How not to learn Spanish: Building too much stuff, not studying enough

[emk]

Very cool stuff, emk! It's really interesting to think about what might be possible now that we have access to all of these fantastic underlying tools. By the way, have you checked out knowclip?

Oooh, that is very shiny and I had not seen it. Thank you very much!

In general, I put down substudy back in 2017 or so, and I've paid almost no attention to what's happening with language learning tools since. I've been trying to fix that recently by reviewing a bunch of popular tools, but there are still neat things out there that I've never seen.

Knowclip is a free, open source project supported by Patreon, and it has very similar goals to substudy—though the author focuses more on selecting "easy" cards at an appropriate level. Which, honestly, a pretty easy feature idea to overlook if:

1. You're an English and French speaker learning Spanish, and
2. You have a high tolerance for partially comprehensible input.

For Japanese, proper card ordering might be essential at the beginner level. So I need to download Knowclip, and do some a proper review at some point.

It ended up being more fun for me to step through the video in Language Reactor versus working through the cards. But a big part of that might be because I've never been able to get into the habit of studing with SRS cards

The exact kind of exercise that I'm doing (turning an entire episode's dialog into flash cards) is most useful at the beginner level. At this point, basically everything is worth studying to some extent or another. At higher levels, I think it makes sense to do a lot more extensive watching, and only make cards when I stumble over something interesting. Sheer volume of largely-comprehensible input is magic.

I do have plans to add a basic video-watching UI to substudy. That's actually the next thing on my list. Though I reserve the right to change the list or ignore it for another 5 years!

Another thing I came across back when I was working on this is that there are various forks of whisper. In particular, I was using stable-ts in order to get more accurate timestamps. I don't know if whisper itself has since made improvements in this area.

Very interesting! I generally can't run serious models on a 4GB nVidia graphics card, and I assume that most users of substudy can't either. If I were willing to spend more money on hosting fees (besides what I've been spending all these years to keep the forum up), I suppose I could host a modified Whisper model as a private REST API. But then I'd need to charge for access, or start up a Patreon, or something. Still, I'm very glad to know that this is out there.

An interesting card, and a conversation with ChatGPT. I'm having breakfast late this morning, and this fascinating card just popped up for the first time:



Quizás lo que sea que tenga que hacer, ¿se me ocurrirá?
"Maybe whatever I have to do will come to me?"

Hmm, those are not familiar verb forms. And we're speaking about possibilities, not facts. It's time to do a little research:

Hello! I'm learning Spanish. Please respond to me as a skilled and professional Spanish tutor who works with experienced language learners.

I encountered the following phrase:

> Quizás lo que sea que tenga que hacer, ¿se me ocurrirá?

There's a bunch of stuff going on there, including some unfamiliar verb forms. Could you please break this sentence down piece by piece and explain what's happening here? Thank you!

¡Por supuesto! Este tipo de frase es muy rica en estructuras gramaticales y expresiones idiomáticas del español, así que vamos a desglosarla para entender mejor cada uno de sus componentes.

"Quizás"

• Quizás: Es una palabra que introduce una oración para expresar una posibilidad o duda. Es similar al "perhaps" en inglés. Puede usarse indistintamente con "tal vez".

"lo que sea que tenga que hacer"

• lo que: Esta expresión se usa para referirse a una cosa o situación de manera no específica. Funciona como un pronombre relativo que introduce una cláusula relativa. Equivale al "what" en inglés en frases como "what I need".
• sea: Es el subjuntivo presente del verbo "ser". Aquí se usa para expresar una idea de incertidumbre o algo hipotético relacionado con "lo que". El uso del subjuntivo en este caso señala que no se está hablando de una realidad específica, sino de una posibilidad o suposición.
• que tenga que hacer: Otra cláusula con el subjuntivo, esta vez con el verbo "tener". "Tenga que" es una forma de expresar obligación o necesidad, y el uso del subjuntivo aquí también indica que se trata de algo hipotético o no confirmado. Es equivalente a decir "that I have to do" en inglés.

"¿se me ocurrirá?"

• se me ocurrirá: Esta es una construcción reflexiva que indica que algo ocurre a la persona que habla, a menudo usado para expresar que a alguien se le viene una idea a la mente. El verbo "ocurrir" está en futuro ("ocurrirá"), lo cual indica una pregunta sobre la posibilidad de que algo suceda o se le ocurra a la persona en el futuro. La partícula "se" indica que el verbo es reflexivo, y el "me" señala que el sujeto de la acción es el hablante. Es decir, "will it occur to me?" en inglés.

Entonces, al traducir la frase completa al inglés, sería algo como "Perhaps whatever I have to do will occur to me?" o más naturalmente, "Maybe I'll figure out whatever I need to do?"

Esta frase expresa una mezcla de esperanza y duda sobre la capacidad del hablante de llegar a entender o descubrir qué acción debe tomar en el futuro. Utiliza estructuras gramaticales avanzadas como el subjuntivo para transmitir incertidumbre y reflexividad para expresar que el proceso de pensamiento ocurrirá internamente al hablante.

Ah, yup, it's el subjuntivo!

And I'm pleased to note that I can actually read quite a lot of that grammatical explanation, thanks to English, French, my prior knowledge of grammar, and my knowledge of Spanish so far. I guess ChatGPT-playing-a-serious-Spanish-tutor thinks it's time I start my monolingual transition. But seriously, having that explanation in Spanish is perfectly fine, and I can follow the key points with no problem.

My whole entire argument in Cheating & Consolidating was that it doesn't matter why I understand something like the passage above. Prior knowledge, context, whatever. As long as I can understand what I'm reading, my brain can use that exposure to "consolidate" my knowledge of what things mean, and move it one tiny step close to automatic knowledge. The entire point of this exercise is to bootstrap my way to the point that I can learn Spanish just by reading books and watching TV. I've seen a few early intermediate learners get fixated almost exclusively on intensive study, when they'd probably benefit more from a mix of intensive and extensive study. Sheer volume is what turns slow, laborious knowledge into effortless automatic knowledge. Both parts are important.

LLM advice. Since people have been asking in other threads about using ChatGPT for language learning, here are a few tips:

1. I'm talking to the expensive model, ChatGPT 4. It's $20/month, but it's smarter than the free version, and it's less likely to hallucinate. Substudy uses the cheapter GPT-3.5-Turbo to translate subtitles, but it contains a few tricks to coax out better results.
2. My question is very concrete. I'm not asking general questions about Spanish grammar, I'm asking how a particular phrase works.
3. I have told ChatGPT what character to play. This is a subtle but powerful trick, because ChatGPT has seen thousands of examples of language tutors, both good and bad. And ChatGPT is fundmentally an improv actor, trying to predict how a conversation would go. But the average conversation is, well, average. And we don't want average, we want good! So we tell ChatGPT which character to play, and it just gets "smarter." Simply because we asked it.
4. I'm not asking for too much at once. I see lots of prompts where people ask for things like, "Here are 10 vocabulary words. Please construct example sentences using each of these words, where each word appears in each grammatical case, and not using any nouns or verbs except these." That kind of prompt is "overconstrained". It's asking for too much at once. And the model is doing a high-speed improv act, and it's trying to make up answers "off the top of its head." So you get mistakes and hallucination.
5. Literally nothing I'm doing here requires 100% correct answers from the model. Like, maybe it explains the subjunctive to me incorrectly. So what? If it gets things 90% right 90% of the time, I'm way ahead of the game.

My skull also contains a highly-sophisticated language-learning machine, much of it operating below my conscious attention. If I feed it the right inputs, and ask it the right questions, it will happily build a model of how Spanish works. I might need to fix up that model, because I don't have a child's luxury of 3-13 million words of input a year for 20 years. But an occasional mis-transcribed subtitle, or a bad translation, or an erroneous grammar explanation just isn't enough to throw me off permanently. Natives get all kinds of weird input and occasionally incorrect explanations, too. If we could only learn a language by doing grammar exercises, then we'd still getting eaten by lions. So we should accept that this process is forgiving, and that we don't need to do everything consciously.

Repeating cards aloud, or "shadowing" them in real time. Yes, it's already obvious that this is an useful idea! Maybe not starting on day 1, before I've had a chance to "marinate" in the sounds. But now that I've been working on it for a while, "echoing" the dialog is almost certainly going to be very helpful. Also I need to go look up Spanish phonetics and learn the high points again.

[emk]

Pronunciation notes. These are notes for me; you should probably not trust them.

Resources. Here are some general resources on Spanish phonology:

• Spanish Phonology
• Mexican Spanish (Phonology)
• Spanish language in the Americas
• Pronunication of the Spanish Alphabet (has a good chart at the very end)
• IPA Pronunciation Guide – Spanish
• Learn Spanish Pronunciation (The Guardian)

Here are some IPA resources:

• Wikipedia IPA symbols. This includes links for each sound, and for each place/manner of articulation.
• IPAlab. Clickable IPA chart with audio.

Vowels. Spanish has the classic 5-vowel inventory seen in many languages, including other Romance languages. In IPA, these are: /a e i o u/. This is the same as Italian, except that I don't need to distinguish between open and closed versions of the same vowel.

Consonants

• b & v. [β]
  
  • (After ⟨m⟩ or ⟨b⟩, start of a "breath group") [ b]
• c.
  
  • (Before e, i) [s] (or [θ] in Spain?)
  • (Before a, o, u) [k]
• ch. [t͡ʃ] (like English ⟨ch⟩)
• d. [d]
  
  • (Between vowels, after consonants except ⟨l⟩, ⟨n⟩) [ð]
• g.
  
  • (Before a, o, u) [g]
  • (Before e, i) [x] or [h] (maybe voiced in ⟨agosto⟩?)
• gu.
  
  • (Before e, i) [g]
• j. [x] or [h]
• h. (silent)
• ll & y. [ʝ] (voiced palatal fricative) or [ʎ] (voiced palatal lateral approximant, like ⟨million⟩ or ⟨papillon⟩)
  
  • y (as a vowel). [ i]
• n. [n]
  
  • (Before v) [m]
• ñ. [ɲ]
• r. [r]
  
  • (Between vowels, or end of word) [ɾ] (flap)
  • (Otherwise) [r] (trilled)
• rr. [r] (trilled)
• z. [s] (or [θ] in Spain)

The are probably some errors here; feel free to point them out.

Summary. There's definitely more funky stuff going on here than I remember from, say, Italian. I see a lot of stuff about how certain phonemes change depending on position in a "breath group." But the only real sounds I didn't have already were [β] and [ʝ]. I can can tap just fine (and indeed spent many months trying to eradicate it from my French). I can trill easily; if anything, I need to learn to deliver the precise amount of trill. And I have [x] from German. But things like the decay of "d" into [ð] in certain contexts are a bit of surprise.

At least I have lots of non-aspirated consonants available. That was one of the weirdest parts of learning French—I had to work for my "r", but my brain figured out the non-aspirated consonants all on its own at some point, to my surprise.

There's always a question of when to start investing into accent. Idahosa Ness likes to do it early and extremely intensively. I think this actually feeds into how he listens. By carving the phonemes deep, early on, I think he's actually improving his ability to exactly memorize chunks of sound, and to hear them better. (I do this by applying an SRS-shaped hammer to my brain and relying on "earworms".) But depending on personal needs and talents, it's possible to get the basics down early and to try to fix them up later. Too much extensive reading too early can be tricky, though.

But accents are one of the few things where kids have a very clear advantage. Before 6, anyone can develop a perfect accent. For people who learn between 6 and 12, the number of perfect accents declines. And after 12, developing a perfect accent is extremely rare. But after that point, individual talent varies. There are some amazing polyglots who are awful at accents, for example. Other people are natural mimics.

Still, if most adults put in the work, they can hope to have a tolerable accent. For example, non-native English speakers can learn to sound more like the people around them than (say) a Californian sounds like a southerner.

Oh, and after ~10 years of sitting in my "Saved for Later" cart on Amazon, I got a Christmas gift card to use on Accents and Dialects for Stage and Screen. Which is supposed to be excellent, and which apparrently goes deeper than IPA.

Card reviews. Currently running 20-30 minutes per day, and I expect this to plateau around 30-40. I have started my cards for Eres para Mí, which will be a fun change. Knowing a bunch of songs well is a great asset; I fixed my French "r" using Un Coup D'Œil Dans Le Métro.

Substudy. I have gotten a surprisingly good transcription of the French Taxi, which has lots of colloquial speed and accents from Marseilles. But the translator can't quite handle the entire film, and I want to make some enhancements, for faster and more reliable translations. On the transcription side, I think that Whisper is likely to be at least somewhat viable even at C1 in the biggest languages. It won't get flawless subtitles, but it may still be very helpful.

[tastyonions]

G turns from a plosive into a fricative [γ] or approximant [ɰ] between vowels as well.

Spain goes harder on the unvoiced fricatives (j, x, gi, ge) than most of the rest of the world. They also front CH more, can almost sound like [ts]. A couple ways to pick out a Spaniard even before they use any slang or pronounce a Z/ci/ce.



Greek and standard Spain Spanish share nearly all the same sounds, with the twist that some distinctions that are purely phonetic in Spanish (like [g] versus [γ]) are phonemic in Greek.

[emk]

G turns from a plosive into a fricative [γ] or approximant [ɰ] between vowels as well.

Yeah, I'm also seeing some claims that it's [ɣ] in those contexts, but I'm hoping I can pick that up spontaneously as a voiced [x] between vowels. Not to be confused with [ʝ], I guess? I'm trying to pay pretty close attention to my audio cards to see if I can pick up these sounds. And I'm hopefully aiming at Latin American Spanish, because that's what I'm most likely to encounter in the real world. But there's a ton of obvious regional variation.

Anyway, learning what different consonants are doing between vowels has already helped some with listening.

French note. Oh, and I'm really looking forward to getting substudy to work reliably on full-length movies. My French listening has always been hit or miss on fast, colloquial stuff, especially with regional accents. But it has always been hard to work on this, because (1) I end up needing to order DVDs from France, (2) accurate subtitles are a thing that happens to other countries, not France, and (3) I have no better tools than watching a movie several times and hoping to pick more up. And frankly, the Taxi films and Les Profs do not benefit from massive repeat watching! A lot of them are funny once or twice every five years.

I'm pretty sure that subtitles and even Anki could be used to help around C1, but the workflow would be very different. I'd need near-100% automation, and I'd want my primary interface to be a full-screen video player. If it takes me more than 5 minutes to prep a movie in French, it's not happening.

[MorkTheFiddle]

French note. Oh, and I'm really looking forward to getting substudy to work reliably on full-length movies. My French listening has always been hit or miss on fast, colloquial stuff, especially with regional accents. But it has always been hard to work on this, because (1) I end up needing to order DVDs from France, (2) accurate subtitles are a thing that happens to other countries, not France, and (3) I have no better tools than watching a movie several times and hoping to pick more up. And frankly, the Taxi films and Les Profs do not benefit from massive repeat watching! A lot of them are funny once or twice every five years.

The three colors trilogy (Blue, White and Red) directed by Krzysztof Kieslowski bear watching more than once. Blue especially has a lot of merit, imho. Of the other two movies in the trilogy, Red is better. Kieslowski is Polish but the dialogues of the films and (nearly all of) the actors are French. TBH I can't remember whether I saw the films in French with English subtitles or in English dubbing, but I think I watched the Criterion edition. At any rate, IIRC, the dialogue does not require much heavy lifting.
Other possibilities possibly bearing more than one watching include Belle du Jour (a frigid young housewife decides to spend her midweek afternoons as a prostitute! but it is not all that racy) and Le dernier métro (in occupied Paris, an actress married to a Jewish theater owner must keep him hidden from the Germans while doing both of their jobs.)

[emk]

At any rate, IIRC, the dialogue does not require much heavy lifting.

Ah, thank you for the recommendations! I have a short list of especially frustrating French films, most from about 10 years ago, where I never got my listening comprehension above 75-90%. Most of them involve a mix of regional accents, fast speech, and often comedy. My main limitation here is that I've spent over a decade listening to the same handful of French speakers. I understand them 99.99% of the time. But I never finished putting in the hours for unfamiliar accents and poorly enunciated speech. So my listening is definitely still below C2.

Some interesting expressions. I noticed some really interesting expressions today:



Temes sentir más de la cuenta. This is a nice little detail. "More than the (ac)count", basically.

Que nos viene sucediendo. Do I spy a... verb form? Ooh, shiny. So I just flip to my phone's browser, open up ChatGPT 4, and paste in this phrase.

Pretend to be a skilled language tutor who works with experienced language learners.

In the phrase,

> Que nos viene sucediendo

...could you explain what viene is doing here? Thank you!

Certainly! In the phrase "Que nos viene sucediendo," the word "viene" is a form of the verb "venir," which means "to come." In this context, it's used to construct a present progressive form that expresses an ongoing action that has been happening up to the present moment and continues to happen. The structure is somewhat akin to saying "What has been happening to us" in English.

Let's break it down:

• Que: That
• nos: to us
• viene: comes/is coming (third person singular of "venir" in the present indicative)
• sucediendo: happening (present participle of "suceder")

The present progressive in Spanish usually involves the verb "estar" followed by a gerund (the -ing form in English). However, in this phrase, "venir" is used in place of "estar" to convey a sense of continuity and duration from the past into the present, emphasizing that whatever is happening has been unfolding over time and is still ongoing. This usage is more nuanced and implies a kind of persistence or accumulation of events or states over time.

So, "Que nos viene sucediendo" could be translated as "What has been happening to us," with an emphasis on the ongoing and accumulative aspect of the situation. It's a more expressive way to highlight the continuity of the action from the past into the present moment.

Which, yup, that seems plausible. I don't need 100% accuracy on these explanations, because I'll be keeping my eyes open for future examples of this form. But since this is a public log, well, the internet confirms this. But seriously, ChatGPT 4 has been really great at answering, "What is going on in this phrase?"

Interestingly, French has venir de (inifinif), which is used to express a recently completed action.

But this sort of thing makes up a big portion of my reviews. I'm watching closely for interesting expressions, and keeping track of how things are expressed. No one card, no one translation, no one explanation is the deciding factor—I'm hammering in those earworms, and I'm either proving or disproving many tiny hypotheses.

substudy 0.6.5: Improvements for longer media. There's a new version of substudy! This includes two new features that should help with longer media:

1. Substudy can now run many AI requests in parallel, which should speed up transcription and translation.
2. Substudy will cache recent translation requests, which should make it easier to recover from a translation that makes it 99.5% of the way and then fails with an error.

I've seen slightly mixed results transcribing difficult French video. Right now, I'm seeing both missed dialog, and subtitles inserted where no subtitles should be. But even so, it has gotten me a lot of subs that I didn't have before, including for fast speech with regional accents. I can use this.

Substudy + Knowclip works pretty well, if you mostly want to watch a movie, plus maybe capture a few cards along the way.

And overall? I'm really liking what I can do with substudy, now that I can get transcriptions and translations for a wide range of material in minutes. This is a good workflow, and my Anki reviews are fun.

[tastyonions]

And Italian has its own idiomatic use of venire + past participle to make a passive construction:

i libri vengono comprati = the books are bought

[rdearman]

If I have a DRM-free music file, I can transcribe it, translate it, and turn it into Anki cards in under a minute. Let's start with a transcription:

Hum... It doesn't seem to work well with a Korean song

Code: Select all

substudy transcribe --example-text Love_Letter_with_you_.txt Love_Letter_with_you_.mp3
Found dialog breaks                                                                                                                                                                                                                          Extracted media items        1/1    00:00:03 ███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████ 00:00:00 Transcribed dialog!          1/1    00:00:09 ███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████ 00:00:000
00:00:04,478 --> 00:00:18,840
나의 온 마음 앞에 서상이잔
바람이 문을 두드리죠

1
00:01:06,340 --> 00:01:22,795
With you, I could die
for you, 바람이 되어 그대

2
00:01:22,845 --> 00:01:26,521
더 높이 오를 수만 있죠

3
00:01:42,818 --> 00:02:02,270
With you, I could die
for you, 바람이 되어 그대

4
00:01:58,179 --> 00:02:04,831
I could die for you,
바람이 되어 그대 더

5
00:02:41,820 --> 00:03:10,320
With you, I could die
for you, I could die


Here I'm actually giving it the entire song lyrics as the example text. (Didn't know if you wanted me to post to github or not). This song (like many Korean songs) has a mixture of both Korean and English.

[emk]

And Italian has its own idiomatic use of venire + past participle to make a passive construction:

i libri vengono comprati = the books are bought

Oh, cool! That's a good thing to keep an eye on.

Hum... It doesn't seem to work well with a Korean song
...
Here I'm actually giving it the entire song lyrics as the example text. (Didn't know if you wanted me to post to github or not). This song (like many Korean songs) has a mixture of both Korean and English.

Thank you for your feedback!

Unfortunately, there are several different things which might be going on here:

1. Mixing of languages in short snippets is likely to confuse Whisper.
2. I think that Korean is supposed to use spaces. But if your input doesn't have spaces, you need to switch to "substudy transcribe --format=whisper-srt ..." to use raw Whisper SRT output instead. This is a bit noisier and has worse alignments, but it should work much better for languages without spaces. If you want to try doing some raw data cleanup of your own, there's also "--format=whisper-json", which is what I actually get back from Whisper.
3. This particular song might be too hard. Whisper theoretically supports Korean, but I don't know how good it is.

If all else fails, you can set RUST_LOG:

Code: Select all

# Print out information about various oddities.
export RUST_LOG=substudy=warn

# Too much information.
export RUST_LOG=substudy=debug

# Way too much information.
export RUST_LOG=substudy=trace


This will give you quite a lot of detail about what's going on.

If you're still failing on a monlingual Korean song, and if your lyrics have spaces, then please feel free to send me the artist and song name, and I might buy a copy of the song to try myself. Especially if it's catchy!

Bertalign. The workflow for turning video and audio into subtitle cards has gotten pretty smooth. Sure, you need to know how to use the command-line, but once you get things to work the first time, converting another song or video is 3 commands and 5 minutes work.

But aligning books is still a bit fiddly.

Kundalini wrote a great tutorial.

Using the Google Colab notebook (https://colab.research.google.com/drive/123GhXwgwmQp1F5SVZ74_uIgyxo6hLRq0?usp=sharing), it only takes about two minutes of work (and some wait time) to align txt files. Then I was able to export the aligned file (The Count of Monte Cristo with alternating sentences in French and English) as epub to put on my ereader. The results aren't perfect, and there are some artefacts, but they are more than good enough for my needs. I'll post a guide in a separate thread for others who may be interested.

I wanted to dump the aligned ebooks as JSON, so I modified the original notebook to do this:

Code: Select all

es = open("../book.es.txt", 'rt', encoding='utf-8').read()
en = open("../book.en.txt", 'rt', encoding='utf-8').read()

aligner = Bertalign(es, en)
aligner.align_sents()

print(aligner.result[0])

output = []
for (src_indices, tgt_indices) in aligner.result:
  src = [aligner.src_sents[i] for i in src_indices]
  dst = [aligner.tgt_sents[i] for i in tgt_indices]
  output.append([src, dst])
 
import json
with open("../aligned.json", "w") as outfile:
  outfile.write(json.dumps(output, indent=2))

Now I have an aligned ebook, which I'll be able to use for all sorts of things! But there's no easy way to integrate an aligner into substudy, so see Kundalini's tutorial if you're feeling like a bit of Python scripting.

The other option is to take monolingual ebooks at input, and rely on ChatGPT 3.5 or Claude 3 Opus to translate them. That would cost more per book (probably) and take a while. But it would be easier to implement. Probably not worth it, overall.

Super Challenge. At this point, I'm basically ready to start a 2,500 page Super Challenge. I can make bilingual ebooks, and I can wade through them slowly. But experience suggests that the first book or two should make a huge difference.

[emk]

Reading Spanish. As an English speaker learning a Romance language, I find it personally helpful to divide vocabulary into several major categories:

1. Spoken language. I've been getting plenty of exposure to this. Lots of spoken language is fairly simple, with a limited core vocabulary. But the trick is finding spoken language in a beginner-friendly format. Hence my focus on subtitles and SRS.
2. Non-fiction language. Newspapers, non-fiction books, etc. This is very heavy on the cognates. Usually not a major problem in a Romance language. It's everywhere, and if the subject matter is familiar, it can be predictable.
3. Fiction language. In some ways, this is the hardest of the three. It uses a larger vocabulary, with more descriptive words. And in Romance languages, it doesn't veer quite as heavily into the Latin cognates.

I'm trying to jump from (1) to (3), and I'm trying to do it too early.

Last night, I tried reading Spanish fiction on a Kindle Paperwhite. I often needed to look up two words per sentence. But at least the Paperwhite has much better dictionary support than Kindle for Android.

I prepared a bilingual interlinear ebook, as discussed up thread. Bertalign is fantastic. And the interlinear format is better for me right now, but it requires a lot of discipline to use effectively.

A better format might be facing columns. And even then, I'd love to hide the facing column under the user asked to show a sentence. Make it a bit more like flash cards, basically.

But if I want to use the Paperwhite, I need to live with some limitations. Maybe I should read a chapter in bilingual format, and then go back and read it again in Spanish.

Intensive vs extensive balance. This is a tricky tradeoff. In the beginning, pretty much everything is intensive. But once you reach the intermediate level, you have the option of just diving in and letting some stuff pass over your head.

And one reason I think that many first-time language learners stall out at the intermediate level is that they focus too much on intensive study, and they do almost no extensive work. But the problem with intensive work is that—in the occasionally overused terminology of my industry—it doesn't "scale."

For example:

• My audio cards are intensive. Extremely so, in many ways. I can expect to internalize 90% of the material on these cards to the point where I can understand it with my eyes closed. But I can only learn 10 to 20 new cards per day if I don't want to devote my life to Anki.
• Reading a bilingual book leans more towards extensive reading. I try to puzzle out the Spanish, glance at the English, and then try the Spanish again. I'll look up some words here and there. But I let a lot of stuff go, and focus on moving forward. Each time I see a word, I maybe increase my "knowledge level" by 5%. But I see a lot of words. And the most common words will be reinforced pretty quickly, over the course of 500 pages or so.

There are no real rules to this, only flexible rules of thumb. I pick something interesting that feels tantalizingly in reach. I figure out some sneaky trick to boost my comprehension as much as posible. And then I try to do a bit every day.

The other big chalenge—no matter what point you choose to tackle it at—is that the first book is a slog.

Active stuff. Probably on hold for a bit longer while I focus on reading before the SC starts.

EDIT: The bilingual book seems to be working fairly well, after a few pages. I'm adapting to the language and the format, and I'm learning to glance at the English just long enough to get a word or two. I think this is workable, though 5 pages a day (the half SC pace) will be a lot in the beginning.