I've been on Duolingo for about a month now, and they must be getting a bit sick of me because I submit a lot of beta feedback---but they're very responsive and usually pretty good about fixing things, sometimes by the next day, including when I type a translation that was correct but marked wrong by the program.
I agree with the OP that it's overall a great site, although there are a few lingering things that they seem unable to fix, mostly minor; for instance, the "practice" is too irritating to keep up with because its randomness is broken and literally every batch of twenty includes at least six to eight occurrences of the sentence "Sie liest die Zeitung" and/or "Der Mann liest die Zeitung". Argh.
But the larger, systemic problem with it is that I don't think it works as advertised for raw newbies to the language. It's perfect for those of us who dimly remember the "rules" of a language and just need a lot of structured practice to get up to speed. But by presenting nothing but example text to translate, they're basically asking the newbie to induce a complete grammar of the language without any guidance, and while children learn their first language(s) that way, adults don't.
(Example: in German, the ending that attaches to an adjective can be one of six or seven things depending on the case of the phrase, the gender and number of the noun it attaches to, and the definiteness of the article that commands the phrase. None of this is explained within Duolingo; it just throws a bunch of example sentences at you, never talking about gender, number, case, or definiteness (two of which are concepts that aren't even relevant to English grammar). That's... tricky.)
On the crowdsourcing translation front, it has an incredible strength, which is that native English speakers will be great at evaluating whether a sentence is fluent, valid English. But it has an incredible weakness, which is that native English speakers who are just learning German are much less good at evaluating whether a sentence is a valid translation of what the German sentence says. I've seen quite a few situations where a particular translation has been heavily upvoted despite missing some crucial part of the source sentence's meaning (and in a few cases, meaning exactly the opposite).
But it's a work in progress, of course. It seems to have a lot of promise, so perhaps they can surmount these difficulties.
Another one of those "why didn't I think of that" ideas. I haven't tried this, but it seems like it would be better than the traditional Anki/Pimsleur/Memrise style in that it would teach (or reinforce or mandate) context. Anyone can memorize a list of words, but the difficulty comes in putting those words into context. Duolingo might help that (or might allow beginners to completely destroy the translation, I don't know their algorithm).
I actually thought of a similar idea before, while I was meditating in rural Philippines attempting to figure out a viable business idea and trying to avoid a return to the rat race in the US. Unfortunately, the idea seemed too daunting to take on by myself.
I don't know if Duolingo already has plans to incorporate it, but I had also originally thought that 'word completion' functionality would assist in the process of learning a new language. For example, a beginner trying to learn the Tagalog language (but knows a few simple words) could begin by typing the word 'Gusto', and then typing a space after the word would produce a drop-down with possible words that could follow, such as 'ko', 'niya', 'nila', 'mo', etc. In essence, it would be 'Intellisense' functionality that provides grammatically correct word-completion to help students learn sentence structure -- which is a major part of learning a foreign language, as opposed to just translating / memorizing fragments (e.g. words). I was even to the point where I was considering how an AI approach using an inference engine / chaining might make this
process more efficient.
Anyway, I e-mailed Luis von Ahn about it, but he was probably too busy to read or react to my e-mail. I am just happy to see someone else was able to make the idea a reality -- another testament that brilliant ideas are virtually worthless without brilliant execution.
I think describing it as flawless is a bit much. It is a great improvement over the status quo. It allows you to put what you've learned to immediate practical use. It is a brilliant example of a satisfying non-monetary trade. It takes a desire for knowledge, provides that knowledge, and allows the recipient to "pay back" the favor by using that knowledge to assist Duolingo. It makes learning fun and enjoyable.
That said, there are glitches. It can be very inflexible at times. There's more than one way to skin a gato, and Duolingo can be very forceful on it's interpretation. I'm also not sure how useful it can be when it comes to learning a new language from scratch. I already had a few years of Spanish before I started, so it was a lot of fun brushing up. However, learning a new language without a human being there to help you and talk you through it can be rough. Duolingo tries to compensate through a forum/qa solution, but there's nothing like talking to a person in the flesh.
That said, it is miles and miles above stuff like Rosetta Stone, etc.
[+] [-] blahedo|14 years ago|reply
I agree with the OP that it's overall a great site, although there are a few lingering things that they seem unable to fix, mostly minor; for instance, the "practice" is too irritating to keep up with because its randomness is broken and literally every batch of twenty includes at least six to eight occurrences of the sentence "Sie liest die Zeitung" and/or "Der Mann liest die Zeitung". Argh.
But the larger, systemic problem with it is that I don't think it works as advertised for raw newbies to the language. It's perfect for those of us who dimly remember the "rules" of a language and just need a lot of structured practice to get up to speed. But by presenting nothing but example text to translate, they're basically asking the newbie to induce a complete grammar of the language without any guidance, and while children learn their first language(s) that way, adults don't.
(Example: in German, the ending that attaches to an adjective can be one of six or seven things depending on the case of the phrase, the gender and number of the noun it attaches to, and the definiteness of the article that commands the phrase. None of this is explained within Duolingo; it just throws a bunch of example sentences at you, never talking about gender, number, case, or definiteness (two of which are concepts that aren't even relevant to English grammar). That's... tricky.)
On the crowdsourcing translation front, it has an incredible strength, which is that native English speakers will be great at evaluating whether a sentence is fluent, valid English. But it has an incredible weakness, which is that native English speakers who are just learning German are much less good at evaluating whether a sentence is a valid translation of what the German sentence says. I've seen quite a few situations where a particular translation has been heavily upvoted despite missing some crucial part of the source sentence's meaning (and in a few cases, meaning exactly the opposite).
But it's a work in progress, of course. It seems to have a lot of promise, so perhaps they can surmount these difficulties.
[+] [-] freehunter|14 years ago|reply
[+] [-] davidpoarch|14 years ago|reply
I don't know if Duolingo already has plans to incorporate it, but I had also originally thought that 'word completion' functionality would assist in the process of learning a new language. For example, a beginner trying to learn the Tagalog language (but knows a few simple words) could begin by typing the word 'Gusto', and then typing a space after the word would produce a drop-down with possible words that could follow, such as 'ko', 'niya', 'nila', 'mo', etc. In essence, it would be 'Intellisense' functionality that provides grammatically correct word-completion to help students learn sentence structure -- which is a major part of learning a foreign language, as opposed to just translating / memorizing fragments (e.g. words). I was even to the point where I was considering how an AI approach using an inference engine / chaining might make this process more efficient.
Anyway, I e-mailed Luis von Ahn about it, but he was probably too busy to read or react to my e-mail. I am just happy to see someone else was able to make the idea a reality -- another testament that brilliant ideas are virtually worthless without brilliant execution.
[+] [-] maratd|14 years ago|reply
That said, there are glitches. It can be very inflexible at times. There's more than one way to skin a gato, and Duolingo can be very forceful on it's interpretation. I'm also not sure how useful it can be when it comes to learning a new language from scratch. I already had a few years of Spanish before I started, so it was a lot of fun brushing up. However, learning a new language without a human being there to help you and talk you through it can be rough. Duolingo tries to compensate through a forum/qa solution, but there's nothing like talking to a person in the flesh.
That said, it is miles and miles above stuff like Rosetta Stone, etc.
[+] [-] adrianp|14 years ago|reply
[+] [-] micah3344|14 years ago|reply
I'm sure it will get really big once they open it up.