I mentioned briefly that I've got a little project on the go to develop some language learning software. I've been writing it under the codename Spanish Inquisition until I recently found a near-perfect domain name to use (but I'm keeping that to myself until I have a brand identity to launch). Why "Spanish Inquisition"? Well, the first iteration was a very simple Javascript page for asking and checking the answers to questions for a Spanish class I was teaching, and very quickly I hit the practical limits of Javascript as a development platform, so I switched to using Python. A Python program that asks lots of questions and teaches Spanish... well, if you don't get it now, it's maybe just too geeky for you!! Anyway, so once I switched to Python, I started making it more flexible -- it generated its own questions and I used it to teach myself a bit of Corsican. It was too labour-intensive to add new languages and new language features, so over the space of the last year and a bit, I've been rewriting it on-and-off to try to get something that I can quickly adapt to any language. There are still big gaps in the software that restrict what types of language features I can get it to generate, and I think I have a solution to this, but as that would mean a major rewrite, and I'm itching to get something together and on the internet for initial testing. So I'm currently neatening up the codebase to iron out a few bugs, and then I'll be looking at the best way to get it onto the internet, as well as choosing some initial languages to input rules for. And then...? Beta testing -- and polydog users will be invited (if the choice of language suits you). I don't have a timescale at the moment, but I doubt it will be before the end of August -- I'm working here in Sicily for another month, and then I'm probably having a holiday, so August will be my month for learning web frameworks! In fact, while I'm learning about web programming, there's another little project I'd like to set up, to try to discover a little bit more about how best to use SRS, and I'll be looking for volunteers for that too. The results will inform the future development of my software, but they'll be made available to the public as well.
Not quite -- I've just reworked standard techniques from the field of natural language processing/computational linguistics. I'm sure someone else would have got there before me if all books on NLP/CL didn't poison people's minds with Noam Chomsky's half-century-old work...
Cainntear, You might want to check out the zhtoolkit site. This guy is Chinese learner and programmer. His latest blog post (first in a year), Busting Personal SRS Myths, and an Anki Simulation in Python not only talks about his use of Python, but makes available some of his scripts.
Interesting stuff. Not sure I could use any of it, mind, but Worth reading-- thanks. Anyhow... after being blocked by stupid bugs and too much work at the day-job, I had just managed to get things moving again, but an intermittent power fault with my laptop has just started getting worse, and my PC's out of action. I'm going to see if I can survive on just the iPad (I haven't installed Python on it yet, but it's in the appstore), as I don't want to get a new laptop right now. I'm leaving the Island in three weeks. Minus one week for delivery (I don't want an Italian computer) and really the extra weight for the plane makes no sense for having it for two weeks.
Hard to say since although he has revealed some details, he doesn't want to reveal them all. And they both seem to share a focus on AI methods. But perhaps not.
I hope you can find a place for the soft cushions and the comfy chair (from Monty P's version of the Inquisition)
I turned down the opportunity to be a tester for Sprachprofi's course due to conflict of interest. There are certain similarities, although Sprachprofi's software is designed specifically to make a course in Chinese. I'm trying to develop my software as a flexible and extensible platform that can be used to create and host my own courses, or to be licensed out to universities and publishers as an interactive extension to their courses. I'm also hoping it will be used in academia to investigate the effects of order-of-teaching on students' retention of language, but that part is a long way off. Chinese isn't high on my list of languages to teach at the moment, so we're not in competition that way. Furthermore, I'm hoping to sell mine, rather than go down the unsustainable free route. Notice how most of the big players in "free" education are starting to divert: Udacity is now selling "micro-degrees" (software training courses by any other name) and signing up for many of the courses on Coursera or EdX will see you getting spammed to kingdom come for (expensive) distance education courses from the host institution (with a nominal discount). The only group currently avoiding commercialisation is the Khan Academy, which survives due to its unusually high profile and the help of several high profile donors.
Unfortunately our almost fanatical devotion to the Pope has scuppered plans for Hebrew and Classical Arabic courses... ;-)
Biblical Hebrew is actually often required for advanced Catholic theological degrees along with Ancient Greek and of course Latin. However the current pope isn't the language scholar the previous two were. Re Sprachprofi's course, I have not investigated it, but personally I would not be willing to donate at the start unless I were convinced the course already included enough material to take one to a very advanced level. This would also be an issue I would imagine with Cainntear's concept which he wishes to commercialize, since there is far greater demand for beginning courses than advanced ones, as witnessed by the lack of advanced resources for many languages. I guess the question is does a course author try to get the beginners to pay a little more incrementally to finance the advanced levels somewhat or just charge advanced learners a lot more to make up for lack of volume. Of course if the course actually had a significantly smaller attrition rate after the beginning phase, then perhaps it would be moot.
My system doesn't really conform to the traditional tiered view of "beginner -> intermediate -> advanced". My first goal is to create a system that addresses learner accuracy, a skill that is necessarily overlooked in language courses due to restrictions on time. Most of us simply don't get enough opportunity to practice the structures we've learned, which is why we fail to internalise them. That failure to internalise is blamed on translation, but as I said: it's normally just lack of practise. Attempts to remedy this tend to fall into two camps: structured drills and free speaking practise. Structured drills don't work, because they overcompartmentalise and decontextualise the structures, and it usually becomes a word-juggling exercise that students have no problems doing, but that doesn't affect their ability to produce language off-the-cuff. Free speaking practise isn't appropriate to this problem, as the complexity is uncontrolled, and students grammar actually gets worse as they spend the whole time thinking about the subject they're speaking about. The student learns to cope with bad grammar, instead of learning to eliminate it. So I'm focusing on something that spans the whole range of levels, as would be necessary in order to break into the university teaching market. I don't want to be pitching myself against TY, Colloquial etc in the first instance, as that's not what I want to create. However, its status as a "supplement" by design will still be useful to the self-teacher, and in particular for those who have already learned a language or two and don't really need as much guidance.
I don't really get the distinction between structured drills which you frown upon, and what you propose. I mean sentence patterns/usage patterns do exist in languages, and one needs to be able to use them effortlessly. For example in German, the differences between words for "but" such as "aber/weil/wegen/etc". I find that sentence paradigms for each along with grammatical rules upon which they are based to be highly effective as one can compare and contrast them, assuming one is already familiar in general with grammatical constructs like conjunctions and subordinate clauses, and differences in relating causation. Now whether in addition to such paradigms one needs to practice each over and over with different small changes is open to question. As for spanning the whole range of levels, what is the practical difference between your complete course and a series of levels of other courses? Surely you are have a gradation from beginner to advanced both for grammar and vocab? Now perhaps if I understand your concept correctly, and the program uses other course media as inputs (to more effectively teach them), then the gradation is the result of the level of the input being used.
A structured drill tends to "practise" the same point multiple times in short succession, in similar settings. Why the inverted commas? Because real learning is about supporting recall. If you do 20 drills of the same thing, you are only recalling it once... at most. As most drills point you to the structure you're supposed to be practising, in practice, you are almost never made to recall the language point supposedly being practised. Instead, the only thing you have to recall is the vocabulary used. The line that the good teacher treads is giving just enough support to aid the recall of the language point in question without giving it to the students and defeating the purpose of the exercise. My goal is to work that notion of minimal required support into the algorithms.
