Languages and a Technology-Intensive Video Approach to Acquisition – Week 47

I’m back from the Andes and back to Mandarin, after a full 20 days off! I’ve created an Off Topic section to my blog. I hope you will enjoy my write-up on a Choquequirao – Machu Picchu backpacking expedition.

2014-12-01 10.16.33 (2)

Although I spent zero hours on Mandarin for three weeks in Peru (see the leveling off near the end of the blue graph),

hours_13-dez-14

I did have fun learning a few words in Quechua and got an extended immersion in Spanish. Working professionally with the language and having the opportunity to travel a great deal in the past few years has taken my fluency to a new level. Before I reveal my actual origins, I am almost always mistaken for a native Spanish speaker, although nobody can quite pinpoint my accent (because, after all, it’s a complete mix, from nowhere and everywhere at once). In Latin America, I am often taken for a Spaniard, and alternately for a Central American, Mexican, Chilean, Colombian, or Venezuelan.

My French, on the other hand, is in dire need of training. It’s so rusty that I’m hesitant to use it at all. On the rare occasions I have the opportunity to speak in French, things often go well for a few sentences. However, as conversation transitions to untested ground, terrifying gaps in my vocabulary emerge like yawning abysses and I recoil back to a more familiar language. I believe I will soon commence a small French project, perhaps devoting 10 minutes a day, mostly to French radio. I expect it will take a year or two to recover full fluency. Stay tuned as I will keep tabs on this blog.

. . .

My Mandarin experiment is a one-sided approach to language acquisition—specifically, to obtaining listening comprehension. My goal in using such an exclusive approach is to prove the utility of an almost universally applicable method—concentrated listening to movies and other authentic video sources—at any level. Ideally, this method should be combined with reading, writing, and most importantly speaking to natives—preferably including some natives who are willing to correct your mistakes.

A natural, communicative approach to language acquisition is far superior to traditional methods that are heavy on translation, rote memorization, and grammar rules.

However, technological advances and cutting-edge experimentation have introduced new possibilities in language acquisition that are tantalizing. I would like to briefly introduce one approach that, like mine, focuses on listening comprehension through the use of authentic video sources. Unlike my experiment, it requires a great deal of technological expertise and preparation, as well as repetition and (dynamic) memorization. Despite its more difficult application, I believe this method has the potential to be extremely efficient and actually speed up acquisition as compared to a purely natural approach, when used in combination with communicative strategies.

Fortunately for me and for anyone who is interested in this state-of-the-art method, a fellow language enthusiast, emk, who has been following my experiment, is currently using and meticulously documenting this method as part of his own experiment. In fact, it seems that one of the inspirations for his project was my experiment, and his project is a bit of a counterpoint to mine. He believes straight video viewing at a very low level of comprehension (my watching Mandarin is an extreme example of incomprehensibility) is inefficient. I am very impressed by his approach and also grateful for his extremely helpful and detailed explanations.

You can learn all about emk’s project and the technical details of his method by visiting this language-learning forum thread.

I will provide a highly summarized introduction to the method he is using and then encourage you to visit his thread to understand it in greater detail. Hopefully, he will also comment on this post and correct anything I have gotten wrong.

First, a few concepts. Spaced repetition is a memorization technique that presents the data you are trying to learn at increasing time intervals. Each time you get an answer right, it will take longer for you to see the same problem again. On the other hand, if you get an answer wrong, you will encounter the problem again very soon, since you obviously still need to review and properly memorize the solution.

Spaced repetition software (SRS) applies artificial intelligence to this concept. It is often used in language acquisition, such as in the Pimsleur system. It is akin to the use of vocabulary flashcards. As you get answers right, you see a given digital flashcard less and less frequently, but if you get them wrong, it will come back more frequently. The great thing about these digital flashcards is that they can incorporate not only text and written translations, but also audio and video. A commonly used SRS is called Anki, and Anki cards have become synonymous with SRS in the self-teaching language acquisition community.

Bilingual subtitles on movies or TV series can be used to generate Anki cards. An entire movie or episode can be parceled into hundreds of digital cards that contain short video clips with audio in the language you are studying, the corresponding L2 (foreign language) subtitles, and the L1 (your native language) translation of those subtitles.

By some accounts, with a relatively small investment of time (once you’ve created the cards), you can memorize these dialogue snippets in a completely unfamiliar language and then watch that movie or TV series and understand it quite well, as if you had already reached an advanced level of comprehension in that language.

The technology involved is dubbed subs2srs, which I assume is short for “subtitles to spaced repetition system”.

My preliminary assessment is that this approach may constitute an excellent tool for some second-semester students to bridge the gap between study of high frequency vocabulary and grammar fundamentals (learned through games and structured communication activities) and the use of authentic videos, reading, and participation in guided conversation classes.

In this regard, I hope to experiment next year with the use of Anki cards and subs2srs in my language institute. We have previously identified a weakness in our teaching approach for some second semester students, and I believe subs2srs, in particular, may be very useful in transitioning students to authentic listening sources and conversation, thus helping to improve outcomes.

With regards to our respective projects, it is very hard to compare results for several reasons. Although emk has apparently made greater progress in approximately 20 hours than I have in almost 200, that comparison is misleading. First, being a native English and fluent French speaker tackling Spanish, I would estimate his progress in general is expected to be at least five or six times faster than someone like me, learning Mandarin without knowledge of any related languages. In the beginning, this speed discrepancy is likely to be much more pronounced, given the huge number of cognates—probably upwards of 70% of words, as compared to practically zero cognates in my case.

Second, his self-described “cheating” and “narrow-listening” approach is specifically designed to give an initial boost to comprehension, whereas my listening approach is aimed at gradually adapting my ear and brain to a new language and is based on broader and more massive input. An open question is whether this broader input approach holds any long-term advantage as compared to narrow listening.

Third—and I will be glad to receive emk’s clarifications—I don’t think his 20 hours can necessarily be fairly compared to my 200. I am carefully clocking my time and any contact I have with the language is being computed in the tally. As I understand it, his approximately 20 hours refers mostly to using his Anki cards and is being computed by the software itself. When I dabbled with Anki cards, the software I was using clocked about 12 minutes for the entire hour I spent on them. In other words, it was vastly underestimating time as compared to my stopwatch. Further, I am not sure he is carefully clocking and including other types of contact he has with the Spanish language.

In addition, it should be noted that long hours over several days are spent in running subs2srs and preparing the Anki cards, whereas my time spent in purchasing or finding videos is minimal. If you include the preparation phase, the time difference would be significantly reduced.

Having said all that, I still believe emk’s approach may hold a significant efficiency advantage as compared by my pure listening approach (which is not entirely pure, since I have been using subtitles much of the time), especially when tackling a language as difficult as Mandarin. It seems to me to be a wonderful use of technology applied to language acquisition, and though I won’t use it for my Mandarin experiment—as it would violate my rules—I do hope to use it in acquiring some future language. I would probably limit its use to beginner and low intermediate phases of language acquisition, and even then, to at most 50% of my time spent, in order to avoid excessive dependence on translations, which can be pernicious.

In that vein, as emk himself has suggested to me as a possibility, I think finding a way to use Anki cards and subs2srs without translations would be ideal. Of course, for a beginner, translations hold the key to meaning, but visual cues or other creative strategies can also be used in their stead.

9 thoughts on “Languages and a Technology-Intensive Video Approach to Acquisition – Week 47

  1. emk1024 says:

    Wow, thank you for the write-up! And yes, moving from English+French to Spanish simplifies my task tremendously, as you well know.

    You’ve raised some really interesting points, and I hope to respond to them with the detail they deserve. 🙂 But in the meantime, I’ve briefly responded to a couple points in my log. One thing worth mentioning: I’m not actually trying to memorize the dialog on my Anki cards. Rather, my goal is to be able to understand the audio when I review the card. This is much easier and less stressful. 🙂

  2. emk1024 says:

    A couple more thoughts, for those who are interested:

    Translation. In practice, I have very little temptation to translate when using subs2srs, for a couple of reasons. First of all, I’m testing my ability to understand the Spanish audio in context, not my ability to give the English translation. Secondly, the English subtitles are never quite accurate enough to trust completely, so I’m forced to interact directly with the Spanish and guess, just like I would if I were learning from context.

    Narrow listening. This is an old proposal from Krashen, a second-language acquisition researcher who was massively influential in the 80s. Basically, the idea is to artificially limit your subject matter in the beginning, so that you can rapidly learn most of the subject-specific vocabulary. In essence, this allows you to “simulate” an intermediate level very early on, and to start learning directly from input much earlier than you could without the specialization. Then, once you’ve established a strong core of grammar/comprehension/fluidity inside your “niche”, you can branch out comfortably.

    This is why I actually dislike films for intermediate language learning: They’re too short, and each one requires you to “start over” much too soon. I much prefer DVD box sets of TV series, because you can go from ~B1 to 70+% comprehension of a series within a couple seasons (especially if you already have good reading skills or know a related language). Then you can just buy a bunch more TV series, and repeat the process. After 5 or 6 DVD box sets, I could channel-surf French TV and watch a large majority of the shows with pleasure.

    This is actually one of the reasons that I think the “Boonie Bears” are an excellent choice for your Mandarin experiment. By tightly controlling the subject matter and the number of voices, you make it much easier to reach a “tipping point” where you can start learning rapidly from context.

  3. emk, thanks so much for your elucidating replies.

    Your clarifications about not memorizing content nor relying much on subtitle translations are important and make me even more a fan of your system. One question, though: how do you determine whether you have accurately understood the audio on a card (and therefore click on the “easy” button)?

    Regarding narrow listening. Krashen’s theory makes a lot of sense and I have no doubt acquiring a language this way offers, at the very least, a psychological benefit. Students can easily become frustrated, to the point of giving up entirely, when watching countless hours of movies or other content while understanding very little of what’s going on. Quickly getting to a higher level of comprehension, by using subs2srs and/or exclusively a specific TV series, as you suggest, can thus obviously help motivation.

    However, here is perhaps the crucial question. If one can take the long view and has the motivation and stamina to stick it out (as I believe I do in this experiment), is narrow listening more effective anyway, or can “broad listening” provide the same outcomes in the long run? Beyond that, does “broad listening” hold some actual advantages?

    I believe your answer, and undoubtedly that of many people at the how-to-learn-any-language forum, would be that narrow listening is much more effective because, as you say, it allows you to reach a “tipping point” where content is actually intelligible much more quickly. This view is predicated on the supposition that listening to unintelligible content is inefficient or entirely useless. Conversely, the more you understand what you are listening to, the more quickly you can learn new vocabulary, structures, and so forth.

    I disagree with that view. The key to learning is attention. Intelligibility provides interest, which in turn spurs attention. However, if you can manage to listen to an apparently unintelligible dialogue with real, focused attention, I believe you can learn, gradually but with efficiency. Becoming familiar with the language’s sounds and intonation is very important, and when you occasionally decipher a word, and then another and another, those memories become deeply ingrained. It takes a lot longer to see the results, but the cumulative, deep effect is eventually impressive. Well, at least this is what I hope will happen with my Mandarin experiment hahaha.

    I would also argue that “broad listening” holds two important advantages over narrow listening. One is obvious. By being exposed to many more voices, styles of speech, subject matters, and so forth, your listening ability is more versatile and well rounded. I understand that in a narrow listening approach, after spending a long period exposed to one TV series, you then pivot to another, and then another, gradually broadening your familiarity with different vocabulary and accents and watching media more generically. However, arithmetically, if you dedicate 200 hours each to 5 specific TV series, you will have exposure to much less linguistic variety than if you use those 1,000 hours to watch 20 different TV series and 100 movies.

    The other advantage to “broad listening” is less obvious, but I believe more important. If you listen to a huge variety of sources, you are acting like a word corpus collector. You will learn those terms that repeat the most often across a wide spectrum of listening sources. In other words, you will tend to learn the highest frequency vocabulary in the language. By contrast, if you watch a TV crime series, and especially if you use a high repetition recourse like Anki cards, you will certainly spend a lot of time learning terms that are much lower frequency, but specific to the context that is portrayed in that medium.

    In sum, I believe both approaches hold advantages. As I’ve stated repeatedly, in my opinion, the ideal language acquisition approach should be balanced. I believe there are important synergies involved in combining listening, speaking, reading, and writing (and even a tiny little bit of formal grammar study). The overall benefit is greater than the sum of the parts. Similarly, I believe there are synergies to be had in “mixing it up” between narrow listening and my so-called “broad listening” approaches.

  4. emk1024 says:

    One question, though: how do you determine whether you have accurately understood the audio on a card (and therefore click on the “easy” button)?

    Well, it’s usually not too hard. Once I’ve reviewed a card a few times, I usually just understand it. Sure, a card might contain an idiomatic expression like, “[b]eso si que fué[/b] asombroso”, which I can’t break down into parts or explain, but that’s OK—if I can accurately hear the words used, and I understand the underlying emotion/idea that’s being communicated, I just go ahead and pass it anyway.

    There’s really not that huge a difference between your experiment and mine in this regard. I’m quite content to tolerate a lot of guesswork and ambiguity.

    I believe your answer, and undoubtedly that of many people at the how-to-learn-any-language forum, would be that narrow listening is much more effective because, as you say, it allows you to reach a “tipping point” where content is actually intelligible much more quickly.

    My reasoning is perhaps somewhat simplistic: I want to learn a language by sitting on the couch and watching a fun TV series. I want to get sucked into the plot and the dialog and the story, and not spend my time straining to pick out a few words here and there.

    So tools like subs2srs and narrow listening are just short-term tools to reach that point as quickly as possible. And so far, it seems to be working: There are now 3 episodes of Avatar that I’ve studied, and which I can watch with pretty good comprehension. As for episodes which I haven’t studied, that aren’t too far out of reach—if I pick an easy episode and watch it with English subtitles enabled, I can sometimes understand over half the Spanish.

    Even given my strong background in French, this is a really nice result for less than 2 months of study, at generally less than 30 minutes per day.

  5. emk1024 says:

    If you’re interested, here’s an update after studying 20 “official” Anki hours (which might really be 30, plus outside study). The good news:

    1. I just read a short children’s graphic novel in Spanish.
    2. I watched several episodes of Avatar without subtitles, and I could follow the plot.

    In the case of Avatar, I watched two episodes that I last watched a year ago in French, without any subtitles. I was able to follow most of the plot without any problems. I could understand a good chunk of the easy dialog, and pick key words and expressions out the harder parts of the dialog.

    This is a pretty major milestone: It means that I can actually watch the series for fun, and I’m getting near a point where extensive watching should allow fairly rapid gains.

    Even with my substantial discount from English and French, this feels like very rapid progress for the modest effort that I’ve invested.

  6. Your progress seems quite impressive. How long do you expect to continue with the subs2srs? I assume from your answer that you intend to continue with your studies of Spanish beyond your experiment. Will you begin doing extensive watching of Avatar soon?

    Your “eso sí que fue asombroso” reminds me of a joke my high school Spanish teacher told us.

    A Mexican guy who spoke no English went into a store in the States to buy a pair of socks. The salesperson, noting his predicament, pointed to some shoes, but the Mexican guy said, “no, no, no.” He then pointed to a belt, but the answer continued, “no, no, no.” This went on successively for some time. Finally, the salesperson pointed to a pair of socks. “Eso sí que es!” exclaimed the Mexican, happily. The salesperson said, “Well, if you knew how to spell it, why didn’t you say so from the beginning?”

    Get it?

  7. emk1024 says:

    Hah, great joke. 🙂

    I’ve already tried an extensive listening experiment without subtitles. The verdict: Out of 6 episodes, I felt that maybe the easiest 3 or 4 were at the very low end of my personal “useful extensive listening zone”. For me, this means that I could follow most of the plot, that I had a good time, and I could understand a good chunk of the easy dialog.

    I think I want to do some more subs2srs before I watch Avatar and Korra straight through. Maybe I’ll aim for 1,500 subs2srs cards, like Sprachprofi? Anyway, I’m getting pretty close, which is awesome.

  8. emk1024 says:

    After studying 3 episodes of Avatar with subs2srs, I’ve decided to mix things up a bit, and start working with larger chunks of dialog. This feels like a natural evolution of my Spanish: I’m getting better at understanding short snippets, so it’s time to start working with longer sections.

    The rules for reviewing cards are still the same:

    1. Look at the pictures and listen to the audio.
    2. Try to understand the audio directly in Spanish without translation.
    3. Show the back of the card.
    4. If I want to verify what I heard, glance at the Spanish or English text as needed.

    But now I need to deal with ~10 seconds of conversation at a time, so I need to hold more Spanish in my head at once, and I need to understand as much of it as possible in real time. It will be fun to see whether this helps. 🙂

  9. Keep us updated! It sounds like increasing the duration of the snippets is a natural way to progress.

    Sometimes I like to think of language acquisition as putting together a huge puzzle (http://mandarinexperiment.com/2014/08/03/experiment-assessment-at-the-10-mark-i-will-learn-mandarin-week-28/). In this analogy (and obviously we’re both focusing just on listening comprehension), you could say that your subs2srs method focuses intensively on specific sections of the puzzle, whereas my extensive video watching is more scattered. I take pieces randomly out of the box and try to put them in their general location, hoping the panorama will gradually come into view.

    The question of efficiency aside, I think these respective methods have their strengths and drawbacks in terms of enjoyment. You clearly seem to enjoy your method, and I can definitely see how gaining comprehension more quickly would make viewing more fun. On the other hand, preparing the Anki cards and doing the continuous exercise of repeating snippets might become a real chore to me. On the other hand, while I often greatly enjoy watching new Chinese movies and Boonie Bear episodes, the incomprehensibility can get tiring and I am sure such a radical approach is not for everyone!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s