it works as I imagine it does I like the fact that we can import online articles and then use the website's tools to read them. Once I fiure out how it works and all I will try to post some more insightful comments.