Show HN: Readlang – Learn a language while you surf the web | Hacker News Reader

Show HN: Readlang – Learn a language while you surf the web

readlang.com

178 pointssteveridout83 comments

Creator of the Readlang here. I started this at the very end of 2012, and save for a few months contracting have been working on it full time since, doing everything myself. Growth has been slower than I’d like but just enough to keep me motivated.

It's a freemium webapp, originally designed to reduce the frustration of practicing my Spanish by reading novels, and later adapted to work on any webpage as a browser extension.

I always appreciate it when people share details about their business, so here are some numbers from the first 5 months of 2015 (Jan - May):

Google Analytics - Sessions: 120,000, Unique visitors: 53,489

Signups - 13,658

Revenue - $4768 (average of $953 / month)

Not spectacular, but when I look back, it took 16 months to make the first $1000 (http://steveridout.com/2014/03/22/readlang-my-bootstrapped-l...), and right now it’s making more than that every month! It’s a long road but I’m very excited about it’s future.

Any questions or feedback, please fire away!

Wow, these are the stories we need to hear more of!! I'm a bit tired of the overnight success stories getting all the attention. If such an awesome product has taken so long to make something in the region of 1000 dollars a month then perhaps creating a startup is not as easy as so many people make it out to be...

Who said it was going to be easy? ;-)

It's cliche I know, but most "overnight successes" involved years of hard work.

Totally agree. For most people, even on HN, it is really hard to identify with all those multi million dollar fundings and billion dollar buy outs. This on the other hand, shows what is possible by one person and how it is not that easy to build something nice and useful.

I hope OP keeps publishing posts on what worked, traffic stats, revenue stats etc.

This is another site that was bootstrapped (I believe) http://www.forvo.com/ and seems to be doing quite well (though the quality is debatable sometimes)

I really like the video feature. Seems like a great way to improve listening skills.

Do you do some kind of text processing, before or after sending the words to Google Translate? Do you analyze the translation mistakes, in order to iteratively improve the results? How much do you pay for the API?

I don't do any processing - the phrase goes directly to Google.

Users can edit the translations so I'm accumulating data which could be aggregated in future to improve translations, but not doing that at the moment.

This is interesting.. I was reading this cool thread a couple of days ago (some really useful stuff):

https://news.ycombinator.com/item?id=5965081

and stumbled across your comment. I thought it was neat (I'm using -not- TransOver for the moment).

Here's what I can think of, off the top of my head, but first some background:

I'm acquiring Spanish right now and simultaneously scratching the surface of German, Russian, Hebrew, and Persian. They have a lower priority, but I'm making progress.

As you may have noticed, English is not my first language. I learned it out of necessity (resources I needed were in English).

I read a lot, and in the beginning, my head hurt as I read stuff written in English. I had MediaDico on my computer to translate from English to French.

I struggled with idioms (I needed to read several examples of an idiom to get the gist of it)... But I can pin-point the exact moment where my learning was about to soar: I was in the living room and I decided that my computer experience had to shift to English. This means that even when I was to search for stuff, I'd do it in English. My brain bled. Something trivial to search for in a language I knew, something I already had the keywords ready, not in my mind but in my fingertips, to search Google for, would take me way longer to think about, formulate, and come up with keywords for, in English.

What would be cool is this:

- A user doesn't have to click on on a word to translate it. The extension makes a sort of histogram of the most used words or expressions (a frequency list of the page), and automatically translates them. Since they're the most used words or expressions on the page, the person will be exposed to them often: "accidentally on purpose" rote repetition. This will be useful for the next step:

- The extension _remembers_ (keeps a list of) which words (should) already have been learned by the user (which words he has been exposed to the most), and in subsequent pages will display them in the target language without the user doing anything. They will no longer be displayed in the user's language.

- The first message it displayed was that I had to drag some link, etc.. Why is that?

- I had a message saying "This is a very large page". Maybe the extension can translate stuff the user is most likely to see first, and as the user browses, it does stuff.. Think Python generators.

All my best !

Hmm, sorry if I'm misunderstanding, but from your suggestions it's not clear if you realise that the extension is intended for reading in your target language and translating occasional words to your native language.

e.g.

> ...and in subsequent pages will display them in the target language without the user doing anything. They will no longer be displayed in the user's language.

The idea is that to learn Spanish, you read Spanish webpages, so everything is displayed in the target language at first, and you click to get the English translation.

Detecting words which are unusually frequent in the page you're reading, and automatically adding a translation to the first instance is a cool idea.

> The first message it displayed was that I had to drag some link, etc.. Why is that?

I think you're referring to the bookmarklet that needs to be installed to translate webpages, this is an alternative to the Chrome extension and works on almost all browsers.

> I had a message saying "This is a very large page"

Yeah, this is pretty ugly, should think of a nicer solution.

>Hmm, sorry if I'm misunderstanding, but from your suggestions it's not clear if you realise that the extension is intended for reading in your target language and translating occasional words to your native language.

I think the idea is that to learn Spanish, you read not only Spanish webpages, but even the webpages you read that happen not to be in Spanish can become so.

The way you intend your extension to work assumes the user is already on a Spanish webpage. This assumes the user has overcome procrastination and reads pages in his target language often.

Now, as a language learner yourself: What percentage of all people who want to learn a language does that case represent? How many overcame procrastination and practice everyday?

These are the users who don't need help and would be fine.

The biggest problems in language learning are two: - Procrastination. - Weak, infrequent and/or sporadic exposure.

Just imagine if in addition to catering to those serious users the way you do now, you can annihilate learners' worst problems and serve the whole 99% of people who've had "I want to learn X" on their mind for 5+ years but have been too intimidated, or think it's too hard..

They're on a page in English sprinkled with foreign words in context. They get value with every single page they read. If they don't understand a word, the only effort they make is click on it. If they somehow read a sign in the streets or see something in TV in the target language and they recognize it, they will have a smile and that will be a turning point. This is where they'll feel "It works". It has more significance because they've tried to learn that language years ago with a bunch of arcane methods, and weren't even able to do the smallest things with it.

Discomfort threshold to the prospect of reading pages in the target language is decreased by the page.

Your extension isn't one that translates words on a webpage, it is as an extension that minimizes friction to learn a langue. It can do it in a variety of ways and already does a great job with the flash-cards. Everybody hates doing the legwork and you're getting them to "Read it and it gets saved".

>I think you're referring to the bookmarklet that needs to be installed to translate webpages, this is an alternative to the Chrome extension and works on almost all browsers.

It's just a guess, but I think a lot of people would benefit if this step gets streamlined. It's already cool that it doesn't require restarting the browser, and making it even more seamless would be awesome. A user has only to click on the install button on the Chrome extensions pages and that's it. No further intervention.

This is for more accessibility. There a lot of people who don't know what a bookmark is, or bookmarklet for that matter.

All the best,

Turning English websites into a Spanish learning experience is a neat idea, because as you say it fits into peoples current habits. It's definitely worth exploring but currently Readlang's focus is learning from real native texts. I'm trying not to lose focus and to Do One Thing Really Well (TM).

I agree that procrastination is a huge problem. I'd prefer try to tackle this by introducing more opportunities for learners to encounter Spanish texts, one way to do this would be to create a browser homepage with recommended Spanish websites, texts, videos, and practice exercises.

> It's just a guess, but I think a lot of people would benefit if this step gets streamlined. It's already cool that it doesn't require restarting the browser, and making it even more seamless would be awesome. A user has only to click on the install button on the Chrome extensions pages and that's it. No further intervention.

It's already very simple to install the Chrome extension, and once installed I think it's correct that it requires a click to activate on a webpage since it does take resources and slow down the page, so I wouldn't want it working on every web-page.

For the bookmarklet - I completely agree that it isn't intuitive, the experience on mobile browsers is particularly horrible, but web-developers have limited control over this without better support from the browsers.

FYI, I discovered this a week or two ago and when I tried to pay using PayPal something timed out and I was unable to complete the transaction. I tried several times, and I don't think it was a PayPal issue. Eventually I just gave up.

Love the site, though.

That's a bit worrying - do you know at what stage it stopped working exactly? Please get in touch (steve@readlang.com) with some more details and I'll be happy to upgrade you for a couple of months for free.

I'm vaguely trying to learn spanish using Duolingo, but haven't been very diligent. I was hoping your chrome extension could be a more passive way of learning that integrates into what I already do, and which would let me stick with it as a result.

Unfortunately, the current version isn't. Unless I'm using it wrong, here's the process of learning spanish with your extension:

    - I'm on a page
    - I decide to learn spanish
    - I click the extension button
    - i click random words and see the spanish word for it

There's a lot of friction (I have to steps 2 and 3) for not a lot of pay off (I can choose to click words and see what the spanish version is). Also, the fact that it gets in the way of every link means I immediately turned it off.

May I suggest an alternate version:

    - on every single page the extension is active
    - pick some number of words (1% maybe?) to translate into spanish, while keeping the text in english. Underline them in green and let me mouseover if I don't understand
    - when I'm reading the text, I will come across these spanish words and learn them in context

Much lower friction, and doesn't get in the way so I won't turn it off.

WDYT?

So something like this?

https://chrome.google.com/webstore/detail/language-immersion...

Yes, awesome, thanks! It doesn't get the translations perfect AFAICT, but I think it's good enough to keep activated!

Unfortunately, I don't think you can learn, for example, Spanish by translating pages to Spanish. You have to read pages in Spanish and translate from it, otherwise you're learning Google-Translatese, not Spanish.

There's a lot of nuance, idioms, etc that translations can't give you, not to mention that they're usually, very, very wrong.

At least personally, I've always been horrible at remembering nouns/verbs in Spanish. I never did listen to any of my teachers and spend time memorizing the 10 or so words per week that I was supposed to. Even still I don't listen to Duolingo and practice consistently.

My lack of vocabulary is directly related to my conversational fluency. It would definitely be nice to randomly see Spanish words in an English sentence while I'm going about my day. Next time I'm trying to speak to someone in Spanish, I may remember some of these words, be more confident about my sentences, and keep the conversation going longer. Honestly, even if I use the words slightly incorrectly (because they are Googleese) at least I'll have said something and have more conversation experience to learn from.

In Portuguese, I just used ReadLang (under C2, Master setting, w/o signing up) and saw the word 'conduzisse' which translated simply as 'lead'. The verb conduzir can also be to conduct, to guide, and to drive. The verbal tense is first/third-person singular imperfect subjunctive, which is important in meaning and context. What I wish to say with my comment is that perhaps the app is best for learning before reaching the advanced stage because this is when we pass from Translatese to real fluency.

I did try Italian, in which I'm intermediate, and found the app useful. The song section is a nice touch, too.

You may find "QuickTranslate" Chrome extension a good fit. It simply translates either the word your cursor points to or a selected text fragment - https://chrome.google.com/webstore/detail/quick-translate/nf...

I've had this reaction a few times, especially since a LifeHacker article (http://lifehacker.com/readlang-helps-you-learn-a-foreign-lan...) which compares it to the Language Immersion extension (https://chrome.google.com/webstore/detail/language-immersion...) which works exactly as you describe.

Readlang is intended to be used in reverse - you should be using it to read Spanish articles, and clicking on the words and phrases you don't understand to translate them to English. Learning from actual native Spanish texts is vastly preferable to machine translated Spanish. Of course the downside is that you need to:

a) have some ability in Spanish already

b) develop a new habit of visiting Spanish websites. You can get an idea from the most popular sites visited by Readlang users here: http://readlang.com/es/links (I've been thinking of exploring another idea for an extension which hijacks the Chrome "new tab" page to encourage this habit)

It might be interesting to offer the reverse: incremental translation of source material that's in the reader's known language, into the target language.

That is, simply by switching into the mode, each page the user visits would have some of its words translated to the target language. They'd have distinctive styling, and a simple hover or click would show the original word. You'd always be reading a mosaic of both languages

As they hover/click for clarification less often, more and more words would be target-translated... building the new language vocabulary over time.

Of course, this doesn't teach the new language's syntax/ordering... but even that could maybe be incrementall mixed-in over time. Perhaps when there's a clear 1:1 sentence map, a sentence could (with some probability linked to how many words have already flipped) flip to the target-language's order-of-presentation, even if most of its individual words are still native.

Sounds like you might like the Language Immersion chrome extension: https://chrome.google.com/webstore/detail/language-immersion...

Great to see your language learning website! I'm developing similar ideas all the time, but I'm having trouble earning the first dollar.

A few questions: Do you feel trying to do every language is effective? Do you have trouble with any particular languages? Do you feel google translate is good enough for learners?

On your site you seem to say audio was a mis-step. Will you be giving it another go?

Is doing every language effective? Most of them cause me no/very little extra effort so I figure why not. The danger is that learning some of the languages gives a worse experience and may negatively impact the users' impression of the entire product. e.g. the site isn't great for Japanese and Chinese which should have extra features such as Furigana/Pinyin, and smart word boundary detection (they don't use spaces to separate words).

Is google translate good enough? Depends. It's certainly not good enough to trust 100%, which is a big issue for beginners. But once you have a basic level of understanding you can generally tell from the context whether a translation makes sense or not. If you're not convinced, you just need to open the extra popup/sidebar dictionary, and I encourage users to check and correct any translations the first time they encounter them in the flashcard section.

I added the YouTube feature very early on, when the rest of the site was rougher round the edges, and felt I should have concentrated more on the core reading experience. Now though, I'm glad the feature is there, it's great for listening practice, some people get satisfaction from syncing and sharing videos, and it's one extra way to impress people when demoing the site.

Regarding plain audio - this isn't on my immediate TODO list but I agree it would be cool. Personally I'd love to listen to podcasts + transcriptions to help me learn.

It is pretty impressive and very usable. I see a great future for it.

My comments:

1 - it is not clear which words go into "my words" e.g. I mark one, then the one after, but I don't want it as an idiom, I just want two separate words.

2 - this causes some confusion when going to the flash cards I get (imagine it's in a foreigh language) "the plane", "the plane is" "plane is taking off" "plane is taking of the ground" - each in a single "flash card item". I'm sure it's due to me abusing / testing marking words

I don't have UI / UX solutions to both issues sadly, I'm sure you gave it a lot of thought.

Otherwise, this is a great, fully functional website, and I wish you will make a lot of money. I'm tempted to become a paying user. Looking at DuoLingo's recent valuation, I would say that there is a market for this.

Great example of a real bootstrapped SaaS.

Hope you get to see it grow further. Sounds like a valid YC submission if you ask me, worth a shot...

1 - good point, the best way to achieve this at the moment is to click to unhighlight the first word before clicking the second one. Even after clicking to unhighlight, the word will still remain in your word list, the idea being that it usually doesn't hurt to collect too many words, and the decision on which words to keep, and which to discard shouldn't take up valuable brain cycles while reading. The flashcards will prioritize the high frequency (most useful) words for you and if any extra words you don't want appear at that stage it's quick to delete them.

2 - the way it's meant to work is that if you select "the plane", then extend it to "the plane is", the larger phrase will overwrite the shorter one. It's definitely true that testing the translation feature can result in a lot of garbage appearing that you don't want.

I could add a collect words ON / OFF setting but obviously that would complicate the UI, so I'm resisting it.

Thanks for your feedback. I actually applied to YC with this a couple of years ago but the product was very immature, I'm a solo founder and I didn't get invited for interview. I'm more inclined to bootstrapping rather than taking VC which seems to be the path most (all?) YC companies take. I may re-apply for summer 2016 though since it sounds like an amazing experience.

Amazing job, I'll become soon one of your paying customers :)

After playing with it for a while a first suggestion. I noticed that it's possible to export the frequency of the words so you have this information, for my use case it would be very useful to have this information in the word list (when the word is expanded).

This is because when I'm reviewing the words in my word list and deleting those I feel are not very usual so I can concentrate on those that are, however this is done just by guesswork as I don't really know which words are more common, having easy access to the frequency of the word in the word list would allow me to do it in a more efficient way.

You shouldn't feel the need to delete infrequent words like this manually.

The flashcards use the frequency to prioritize the useful words first, so provided you have a large enough pool of translated words you will never see the really infrequent ones. (The exception is that if you star any word, it will be bumped to top priority)

Agree it could be nice to show the frequency information in the word list somehow, but it needs to be subtle / hidden enough not to clutter the UI.

There’s a lot to like here. It reminds me of this great article I read about someone learning French by reading Harry Potter many years ago. I can’t find the first reference to it, but it seems a lot of people take this approach now and numerous articles have been written about this method. https://www.google.com/search?q=harry+potter+language+learni...

I was delighted to see that you even wrote about it yourselves on your blog. http://blog.readlang.com/2014/03/08/learn-languages-with-har...

Your content marketing is great, but you don’t do it enough. Good startups ship both code and marketing efforts on a regular basis…ideally, you should shoot for something going out weekly.

It’s great that you shared your numbers, but snapshots are less helpful than data or information that shows momentum. When I’m trying to figure out if a company is doing well, numbers at a fixed point in time usually isn’t enough data to help me determine what the company looks like in the future.

Early stage startups should try for at least 10%/weekly growth on your core metric. If you’re not hitting that, make sure you’re only spending time/energy on tasks to hit those numbers.

Onboarding experience can use some work. There’s not enough guidance in the app to make me feel confident that I can use your site to learn a language on my own. For instance, when I first login, I had no idea what I was supposed to do first.

http://cl.ly/image/253K373i1M2Y

Even though I saw Upload Text button and the Web Reader link/button…I naturally clicked on the first story…which lead me to a page with even less guidance. I’ve never learned a language by reading and I couldn't find any help for best practices to using your site or how I’m supposed to turn this exposure into language proficiency.

Highly recommend reading Kathy Sierra’s article about how learning isn’t a push model. Might give you some ideas: http://headrush.typepad.com/creating_passionate_users/2004/1...

The highlight to translate UX feels promising, but I found myself wanting a way to see the full english text side by side with the translated portion.

Why make me click twice to learn words through the flashcard interface? (tab then button) Also, if you want me to do it everyday, offer me a daily email or sms to practice. Each one of those emails is an opportunity to upgrade them down the road.

http://cl.ly/image/343r002E071K

Speaking of upgrade triggers, it sure feels like you make it hard to upgrade or pay right away right from the get go. I think it’s because you’re shooting to upgrade people who are using it the most. That seems way too nice. :)

As far as your feature list is concerned on the marketing side…they feel very YOU centric rather than USER centric. http://readlang.com/features

I have to do all this work to figure out why the feature is better for me. If the copy is one step removed from me selling myself, it’ll be two steps removed from me being able to sell it to a friend.

Here’s a link to another Kathy Sierra article talking about focusing on helping users kick ass over showing off how you kick ass (she’s the best). http://headrush.typepad.com/creating_passionate_users/2005/0...

I see on this page that you do have daily reminder email for flashcards…funny, I never found it when I was playing with it.

Anyway, hope this helps. Keep up the good work!

I read a couple Harry Potter books in Spanish a decade ago. It's always helpful to read. Those books are a little hard for a beginner. I made words list for each chapter, which I lost. I think it's helpful to study the words then read chapter.

I don't suppose there are digital copies that can be parsed floating around?

Harry Potter is one of the very few bestselling fiction eBooks that can be purchased DRM free: https://www.pottermore.com

If only more publishers would do this, it would be far easier for me to recommend novels for users to buy to read on my site!

Thanks very much, this is great advice!

(It's 4am here now and my brain is fried so will go through this in detail tomorrow.)

Any plans for Bitcoin integration? My preferred methods of payment are, in order of preference, Bitcoin, PayPal and credit card as a distant last.

Sorry, no plans at the moment. It still seems too niche to be worth adding.

Would love to hear if anyone has achieved significant increase in payments after adding bitcoin in addition to paypal & credit card.

Apart from that one Harry Potter article, I haven't found content marketing to be that effective in attracting new users, mainly because I haven't invested enough time in it. The low hanging fruit I'm going after first is to add more landing pages and product information about Readlang. Later I might try getting language teachers and bloggers to help out with the content marketing, I can't see myself doing it frequently since it takes too much time from working on the product. Definitely need more content for SEO, almost all my search engine traffic queries at the moment contain the term "readlang".

Being bootstrapped, my core metric for the moment is revenue and the graph is looking pretty bumpy at the moment, I may follow up with a blog post digging into the details.

Onboarding - I completely agree, I started playing around with more walkthough hints, using a rabbit character (a la Clippy) that guides you though the process, but abandoned it since adding more complexity felt wrong when the underlying UI should probably be clearer. Still plan to add a nice checklist of items that new users are encouraged to complete.

Love that Kathy Sierra article about learning - remember reading "Head First Design Patterns" about 10 years ago, first impression was how tacky all the clip-art looked, but I ended up really impressed with how engaging it was. At the moment, Readlang is still best for people who already know they want to practice by reading and just want a better, more convenient method. Long term, I want to make make it a slightly more guided and engaging experience. This could include personal recommendations of articles, gamification around both reading and flashcard practice, and social features e.g. write a summary of the previous chapter in Spanish, discuss the text with other users.

I agree a full English translation would be nice for beginner learners, ideally a real human translation, similar to a parallel text. For now the product is targeted at intermediate and advanced learners but definitely something to consider as it broadens to appeal to beginners.

Opt in to the daily flashcard email is currently presented once you finish a flashcard session - pretty hidden - I'll make it more visible! (note: currently all users who've translated a bunch of words get a lifecycle email 2 weeks after signing up with a plain text version of a flashcard)

I'll experiment with adding a way to upgrade from the get go. I was concerned that upselling before users had a chance to play with the product could scare them off, but will try it out!

I'm gradually learning and will work on making my marketing more user focussed. This image really sums it up and sticks in your head: http://www.useronboard.com/features-vs-benefits/

Thanks again, really awesome feedback!

EDIT: slight readability tweaks

Great work here! I wonder if you'd be interested in some form of integration.

I'm a partner at http://green-bridge.org and we've been bootstrapping our stuff for a year now. It's a tool used for over 15 years in the classroom but we only just begun to bring business aspect to it. Grmmr shows visually the grammar of English (we're slowly expanding onto other languages). We're focused just on the grammar aspect of things.

You have exactly what we were planning on building. We would love to add our shapes atop of the translations so it's easier to visualize. A lot of schools and ESL groups in colleges use our tools and they definitely will benefit from using your stuff along with our visuals.

Let me know if this is something you might be interested in.

Looks cool! Some of it seems very intuitive, like the arrow pointing back to indicate the past, or the double arrow to indicate plural. I could imagine a set of symbols like this being a very nice addition to a dictionary allowing fast identification of the part of speech and conjugation.

My TODO list is already way too long with core Readlang stuff to do this kind of integration at the moment, currently I don't even detect whether a word is a noun, verb, etc. It's all handled by google and the external dictionaries. I look forward to following your progress though. Good luck!

I use Rikaichan (which works really well for Japanese, not sure about other languages as I've not tried) It covers a significant portion of this in my opinion. The original is for Firefox, but there's also a chrome port under the name "rikaikun".

That said, I'll take a look at this, maybe it's cool. :-)

I think for Japanese, Rikaichan probably beats Readlang right now because it can detect word boundaries, and probably has other features specific to Japanese that I can't afford to spend time on with Readlang yet.

It would be awesome to improve Japanese support later after it's achieved success with the main European languages.

Yeah, I took a look at your Japanese mode, and while I'm impressed that it works at all, it has some problems with Japanese's high concentration of homographs and homonyms.

First thing I ran into (second character in the text!) was 月, in this case it was part of 夕月(approximately "evening moon"), but it separated the two characters and translated 月 statically as "month" instead of "moon", which it meant in this case.

This stuff can be hard, I've tried for myself and honestly, I'm not up to the task unless I can make a living at it.

Watched the videos to fully understand what this is and wow, just wow. I used to be fluent in Spanish and lost it years ago for lack of use but this is going to be an easy way for me to catch up seeing as how I read so much during the day. Congrats, this is an amazing project.

I tried German, where the first text is the German version of Let It Go, and there are quite some mistakes in the text (wrong capitalization, wrong spaces). Even more evident when the YouTube video itself actually contains the text.

Is the text user-generated?

Yes, the texts are uploaded and synchronized by users. I should really add a way to easily flag up mistakes and contribute corrections.

I like the idea of doing two things at once but you often end up doing both things poorly. There's a high likelyhood that you will learn somethings incorrectly due to machine translation errors, even in popular languages. For other language combinations, English-Chineese for example, Google translate is really bad. Frequent and repeated exposure to a language is the best way to learn. This product could help but learning a language is like loosing weight, there aren't really any short cuts. You still have to work hard.

Learning a language is going to take a lot of time and effort, no argument there.

But not all methods are equally effective so anything that reduces the time or effort required, even if just by 10%, is technically a 'short cut'. And anything that increases your enjoyment of the process is a win, regardless of whether it saves time, since it will help maintain motivation. Losing motivation is surely the most common reason people fail to learn.

The main idea here is to reduce the friction of reading as much as possible, freeing up your attention to focus on understanding and enjoying the text. The machine translation is only used here for words and short phrases, where it's more effective than for complete sentences.

(I'll take your word for it that google translate isn't so good with English-Chinese, I'm mainly focussed on the European languages for now)

I can't select more than 3 or so words at a time before being bugged to go premium, and some words refuse to translate altogether, just bringing up the "go premium" box.

Neat idea though.

If you select 2 adjacent words it will try to combine them into a phrase and you only get 10 phrases / day for free on the free plan. You get unlimited single words for free though so you CAN continue using it heavily without paying if you like, it's just not quite as useful.

Have you played with this number at all? 10 phrases per day seems awfully low in my mind. If I was to download this extension, I would want to use it for a day or two before paying for an upgrade. I imagine you hit 10 phrases quite quickly, and this might turn some people off of the product before they fall in love.

You could be right. It used to be 20 for free, but I lowered it so that more people would actually hit the limit and get the "Go Premium" message. It's hard to know what the optimal number should be. For a while I had a split test running where some users had 5, some 10, and some 20 per day. But there weren't enough users to see a significant difference in conversions.

I think the sweet spot is probably closer to 100... you want people to see how great it is when they use it a lot.

Isn't Google translate API horrible at actually translating? Both me and my wife are bilingual, and GT goes full retard for any non-trivial sentence for both of us.

It's actually pretty good as far as machine translation goes. For long complex sentences it can be misleading, but the use case here is to translate words and short phrases good enough for you to understand the meaning and enjoy the text, and for that purpose it works well 95% of the time. For the other 5%, you can usually tell from the context that it's not correct, and there's a popup/sidebar dictionary you can refer to. Please give it a try and let me know if you think it's useful or not.

How to you avoid falling foul of Google's 'unusual activity' algorithms?

There's another extension (Language Immersion for Chrome) that could've been great - it's designed to be used during regular browsing in your own language, and translates a user-defined proportion of words from your language into your target language. Unfortunately, it only works for a few pages before being blocked.

Not sure what unusual activity you're referring to, I pay for their translate API - I don't know how Language Immersion works but perhaps they are getting around paying for the API by loading a hidden version of the google translate page and scraping it - that's pure conjecture though!

I am saving aggressively to do the same in two years.

How hard is it to get out and be social while literally everybody else is working at a 'real' job?

Not that hard! One trick is to move to a cheaper part of the world, I'm currently spending most of my time in Madrid which is a lot cheaper than my old home of London, so socialising is easy financially. The tougher problem has been pushing through and continuing when the usage has decreased for 2 months in a row, but things always seem to turn around and improve, and as long as things are moving in the right direction long term, I have a roof over my head, and enjoy what I'm doing, I don't see the point in stopping. (My thoughts may change slightly in the coming months since we have a baby on the way!)

EDIT: slight tweak

Wow, just wow...

In my mind I could never afford having a partner while still bootstrapping, much less having a kid! That's very interesting to hear. Best of luck to you.

My partner works, so up till now we've been 50/50 on everything, and in that case, having a partner reduces costs - since you can share lots of things.

Looking back, I'm not sure I could imagine bootstrapping without having a supportive partner. Mine was earning slightly less than me, but she covered the insurance, so that was helpful too. Much more importantly, when there were times that there was no time to see anyone non-business related, when I would get buried in a problem and lost my ability to see the larger picture, and when somebody needed to just drag me away from the computer for a bit, she was there. Not to mention all the help on the domestic/not living like an animal front.

It seems like some people need work, problem solving, and/or success like they need to breathe, so their pursuit of those things seems to be enough fuel for them to live off. For the rest of us trying to solve big or small problems, I would say that a strong relationship is an incredibly useful asset. That's even bracketing all the important benefits in all the non-work areas of life.

Completely agree - I would have probably gone crazy or given up by this point if it wasn't for my girlfriend.

I use a great extension for this purpose called mingaling. It replaces words you designate with translations when they occur on any page. So as you surf you passively learn vocabulary.

https://addons.mozilla.org/en-us/firefox/addon/ming-a-ling/

I tried to sign in with google email. The site asks to view my email addresses. I cancel.

> The site asks to view my email addresses.

what do you mean by that?

Great job. A few years ago I did something very similar myself but just for arabic - www.arabicreader.net (still running but needing some TLC) - but didn't pursue it the same way you've done. I take my hat off to you!

Really nice, i was just looking for such a tool as i'm learning esperanto

This is a great project. Please make the checkout page SSL-enabled only. At the moment, the extension's "Go Premium" menu link redirects to an insecure version of the page.

Yikes! Thanks very much, just fixed it.

Can I use this to improve my english vocabulary? Sometimes I read articles but I don't know what the words mean. It would be nice if you could implement this, if not already.

Only if your native language is not English. There's been a uservoice suggestion to allow it to work in monolingual mode for ages (https://readlang.uservoice.com/forums/192149-general/suggest...) - I've been avoiding this so far since it's a big change to how the inline translations and flashcards would have to work.

Loved the tool. Congratulations. Just wondering, how do you manage the phrase translation? It is your system learning from itself?

I pay to use the Google Translate API, it works very well for popular languages to/from English. Other language combinations aren't so great.

How much you pay for that?

It's $20 per 1M characters.

"Learn word in context", that's what I've been working on for a long time. Just from a different angle:)

keep going with this - it's genuinely useful - you've actually automated my language learning process.

"Read and change all your data on the websites you visit"

Srsly?

Yep, I know it's sounds scary, but I believe it's the only way for the Web Reader extension to work. I don't do anything malicious (just trying to upsell you on the premium plan) but I'd completely understand if you were freaked out by that.

An alternative is to use the bookmarklet, which will only run javascript on those pages where you explicitly clicked to open it, it works exactly like the extension:

http://readlang.com/webReader

I just tried it out. I take it all back.

It's absolutely awesome. I regularly read news in half a dozen languages and this is what I've been waiting for for years now.

Awesome!

i don't see japanese

Try the "More languages" dropdown when selecting a language. I think its there near the bottom, but it's actually spelled in Japanese.

It's also labelled as in beta. After signup, it clarifies what that means at the bottom of the page:

> Note: Readlang has only limited support for 日本語. Words are not prioritised based on frequency, and texts are not graded by difficulty.

Slightly disappointing, as I really hoped they did that, but I'll probably still give it a try. EDIT: I'm very impressed by the translation.

For Japanese, you can also check out rikaichan for Firefox. I guess rikaikun is the port for Chrome.

Reply on news.ycombinator.com