neofight78 wrote:Cainntear wrote:But that word -- subscriber -- is what I really dislike about LingQ. $10 a month -- for what? There's practically no ongoing development costs for the site
That's simply not true, the site plus iOS and Android apps do get updated, and developers don't come cheap. I'm quite happy to pay $10 a month not to mess around trying to install LWT and for the convenience of having a proper iOS app. For other people it may not be worth $10 a month, but the price is not unreasonable.
OK, I'm seriously overstating. But all they seem to be doing is tinkering with the appearance. The underlying functionality appears not to have changed in at least the last five years. The assumption is that it's done, finished, complete; and that the only ongoing development goal is to make it visually more appealing to consumers.
What's there seems incredibly basic. Crowdsourced glossaries that are short on information (eg "mora" in Portuguese as "live", but no information on which conjugation it is); flashcards based on these glossaries; "cloze" tests (a misnomer) that just pluck out bits of a text and drop out a word. None of that's particularly difficult or particularly clever.
Even the basic idea of "i+1 comprehensible input" is very poorly implemented. They have a vast database of texts, and it would be pretty straightforward to suggest what to read next, but the user is left to choose, and the only real data available is "number of new words". Choice seems like a great thing, but you can't make an informed choice until you know what it is, and you don't know until you've read it. The number of new words is done per lesson and per course, but there's no clear and unambiguous indication of how much repetition and reinforcement the reading is going to give -- this reinforcement is a key part of comprehensible input theory, and Kaufmann has just hacked around the problem and put the words into SRS instead, which is essentially the opposite of CI.
Given how little direction is given to the user (n.b. "choice" is often synonymous with "enough rope to hang yourself") and how little scope there is for meaningful control of variables, there's really no data Kaufmann can draw on to prove the effectiveness of the technique or to help him refine the algorithms.
With no path for continuous improvement of the service, it all falls back on whether Kaufmann was 100% right first time, and no-one in history ever was. Kaufmann certainly isn't the exception.