What is the perfect language course?

[Cavesa]

My addition to the perfect course would be...Make it entertaining! Maybe have an interesting storyline? How about make it a murder mystery, and you learn the relevant topics while learning about clues and character backstories. And make the conclusion in the target language, so you can't look at the back of the book and cheat.

Yes please. Or at least, make it anything but a family with an exchange student. I hate those and I actually am an exchange student, what about the rest of people.

And make it long with tons of exercises. Sure, grammar explanations, exhaustive vocab lists (of the kind "needed for the lesson" "more vocab for the topic" "more words for enthusiasts"), good audio. But what I miss the most are exercises. I always need to get several courses and workbooks to have enough of those! From the classical filling the gap ones (which only got a bad name due to overuse in classes) through substitution drills and rephasing, up to translation exercises.

My ideal would be lessons with bigger chunks of grammar but enough explanation, examples, and exercises. I really dislike to have one thing chopped over four lessons (with a few other lessons between them) and have tons of time to make some wrong assumptions. I like to see the bigger picture and then dissect it, instead of getting tiny bits of the puzzle with no idea what am I supposed to see in the end or at least how many pieces are in the particular puzzle.

[garyb]

I'm fairly spoilt since I've studied very mainstream languages with many resources, and for these you can get quite far with two or three well-chosen courses. The main thing that I feel most courses are missing is decent instruction on pronunciation: most either rely on the learner having a good ability to hear and reproduce sounds (e.g. Pimsleur) or give explanations comparing the sounds to English ones ("a" as in "father" etc.) that can do more harm than good. I like Assimil and Michel Thomas, so my ideal course would probably be a combination of these: MT-style introduction (with a native speaker) with additional pronunciation guidance and then lots of Assimil-style dialogues.

Other than that, the ultimate resource for me would be a kind of chat-bot for practising productive skills (written or spoken) and getting corrections without the need for a real person. Obviously it would never be a substitute for a real thing, and language learning has always been a social thing for me, but it would be a great supplement to tutoring and conversations which can take a lot of time and effort to find and, in the case of tutors, cost a lot of money in the long run. I realise that there are apps like Hellotalk and whatever has replaced the old Verbling, geared towards short and instant conversations with real people, but real people are flaky .

[Cainntear]

Also developers focus too much on trying to get all languages available for their beta when they should stick with just one (especially if you are trying to get the users to C1). Build and learn from that language then try to add more. Like with Lingvist.

I agree completely. But the underlying problem is actually worse -- by the time they reach beta, most language apps seem to be wedded to a particular way of doing things, and their codebase isn't flexible enough for them to implement any substantial changes based on tester feedback... or indeed any findings from research.

For example, Duolingo added the "timed practice" option very shortly after a paper was published that claimed to prove that time pressure in computer-based language learning resulted in better learning (the methodology was pretty weak, so I don't put much stock in their results). Now, I can't say for sure that Duolingo did this in reaction to the paper, but the timing was definitely interesting.

Unfortunately the timed practice element in DuoLingo puts the time limit against the whole session, so if you stop to think about (or even just look at) the corrections the software gives you, you lose time and you lose the game.

The system itself seems to be poorly designed, offering little scope for changing or experimenting with the interface. Good language software has to recognise that it's not perfect, so it has to be designed to be easy to change.

[mastrinka]

Hm, perfect language course would:
1.) teach me most useful vocabulary subset (I don't really need to know the name of exotic fruits or animals that I don't know the name even in my native language)
2.) should be repetitive, designed with forgetting curve on mind, i.e. should be form of a space-repetition system
3.) include me as an active participant (like asking me to combine known vocabulary and rules into new clauses, questions etc.)
4.) should use any generalization available (i.e. grammar) to help me understand and remember the structure of a language
5.) should introduce new words in context and with articles
6.) should include some kind of pronunciation practice or break down of hard words (Pimsleur does something similar for Swedish, breaking the words into syllables)

[reineke]

"The ultimate goal of second language acquisition research is the development of a theory of second language acquisition. I think there is fairly widespread agreement that no such theory exists; beyond that rather minimal point, the consensus starts to dissolve."

Kevin Gregg (1989)

Not a solid foundation to build upon.

[BOLIO]

A series of graded readers, say ten volumes, that has a native speaker reading the audio. It would take you from toddler level to native materials. It would start as interlinear dual text. Then move on to L2 only but newly introduced vocabulary highlighted and explained in the margins.

Also, grammatical explanations would be at the bottom of the page. It would have a complete glossary at the back of each book of the series. It would cover 5000 of the more common words of the languages.

[dgc1970]

A series of graded readers, say ten volumes, that has a native speaker reading the audio. It would take you from toddler level to native materials. It would start as interlinear dual text. Then move on to L2 only but newly introduced vocabulary highlighted and explained in the margins.

Also, grammatical explanations would be at the bottom of the page. It would have a complete glossary at the back of each book of the series. It would cover 5000 of the more common words of the languages.

It amazes me that Spanish for Reading, French for Reading, and German for Reading don't come with audio.

[Whodathunkitz]

A series of graded readers, say ten volumes, that has a native speaker reading the audio. It would take you from toddler level to native materials.

It amazes me that Spanish for Reading, French for Reading, and German for Reading don't come with audio.

With kids books, you can get sounds operated by buttons. Birthday cards can have songs. Apps can do anything.

I think this idea could be a winner. First thousand words with native audio example sentences.

My problems are all to do with production, I really need a chatbot / quiz for written and no idea (apart from wetware/people) for speaking.

Oh and time. Good quality not tired time.

[Käfer]

A series of graded readers, say ten volumes, that has a native speaker reading the audio. It would take you from toddler level to native materials. It would start as interlinear dual text. Then move on to L2 only but newly introduced vocabulary highlighted and explained in the margins.

Also, grammatical explanations would be at the bottom of the page. It would have a complete glossary at the back of each book of the series. It would cover 5000 of the more common words of the languages.

Wow, that sounds amazing. I would just add some additional focus on pronunciation: descriptions of all the sounds with IPA and physical descriptions of mouth positions, notes on allophones, stress patterns, clusters that change the pronunciation of their components.

And maybe I'd also add a few choices of speakers for the audio, or at least one male and one female speaker.

[Cavesa]

I think this idea could be a winner. First thousand words with native audio example sentences.
My problems are all to do with production, I really need a chatbot / quiz for written and no idea (apart from wetware/people) for speaking.

Most courses have sound for the dialogues and texts. Sometimes not for all of them, or some audios are without the text. Sure, having both for everything is a good one. Easy dissection of the course material for sentences to put in your mp3 or SRS would be a huge plus.

But the 1000 words are a problem. That is waaaay too little even for a beginner course. Good quality beginner courses (of various kind) that I have in my bookcase introduce at lest 1500 words just for the A1 level. Courses A1-B1 something like 3000 words. And no, they are not counting various forms of the same word as one. I would never invest in a course that teaches so little. Perhaps for an exotic language, if there was nothing else of comparable quality, but I know it would leave me wondering "nice start, but what to do now?".

And the first 1000 are usually not that hard to remember, compared to the second 1000 or the fifth. While there are individual preferences in this, I'd say the importance of good learning material is quite the same for all the common vocabulary. It is a very common problem for learners, who would otherwise be at solid intermediate, to have very poor vocabulary, as they didn't get so much support for the words after the first 1000 or two.

I like the idea of integrated work with the vocabulary. The fill in exercises or substitutions are good but I don't think they suffice. More translation exercises help with various things. But I would love to see a coursebook with official decks for Anki, Memrise (as I find the typing very efficient), Closemaster, or perhaps a "mutated" version of Lingvist (as I think Lingvist is a quite good platform with sometimes dubious content). Such coursebook lists are available to Chinese and Japanese learners using Scritter, I miss them for my languages elsewhere.

A series of graded readers, say ten volumes, that has a native speaker reading the audio. It would take you from toddler level to native materials. It would start as interlinear dual text. Then move on to L2 only but newly introduced vocabulary highlighted and explained in the margins.

Also, grammatical explanations would be at the bottom of the page. It would have a complete glossary at the back of each book of the series. It would cover 5000 of the more common words of the languages.

Yes please. Such a cover of the first 5000 words or more would be awesome! Such an approach Bolio suggests would also mean higher retention. From my experience, the usual courses (those with one volume per level meant for classes) offer lots of vocabulary in the four or ideally five levels. The problem is the relative lack of examples and exercises. Therefore the rentention rate from each level is less than optimal for most students. When I was young and despite being a rather independent learner for my age and the availability of resources at that time, I simply had no clue about stuff like SRS and I always failed at doing the paper cards for all of that.

I don't think it would necessarily have to be the toddler level. We are not babies and we do not learn like them. Assimil doesn't give toddler talk in the first lesson. If it did, I would never use it. But the graded attitude is surely good.

Real stuff is good. I can imagine various kids for children (not toddlers) or various media being introduced rather early.

Thinking of it, I would like the course to go from grammatically simple things to the harder ones. Not to order them by the touristy importance.

A typical example: My old French textbook from childhood and other older books that I've seen started the first unit with a simple construction of a pronoun and a verb in present with no complications. Either with être and avoir, or sometimes with regular verbs, leaving those irregular to the next lesson or two. Introduction of pronouns, the present tense, pronunciation, and a few words with very simple and regular ortograph. Nowadays, learners with most coursebooks start with "Je m'appelle" and "Comment tu t'appelles?" They get discouraged as soon as they start learning, because "French is so hard, there is too much memorisation and complications". Because it is simply impossible for the teacher to introduce them to the pronunciation, the present tense, the ortograph, verbes réfléchis, and question making, right upon encountering these examples.

There is more stuff like that. I would like a course that goes gradually by the grammar and shows me how to use it in various situations. Not to get to a narrow situation and miss the bigger picture. I like to start with what is easy, not with what is expected a tourist wants to know.