Do you have an exit plan for isolated vocabulary study?

Don't know about Luca. Although Dr. Arguelles and Linguamor did not use nor compare with Anki, they both said or implied they read for many hours a day as learning required massive exposure. Only Linguamor gave figures, which might be familiar to you since you quoted her on HTLAL . While for the purposes of the comparison below, I will accept her figure for ER, I don't accept her conclusion, simply because she did not know the time figures for someone using Iversen word lists or Anki lists every day.

Super-fast vocabulary learning techniques:

She gave the same figures in another thread: Better to read without translation?

So she said 15 words per hour learned with ER, which she said was faster than could be done with word lists. As you know I am conducting an experiment using Iversen's method up front, and then Anki as a control for well-learned words and to mop up those not learned. I believe Iversen has said he can learn 30-40 new words in 20-30 minutes, and has an attrition rate of 15%. So far, not being efficient with associations as stated earlier, I am taking twice that for the up front work, and half that in addition for the subsequent reviews the next day (Iversen method) and the day after that (Anki). 15% is my best case scenario for attrition/loss, and I will guess and say the end result when I tally up all my figures to a less impressive 25%.

Now lets compare the maths:

Iversen: 30 words in 20 minutes plus 10 minutes review next day =30 words/30 minutes = 60 words/hour * .85 retention rate =51 words an hour (note he has said in the past he couldn't keep up a very fast pace and learn 2000 words in a fortnight - How long for your to learn 2000 words?)

Peregrinus: twice the time for less retention = 30 words/hour * .75 retention rate = 22.5 words per hour (I am trying to be conservative but my experiment is not over yet). For Anki alone, as a rough approximation using 2 hours per day for a year to learn or partially learn 10K words with no words omitted via attrition = 10,000/(2*365)=13.68 words /hour (but not all learned as completely as with the Iversen method).

Notice that Iversen's rate is 3+ times Linguamor's ER rate, and that my rate for the hybrid method, unimpressive as it is, still looks as if it will end up being 1.5 times her rate of 15 words/hour.

N.B.:
1) The rates for Iversen and myself are explicitly for passive knowledge, and Linguamor did not claim otherwise.
2) Iversen and Linguamor both know many related languages which allows for better morphological associations.
3) As far as I know, Linguamor made no systematic attempt to verify her long-term retention rate and I am giving her full credit, i.e. no attrition rate.
4) A couple studies have shown that ER is not as effective as Linguamor claimed it is (I'll have to dig those up), i.e. that her figure asserted elsewhere and similar to Dr. Arguelles', of needing 4-5 exposures via ER for a word to stick, are (far) too low, at least for average learners.
5) My rough approximation rate for Anki alone is less than Linguamor's claimed rate for ER. While I generally could not spend more than 2 hours a day on Anki, if even that, she apparently was able to devotes several hours a day to reading, and the same for Dr. Arguelles. So they probably learned words faster in calendar time by virtue of being able to devote more time per day to the task.
6) My figures are for the first language (German) that I have used Anki for this extensively (less for Spanish), and the first language rates to be harder and subsequent related one easier, with any method.

Edit to add for one of studies I had in mind re ER:
Nation & Waring (1997): Vocabulary Size, Text Coverage and Word Lists

HTLAL thread discussing this: Amount of Vocabulary Research

Poster irrationale's analysis:

Kind of… sort of… maybe? If I pull a sentence from a novel or a newspaper article (my two most common sources), I still remember much of the original context, even a year or two later. "Oh, that was the annoying political editorial about French actors with des grandes gueules, and whether the government was fawning over them sufficiently," or "That's from the scene in Taxi when the cops arrive at the pizza joint, and he starts making jokes about pizza poulet." When I can remember the surrounding context like this, it really does make it easier for me to remember a word and exactly how it was used. So for me, Anki sentence reviews are more like an ultra-concentrated form of extensive reading.

I have a model for receptive language acquisition that I've mentioned before on HTLAL. It goes something like this:

opaque —(1)—> decipherable —(2)—> automatic​

Isolated vocabulary study helps with (1), the transition from opaque to decipherable. Extensive reading helps enormously with (2), because it exposes the reader to decipherable text so many times that the process becomes automatic and instantaneous. But extensive reading does help with (1) as well, because as more vocabulary and grammar becomes automatic, that will make previously opaque portions of the text decipherable.

Now, you can optimize for (1) by studying huge amounts of vocabulary in isolation, and that may make lots of sense for many people in the beginning. But that still ignores (2), and so there's a good chance you'll find yourself doing laborious dictionary lookups in your head when you're trying to read. To correct that, you'll need to tackle (2) eventually. On the other hand, extensive reading by itself might not be the most efficient way to tackle (1), but it's not horrible, and it also provides massive amounts of (2).

Personally, this is why I personally like "extensive reading plus distillation", using Anki sentence cards. It gives me tons of practice with (2), but it allows me to mix in as much (1) as I want by setting aside short snippets of text, looking up unknown vocabulary, and then reviewing those snippets at rapidly lengthening intervals. The point is that this process mixes two entirely different things, and I can adjust the proportions as needed. This allows me to tackle French, where 99.9% of the vocabulary in a typical text is either known or trivially decipherable. But it also allows me to tackle Egyptian, where my vocabulary is a few hundred words and the grammar is delightfully unfamiliar. The specifics vary considerably—I make heavy use of parallel texts with Egyptian—but the overall process is largely the same.

I'm not saying that this is any kind of miracle method. It isn't, and I'd happily believe that Iversen's method is faster. But nor do I think that my approach is massively slow or inefficient. When I started the Super Challenge, it took me several evenings of serious reading to finish a volume of Tintin. Twenty months and ~40 books later, I was comfortably reading 40 pages/hour with an opaque word every other page. Perhaps this was an inefficient use of my time, but if so, the waste was not so ridiculous as to be irresponsible. And in any case, I enjoy reading, so it was only a minor sacrifice at worst.

Language learning doesn't need to be fun, and some people make very rapid progress using rather brutal methods. But a mix of shameless-but-sustained hedonism and occasional study can actually work pretty well for many people, too. The "correct" answer for a given learner probably comes down to whatever keeps them motivated, since far more people fail by giving up than fail by using moderately inefficient methods. Language acquisition is quite robust, given large amounts of comprehensible input, a need to communicate, and a sufficient quantity of time.

I'm definitely part of the "fun crowd." I don't really mind if my methods are somewhat inefficient compared to Anki or FSI drilling or whatever more "disciplined" thing we care to discuss, as long as they are enjoyable. I mean, I am not on a deadline or anything, and in the areas that matter to me (talking with natives + reading and listening to native media) my progress is reasonably fast. It's possible that if I tackled a very different language I would have to be more systematic about things...if I decide to hit up Arabic or Japanese or Turkish in the future, we'll see.

You remember that I have phrases and sentences on my cards right? And that though not all, a large proportion of my vocabulary was taken from DW, LL and other courses, and now from a novel? And that I read/listen every day to various German media to reinforce my vocabulary?

But let's say I learned a certain word in fact in isolation. Bare vocab card with no context and definition I found in a dictionary entry with no context (the reason I use more than one dictionary). But you learned the same word in some context like a novel or a newspaper article. Obviously, and a point Linguamor stressed, you get to learn some practical usage of the word with surrounding words, what she called how the words pattern together. If Anki is the only place that I will ever come across that word, then you win. But barring perhaps some obsolete vocabulary, I will in fact come across the word again in context, and will have an initial grasp on its meaning which can and will be adjusted by that context. Additionally it is very likely that I meet that word in context, already knowing more than one possible meaning, since I use detailed definitions, and will quickly recognize which applies.

But I mostly do learn with some context, and below is a word I harvested from that novel yesterday and entered into Anki:

Front:

aufraffen

sich aufraffen

sich zu etwas aufraffen
ich kann mich dazu einfach nicht aufraffen​

Back:

to pick up, gather up, snatch up

to brace o.s., pick o.s. up, pull o.s. together, bestir

bring /rouse o.s. to do sth
I just can’t be bothered​

I have already learned the base word raffen which means to grab, snatch up, amass, close to this typical German derived separable form. While on the first couple Anki reviews I won't memorize those phrases, sentences, and only try to get an initial toehold on the meaning, with subsequent reviews I add to my memorized definition, and especially will want to internalize the use in the reflexive form, i.e. sich aufraffen. Most often in German, a reflexive form used transitively is fairly clear. But the intransitive use is somewhat colloquial and not as easily guessable. So I think I have a better chance on first meeting in subsequent extensive reading to understand that meaning nearly perfectly, than if I had not previously learned that intransitive reflexive use.

Again, it will take me several Anki reviews for all that to be internalized, but those reviews will happen in fairly short succession. Whereas with ER alone, it could be a long time indeed between meetings. So although as I have said I will at some point likely stop Anki for German except for specialized thematic areas perhaps, and thus don't intend to learn 100K words in Anki, for those I do learn in Anki I think it more efficient to meet the word later in reading already having a grasp of more than one possible nuance of the overall meaning, and will have learned it more quickly, simply by virtue of the fact that relatively infrequent words like this one can only be met multiple times in context with truly massive exposure, which Anki artificially provides.

But Peregrinus, again, your example shows the vocabulary item in only two contexts of millions. That's only 2 more than 0, and it can't result in a balanced, rounded understanding of the word. Also, neither of your examples actively demonstrate this is a separable verb. It's very, very difficult to get all the information in there....

That would be a problem if I never followed up the initial learning of Anki with extensive reading and listening and continued to add to my understanding of the word in actual usage. And with low frequency vocabulary like the example I gave, they don't tend to have many distinct meanings. And it will generally take you far longer to come across just those two meanings in ER alone because of the low frequency.

I wasn't trying to get all the information in there, only that which I don't know. I don't try to copy everything from a dictionary entry. I know from looking at the verb that it is separable despite it not being used thus in the 2 examples. And I know how to use and understand separable verbs, or other aspects such which auxiliary verb they use depending on transitive or intransitive use.



Since all of us who are advocating Anki, Iversen word lists or Gold List, in fact intend to follow up the initial vocabulary learning with readings in a course, specific text or just free extensive reading, then perhaps the easiest way to conceptualize the methods is to imagine them to be a form of extensive reading, just preceded by some initial vocabulary learning to make the ER process (much) more efficient.

If we had no such followup, then that would be isolating and also limiting as to learning more complete usage of those words. To use another metaphor, we grease the ER tracks before setting off, and as a result, don't need to encounter low frequency words nearly as often as required for such vocabulary to stick, as would be the case with ER alone. We are merely artificially adjusting the frequency.

See how many words you have to read and how long with free ER it takes you to encounter that verb even 4 or 5 times. Get back to us hundreds of thousands words and months later and let us know a more precise figure (excepting specialized contexts).

You sure you didn't just strawman yourself ?

With Anki, Iversen or GL, you are focusing on the initial, partial and incomplete learning of a word. That learning continues with the use of that word in course texts, selected texts from which the vocabulary was taken, or free extensive reading. Yet you are not doing this with ER, i.e. you are ignoring how many times you have to come across a word before it sticks, and how many thousands and thousands of words you have to read for that to happen, and thus the time for same also.

If I see a new word or word combination in a text I may be so lucky that there is an explicit explanation, or the context may suggest a meaning which some learners might remember precisely because it is 'unfinished business'. Others, like me, would find it irritating not to have confirmation, and I am fairly convinced I would be less likely to remember dubious halfknown words. Unfortunately that will be a large portion of the vocabulary in the things I read, and potentially maybe even the most relevant part of it. I understand how emk can remember a number of interesting words/combinations by the place he saw them - I also have a store of such memories. But the interesting things to notice here are 1) the number of words you can connect to their sources is probably limited, 2) it is highly questionable whether you also remember less pittoresque terms. I personally don't remember the first time I saw "poulet", but I remember "tripes" because I was dumb enough to order it during my first visit to France without even knowing what it was - revolting scraps of stomach walls and other intestines in some gooey soup. One word learned the hard way - and a deep distrust of French cooking! But with a few notable exceptions of this type all vocabulary I just see once somewhere is forgotten to the extent that I don't even remember having forgotten it. It is simply gone with the wind - and I have no way of knowing how many words I have let slip away in this manner, but it must be thousands of words. You can try to retain those ephemere terms by taking notes and lookings words up, but if you can live with that you can just as well go all the way and use wordlists or SRS methods.

Wordlists/SRS based on intensively studied texts are efficient in making sure that you don't just let your new words disappear without a trace. What then about wordlists based on dictionaries? Well, I have already stated that I see any formal vocabulary training as a preparation for later activities, but it is more than that. It is also a way of putting ressuscitation of once seen, but immediately forgotten vocabulary into a system. Many of the words I see in a dictionary ring a bell, and if I think I can remember them without further ado I may not even including them in a wordlist - seeing old forgotten words in a dictionary may be enough to put some life back into their desiccated carcasses. And I also often have the experience that I hit upon a word in a text or podcast and remember that I once used it in a wordlist. So that experience is not just something you can have with literary works.

It depends. I can think of several case scenarios, for example:

1) You used SRS to help you out in your way to proficiency (C2) in a language that you keep using every day (because you're learning English or the language of the country you live in). You're deck is learned in terms of SRS and the average card in it is scheduled for August, 2025. You made your flashcards yourself and many of them reminds you of that lovely novel you read about WWII or the third season of Lost, so you grew really fond of them. It takes you about 7-8 minutes a day to keep 15k cards and it's real fun. You don't need SRS that much any more, but what sort of criminal would you be if you get rid of it all?

2) You used SRS to help you out in your way to basic fluency (B2) in a languages that you're never going to use much (it's just your n+1 language), so the real challenge is to maintain it. You think that your 15k cards deck might come in handy for that, but on second thoughts, the whole thing is becoming a nightmare: the daily review takes forever, you keep failing cards, most of the cards came form pre-made collections on the Internet, so the damming thing bores you senseless, but worse, worst of all, please awake if you fell sleep, who in their right mind would read twice the same book? No one, of course. Chances are that the second time it will contain the same words in the same order. That's isolated vocabulary!!! Please think that if you add a sentence to your SRS collection you're going to read it not just twice, but 7-8 times, maybe even 10 times!! So you end up deleting the deck because you'd be better off just reading 'real' stuff.

That would be Jan-Arjen Mondria's fifth myth of vocabulary acquisition (remember them?). I don't believe in this idea that we all learn differently, and the figures seem to back up the idea that the brain doesn't really start learning until it has some kind of certainty.

And this also leads to one of my favourite cognitive gotchas: quite often, a technique seems effective to the learner because he recalls doing it. But events we remember are events that are in some way unusual.

You might remember what you had for dinner on your wedding day, but can you tell me what you had for dinner the other 364 days of that year?

I remember how I learnt to stop confusing casado and cansado after embarrassing myself in a pub. I remember how I learnt the Spanish for hazelnuts, via a similarity to oats, and a lot of confusion with a box of muesli. I can tell you that I used the mnemonic "ankle biter" to help me remember the French "un cobaye" (but I can also tell you that I remember the mnemonic better than the meaning.... it's either a hamster or a gerbil, but I wouldn't know which).

But there are thousands of other words I've learnt and I couldn't tell you how... and as these are the normal ones, they're probably more important.

That Mondria study is a great resource, and one that I have previously linked to myself. But I think you are misapplying it to what Iversen said. That 5th myth is actually a slam against extensive reading and learning from context alone. Yet that isn't what Iversen said if I understood him rightly. Even if he infers the meaning of a new word from context, he thinks it less likely he will remember it without confirmation (looking up or studying). And that is precisely the problem with ER, one may make a mistaken inference of meaning and try to adjust to that in the future, or have it take a really, really long time to come across it enough for it to stick even if one does correctly infer the meaning (of a totally new word that is not a regularly derived variant of a know root).

You are essentially just restating what Iversen said. Less pittoreseque words are hard to remember = ordinary words not connected with an event to intensify them are harder to remember. Which is why he uses word lists and artificial mnemonic associations with same.

I certainly remember Mondria. And I think my methods are in full accordance with almost everything he writes during his debunking of a number of popular myths in language pedagogics. And as Peregrinus points out even Cainntear actually doesn't contradict the things I wrote above, namely that things you trust probably are more likely to be remembered later and that memories concerning single words probably are biased in several ways. I do however admit the possibility that people who don't do wordlists or SRS may be more likely to remember things they had to infer, but the reason could then simply be that they spend more time and more thoughts on things they find puzzling than on things they can guess immediately. And the sad consequence would be that they then remember a lot of most fuzzy, shaky and unreliable word meanings, which they then have to get rid of later.

As expected, Mondria rejects the myth (no.2) that wordlists should be of limited value. However he does seem to have a fairly narrow concept about them: "word lists and frequency lists are excellent checklists for determining which important words learners do not know yet". Instead he advocates the use of something that resembles flashcards and SRS systems. Personally I think that there is more pedagogical potential in wordlists than he makes allowance for, but it is worth stressing that he finds wordlists valuable even in their most primitive form. It would have been interesting to know what he would have found with a broader and more flexible wordlist concept.

OK, I was skim reading, and the measures they serve whisky in here are larger than back home. (I've been saying goodbye to various people all week.)

I misremembered Mondria's argument - either I confused it with some else's or I added my own interpretation, that the additional time taken to learn context-inferred vocabulary is down to uncertainty.

I never try to infer from context, because I'm wrong as often as not; instead, I keep a dictionary beside me while reading. I don't look up every word, I just (if the night was as black as a plergly-plink's whatjimadoobrit, I don't need to know the words to understand that it was a very, very dark night).

Ô_o
Do I come across as so fractious that you can't accept that I might agree with someone else from time to time?

But actually I was expanding on what he said.

Iversen's point is that certain techniques are only effective with intrinsically memorable materials/techniques. I was trying to add that we are cognitively biased to be aware of such techniques as a side-effect of he memorable material, and that makes us more likely to identify them as "core" techniques and advise them to others as such, because we are blind to how insignificant they really are.

I think the problem with SRS is that it's too easy to overdo it. Nothing is fine, a little is a help, but too much and you're shooting yourself in the foot. You have to be selective, or else you quickly slip into a pattern of adding everything and you become a slave of something that was meant to be only a tool. Someone talked about inflections, well, you can add inflections, why not? Actually SRS is all about the skill of deciding what cards have to be made and how. If you really want to own a word, expression or grammatical feature, you have to ask yourself how many facts are there to be learned, and then go on to make one card per fact. One card per word, per sense, per inflections, per syntactical feature. In highly inflected languages it's up to you how deep you want to go with this, but obviously the more depth the less words you'll be able to manage.

So, I think it's a bit nonsensical the complaint that's always the same context, the same inflections: add another, mate, but what really puts me off is how SRS critics seem to completely overlook the spacing effect. If you know nothing about SRS and you hear them talk about people practising the same phrase with the same context and the same inflection over and over again during a lifetime, you'll think oh God, those fools, how may reps are they doing for each word, millions?, maybe even trillions? In fact we're talking about 10-12 reps if you don't do unnecessary 'learning' reps (I for one don't do them). Just as an example, I'm gonna show you the furthest away scheduled card in my English DB at this moment:

Q: a person whom you know well and whom you like a lot, but who is usually not a member of your family:
So I'm Kim and I'm sexually obsessed with my best ____ Sugar.
R: friend

Next rep: in 18.3 years
Last rep: 1 month ago
Learning reps: 1
Review reps: 7
Lapses: 0
Created: 08/25/2007

Now, some of you might be thinking, oh God, this guy has always seen the word "friend" in the same context, he surely ended up thinking there's something sexual connected with friendship. If that's the case, please do the maths: 8 reps x 10 sec = less than 2 minutes in 25 years. In fact, I must have seen or heard or said the word "friend" thousands of times in real contexts for each time I've seen it in the program. That's what makes it an 'easy' card, but I wouldn't have known that when I created it, would I? But what if I don't come across the word that much? What if it's a 'difficult' word? Well, maybe that's when SRS turns out to be most useful, and if it comes down to a choice between seeing a word several times with the same context or not seeing it at all, well, it's up to you.

Using SRS as a "memory amplifier" improves my ability to remember these contexts considerably.

Used well, the forgetting curve is really powerful. But I think the standard theory of how to exploit it leads to needless misery. The idea that you should create a big database of facts, and review them just before forgetting them is pretty painful in practice. Letting memories decay to the point of near death and then laboriously pulling them back from the brink might be efficient, but it's also pretty miserable.

Over the course of more than 31,000 Anki reviews, I've stumbled over a model that I find far more agreeable:

1. Ideally, cards should be made from things I would enjoy re-reading or re-watching.
2. Cards should be easy to make in large quantities.
3. Cards should be easy to answer correctly.
4. Cards should be disposable.

This isn't about fighting an epic battle against my memory and saving each word at the last possible moment. Instead, I'm exploiting the effect that causes annoying song lyrics to get stuck in my head. Here, the standard forgetting curve is transformed into "the minimal review interval necessary to create an earworm." And any vocabulary which I can't learn easily gets deleted.

I just wanted to say that I love this. My use of Anki isn't quite this lovely and stress-free, because I sometimes ignore it for a while and spend a weekend catching up. But you've definitely captured the overall feeling.

This can be optimized, however:

1. Use materials that are intrinsically memorable.
2. Exploit SRS software as a memory amplifier.

Right now, I'm on page 384 of a French action thriller, and I've marked 84 passages that contain an unknown vocabulary word. There were perhaps another 20 unknown words that I left unmarked, generally because they were obviously useless—a local Caribbean word for "hill", a kind of African dance, a noun derived from a brand of shipboard deck chair. There were also other unfamiliar words whose meanings were clear from cognates and context, most of which I left unmarked, because those generally take care of themselves.

Overall, I would say that I've succeeded at acquiring the most common French vocabulary (though I still have more work to do, of course). Looking at the combined statistics, I have 2,805 French cards (more than I thought) and 20,606 revisions over the course of 90 hours during two and half years. These cards were created over the course of 10,000+ pages of extensive reading. I don't have any good way of estimating how many words I know, but online vocabulary tests suggest 20K+. I usually round this down to "more than 15K". Still, this suggests that I learn over 80% of my vocabulary through non-Anki methods.

If you like to read a lot, it's a surprisingly good strategy for an English speaker learning a Romance language.

I'm actually very laid back about this. I have cards with a passage containing three boldfaced vocabulary words on the front, and I mark them as "pass" if I can understand the passage without looking at the definitions. When I miss a card, sometimes I'll mark it as "Hard" instead of "Failed", or just suspend it permanently. And if a card makes me say "Ugh", I just go ahead and delete it. If I actually need to know the word, I'll just wait until the next time I see it and make another card.

I first discovered this approach to Anki when using Subs2SRS. Subs2SRS will take video file and bilingual subtitles, and output roughly 1,000 cards for the average movie. Out of those, perhaps 100 cards will be useful at any given level. So I very quickly learned to delete aggressively. There's a great feeling of freedom in knowing that I can easily make far more cards than I could ever review, because then I don't care very much about any given card—it's OK if I forget or delete, and it's OK if I abuse Anki so much that some words slip through the cracks. Whatever. Anki still gives me an enormous boost over my natural memory, and 90 hours of reviews in 30 months is a pretty low price to pay:

I've come to the same conclusions. When I fail a card, I try to add more clues in the question and make it clearer so the next time it will be easier. Sometimes you just have to get rid of it, but in general the less the question resembles a question and more a reminder, the better. A question may have multiple answers, but we are not talking about theoretical questions here, what we want is retrieve a very specific memory. Maybe that's not ideal in terms of knowing a word in multiple contexts, but when you have a phrase fresh in your mind, like when it's part of a catchy song, chances are that something from it pops up when your talking or writing, and the pattern-capable part of your mind will choose the appropriate case or inflection for the current context. So, the initial specific memory grows and expands as if it was a seed.

Now define "fact" in such a way that it is possible to make one card per fact.

I'll give you the benefit of the doubt and assume this is a non-native, thing, but I'll warn you that using "mate" that way is extremely patronising.

Anyway, the problem with just adding another is that you're overlooking...

...the spacing effect. If you put two cards with the same feature, you break the algorithm. If they turn up in close sequence, the second one will be easy by association, yet the software will consider them individual "facts" retrieved from long-term memory and won't realise that the second one's only easy due to short term memory

Have you got "strawman" in your deck? If not, you might want to add it...

There are two problems with your argument:

1) My point about "one context" is that it's often presented in opposition to contextless words, with the argument that lack of context leads to making errors. If that's true, all "learning in a single repeated context" can do is reduce the number of possible errors by one, which is a pretty insignificant improvement.

2) Just because I didn't die last Friday doesn't mean that my landlady's iron is safe - it isn't. It's a steam iron, so it has a water reservoir, and it leaks. An electric device leaking water is inherently unsafe. A learning technique that leaves open the possibility of making mistakes doesn't invariably lead to mistakes, but if we can remove the risk, we should. For example, I object to wordlists because sometimes - not always - you end up remembering the list as a sequence, not a collection of independent items. That Iversen doesn't have that problem doesn't mean that it isn't better to use a technique that doesn't impose an arbitrary order on the items (eg flashcards, SRS).