Was originally planning to write this a couple of weeks back, but a side-project and Horizon: Zero Dawn took the extra time. Both are now concluded with great success, and I’m back to regular programming (so to speak).
One of my longer term projects has been to learn Portuguese. I tried doing this the hard way some years back by enrolling to an evening class. My capacity of memorizing “arbitrary” non-logical things coupled with general tiredness after work days caused this experiment to fail. Years came and went and somehow the project bounced back into my TODO lists, so I thought that this time I’ll use software to drive this thing forward. At my own pace, dang it.
So, last summer, while on a vacation of a sorts, I tried both Duolingo and Memrise on Android and ended up with Duolingo. Now, I know that you’re saying: “Man, that is crazy, Duolingo’s Portuguese is actually Brazilian!“, but I thought it was close enough (and I liked Duolingo better). My regime was perhaps 30-40 minutes of Duolingo each day for a couple of weeks and then I switched to Memrise since it had proper Portuguese, but kept the same learning regime.
After coming back to work after the vacation, this regime fell apart quickly. I didn’t have the time to do it in the mornings and certainly didn’t have the time to do it during my work day and after coming home, could think of better things to do (I was working quite long hours back then).
I decided then that I’ll try Memrise again later, but perhaps with a more “permissive” regime.
After the winter holidays, I noticed that the car battery was getting tired and long story cut short, this forced me to move from a relatively random sleep patterns to a highly periodic one.
Having a highly regular sleep pattern now also means that I have a more or less set time at which I go to sleep, and actually get enough rest, which is quite awesome. The other positive side is that I can schedule some time to train my Portuguese in the morning. Each morning, without exceptions (so far).
Amateur or Pro?
Memrise is a split “no-cost” / “subscription based” service. I recommend you delay your start of subscription until you’re truly happy in the method of learning with the service.
If you “go Pro”, you’ll get access to “Listening Skills” (hearing comprehension) whenever you want it (otherwise it’s quite rare for Memrise to select those types). You’ll also get unrestricted access to “Meet the Natives”, which is a series of short video bits of regular people saying the sentences that you’re training. (The videos are a relatively new addition, so do check whether they’re available for the language that you’re learning before going Pro).
Going Pro also allows accessing some extra “BIG DADA” via the Memrise web site (which I normally do not use). The metrics that are collected are.. well.. quite basic. Personally I don’t think it’s worth the Pro and was somewhat disappointed that they didn’t have more metrics based stuff in the web.
I do have Pro, but mainly for the “Listening Skills” and the native videos. Also, if you score highly (yes, Memrise is somewhat gamified), you can get an automatic 50% discount Pro offer for the yearly subscription, which is actually a good deal. The offer was presented to me automatically once I got highest score of the week or the month (can’t remember which now). Not sure whether this is a permanent “award” mechanism, but at least it worked like this for me. Achieving high scores is relatively easy during a vacation.
There is a slight disconnect between the way that the web service thinks of things and how the Android client thinks of them. For some reason, my unbroken streak of 78 days is not actually unbroken according to the web service above (evidenced by the 4 holes in the heatmap since January). The image also shows quite nicely the more intense regime I had during my summer vacation and a bit after it.
So, not really rocket science in metrics, but I guess something that “performance” oriented people like.
Perhaps slightly more useful metric is the “learning time”:
As you can see, I don’t actually spent a lot of time each morning on Portuguese. However, doing it every day in a repeated fashion does work (at least for me).
The last graph is supposed to inform me of the best wall-clock time to learn new things (when my “performance is at peak”).
However, in its current form it doesn’t really work that well. I’m not sure what information I can get from the graph other than that I never use Memrise between 1AM and 5:30AM, which is not even true. But, international timezones, it is an impossible problem to solve in so many ways. I do occasionally use Memrise in the evenings during weekends, but the main bulk of learning happens each morning.
That’s more or less it. Getting Pro also allows one to mark words as “difficult” for further training later, but at least I didn’t find that very useful yet.
I’ve completed the first series of lessons (Portuguese 1) and am at about one third into the second course. My experience with both has been similar, although the content in the second series is somewhat more complex (as to be expected).
I’m using the default setting of “5 new words” per day, which sets the minimum score that Memrise considers as “daily goal achieved” (it doesn’t actually track words learned, so any way of getting the score is accepted to keep the streak).
Each morning, I start with “classic review” for the current course that I’m learning. The number of words to review depends on how many new ones you’ve learned in recent history. Older words are also “recycled” once a while, but I’ve yet to determine the exact rules. It seems to work quite well. For scoring, this is very good until you run out of words to review. Classic review is the only mechanism where you have to remember how to spell individual words. Even then, you have a subset of full keyboard to help you remember which letters were included in the word/phrase. So yes, cheating a bit.
If I feel that I can introduce new words, I do one or two rounds of “learn new words”. This is the main mechanism to expand the vocabulary and learn things in Memrise. I quite like the way this part is constructed.
The “speed review” type of “quiz” acts as a very good meter to test whether I got enough rest during the night. The maximum length of the speed review is 100 words/phrases (once you learn that many in a course) and going through that without 3 mistakes with the time pressure works very well. Speed review is the best way to increase your score, but you need to have at least 100 “learned” in a course for this to work well. This is also the most stressful way of scoring, but some people might enjoy it (it’s a game).
I then switch to the previous course and do “classic review” there, just to keep the words alive in memory. I used to also run the “speed review” for the first course, but that ended up taking several minutes of extra time each morning, so had to drop it once I got to 100 words in course 2.
During weekends I might do some Speed Reviews just for the fun of it (depends on whether I have Important Things™ to do or not).
Learning new words is sometimes a challenge for me, when multiple words are close to each other (same context and similar semantics or opposing semantics). I suck at memorizing things for which I cannot form any logical connection or some kind of “collation order”. I’m pretty good at extrapolating and intrapolating, but I’ve found that natural languages are way too chaotic for me to infer too much.
Once I notice that I’m having issues with the new words, I stop doing the “Learn New Words” and just run the speed review until I’m comfortable with the set again. I’ve found this to be more useful than using the “difficult words” mechanism. I’m not sure what will happen if one is unable to rest multiple days in a row however. I think the maximum number of days that I was stuck on “speed review” was 4 or 5. This “block” has now happened twice, and so far this technique has helped me go over it. YMMV.
Comparison to Duolingo
Duolingo’s objective (as a project) is quite different from Memrise’s and I really wanted to use their software. In many ways I found it much more “fun” due to the following reasons:
Variety in voices (Memrise only has one femaly voice, or at least in Portuguese).
Pictograms form the core of pedagogy in the beginning (while words form the core in Memrise, meaning that you have the extra baggage of loosely defined origin language like English). I realize pictograms mainly work for nouns and once one gets into more abstract concepts, pictograms lack expression (draw me “loathing”, or “hurry”).
Other differences between the two:
Selection of languages is more limited in Duolingo (due to the core objective difference).
No nagging to upgrade to the “Pro” version of Duolingo (this wasn’t a biggie for me in Memrise, but might be something you won’t like).
What might also be problematic is that when I last checked (this was some months back already), I couldn’t really understand what the financial model of Duolingo was. Once I commit to a tool, I like to know how the developers plan to finance the maintenance and perhaps even new development of the software. If the tool is completely without cost (and even advertisements), then why would anyone spend their spare time maintaining the software into the future? Obviously selecting subscription based software does not guarantee that the developer has any interest in maintaining it, but at least there will be some monetary interest to do so.
Yay or Nay?
Definite Yay, at least based on the past 78 days. There are many things I would perhaps tweak, and a bug or two, but overall I’ve been quite happy with the process so far.
And what about Portuguese? I think it’s awesome too. Yay.
If you wish to “friend” me on Memrise my username there is
czrczr. Then we
can compare our scores! (Much wow.)