How to learn a thousand words each one-two months

[Pall]

Hi,
In the mid-summer I made a proposition to add to transcription a radical of the syllable. I proposed even to add its name to the sound of the syllable, but the latter appeared exessive. However the idea to add something in front of the syllable transcription proved to be fruitful. I decided to add firstly some leading feature of the character. In the attached list of the 5,000 HSK words (with some extra words)  I arranged them in six parts. In the first four parts words are distributed by the leading feature of one of their characters.  The leading feature is something, which catches your eye first of all. It occures firstly on the margins of the character, at the bottom, on the right, at the top and on the left, I preferred going clockunwise, see pic. Some of the leading features coinsides with radicals, those that are exclusively distinct, others not. For example, in the part 3 item 6 you can see cases when two strokes happen over the character  above some line, which can be seperate or a part of a figure. When there are two strokes it is 八 radical, which also can happen at the bottom and in other forms, but that's ignored. In the part 3 item 5 all these characters are considered of the same leading features: 亡, 交, 方, etc. 
The parts 1-4, in which leading features can be seen easily, covers some 3,000 words of the HSK list. Corresponding radical or a symbol for a groop of leading features is put in front of the syllable in transcription. For the groop of "creepy" a five-end star is used with a line at the bottom, for "chandeliers" - crossed W. 
The part 5 includes words of some rather frequent radicals. In the part 6 the rest of the words is organised in five groops: 1) thoses having somewere in their outline the elements ヌ, 夂 or 攵, 2) having somewhere 人, 入 or similar element like in 个 (dispite it's considered 八 in fact), 3) having some square or rectangle with something inside (in the pic. it comes No 4, but it's No 3 instead - all these are relevant: 中, 里, 翼, 间 ), drawn as double crossed square, 4) having some square or rectangle , which is not the 口 radical - can be a part of another element (况, 舔, 点), shown as a square crossed with a single line, 5) just "lines" - all the remainder - not indicated at all. 
Also the fleshcard type is important. On the face side the transcription with leading features is written, on the reverse translation at the top and the word in characters at the bottom, with some horizontal space between them. Firstly they are learnt in these two directions simultaneously: from the transcription with leading feature (for each syllable) to the translation and to the character outline, which is to be drawn by hand. Then looking at the other side and covering with the big finger the translation we train to recognize the words in characters. I do it even holding the fleshcards in the upward position, then it's also more covenient to cover the translation with my finger. When revising the words difficult ones should also be learnt from translation to the characters, which are to be written by hand. In addition I duplicate fleshcards for difficult words changing the order of the meanings. It's important to reduce the number of meanings to one-two. However because of the synonims it's better sometimes to give a couple of corresponding translation words even for a single meaning. 

The words are learnt mainly by sections of the same leading features, though often mixed for several of them. 
That's all in principle. Also it worths to mention that I used mupin as transcription. It seems to be better percepted in the combined form of the transcription, with leading features. Yes, it also can be due to that I'm familiar to Cyrillic. However, I'm quite familiar to Lathin cript, and I learnt pynuin well before I encountered at mupin, and it take efforts to switch to the latter finally. Anyhow you can start with pynuin, if it works OK, then no problem. 

See rearranged HSK list enclosed, it's not only "radical organized" now.Radical_Organized_New_HSK_Wordlist1.xlsx 
 

Radical_Organized_New_HSK_Wordlist1.xlsx

[Pall]

It turned out to be even easier. Words are learnt in an unusual way. Firstly, they are learnt in the direction from the "sceleton", i.e. transcription in mupin plus some indication in front of each syllable, to the translation. After several days of reviewing them they are checked for recognition of the "sceleton" from the translation. The recognised words are put to the horizontal pile, the unrecognized words remaning in the vertical pile. It's not necessary to write the sceleton by hand, it's enough to reproduce it orally plus imagine the indication - the leading feature or something instead. Checking for such recognition happens every several days, the review for remembering translation from the "sceleton' is underway constantly. When almost all the words of a section appeared in the horizontal pile, they can be checked for recognition of the words in characters. It's a sort of mistics, the result of the recognition is 80-90% despite the fact that words have been learnt from the sceleton and to the sceleton only. So, such approach saves much efforts since dealing with the word sceleton is a way easier that that with the word in charcters. The words, which are nor recognised from the characters, are either put back to the vertical pile, or learnt in traditional way.

[Pall]

No, I was wrong, but not completly. The order of the first two actions should be the opposite. Firstly, we review the fleshcards in the initial vertical pile of the section to recollect the 'sceleton' by reading the translation while a finger or something covering the characters. The 'sceleton' is reproduced orally with imagining in one's mind the indicators before the  syllables. For better results we can write down the sceleton with all its components. Secondly, once in several days we check how well we remember the translation, looking at the 'sceleton'. The recognised fkeshcards go to the horizontal part of the pile. And the third action remains the same. As soon as almost all fleshcards appeared in the horizontal pile we check the recognition of the words in characters for translation and oral reproduction. 

[Pall]

It's going on very well. Nevertheless, I decided not to learn all HSK(1-6) words, 5,000, in this way. I'll limit it by the parts 1-4 only, based on leading features. Moreover, I'll split the current Part 4 into two parts, limiting the part 4 by only 5 leading features. Then the parts 1-4 will make up somewhat more than a half of the whole list, about 2,700 words. The rest of the part 4 will be "Other leading features", covering 7 of them. So, the present part 5 will be renamed to part 6, and the part 6 to part 7. I hope to finish learning the set by the end of the year (in six months totally).
Why do I pay much attention to this division? Because the sections, relating to the parts 1-4, represent complete series. It means that each section includes all words (within HSK), one of the characters of which has this leading feature. If one of the characters of a word represents one leading feature, and the other represents another leading feature, the word is duplicated it the list and one more similar fleshcard is added to the other section pile. In the list the duplicated lines are marked in red, in the fleshcards there is indication, too. It's important to be aware, that all the words of a certain series must be known, it provides more confidence. The words consisting of 3 or 4 characters may be duplicated even 3 or 4 times. And it's worth to do since they are more difficult to remember.
As to the rest of the list, I'll make myself sure only that I know the words up to HSK 3 in advance. The words are not duplicated there, each next section doesn't include words listed in the previous sections.
What next? As soon as the parts 1-4 are learnt, I'm going to start reading graded texts up to HSK 6. I'll be retyping them. The words, which are from the rest of the list, firstly will be in Russian, further in the 'sceleton' view and finally in characters. I hope it'll be better way to learn them. 
The only inconvenience is that I cannot type all the leading features and other indications, and even if I can, there'll be space typed between them and the syllable name in mupin. So, I'll have to type only mupin transcription with the help of the macros function of the MS Word and then fill in indications by hand.
Looking in the future beyond the HSK, I suppose, the learning will retain such mode. New words will be added to the list to proper places. 

[Pall]

There can be a compromise solution. The thing is, that it's easier to learn the words, one of chacracters of which has a distinct "leading feature" on the brims of the character, by whole sections and complete series, but it's not the fact that it concerns the words from the parts 5 and 6, in which only some radicals that can occure in any place or even some elements of characters are used for their identification. Then the procedure may be as this: the words from parts 1-4 are learnt by sections, while the words from the parts 5-6 are memorized according to the HSK grade. As soon as the HSK4 words from the last two parts are memorized, it's reasonable to begin to read graded texts and other materials up to HSK4, as soon as the HSK5 words from the parts 5-6 are learnt, one can proceed with HSK5 graded texts. So, only some 3,000 words of the parts 1-4 should to be learnt in advance in the way discribed above (with 12-13 distinct leading features in the part 4). At my current rate of learning I mat approach the acomplishment of this initial task by the year end. By the moment I have learnt abot 1,500 words very confidently and some 700 words not so confidently. Although, I havn't moved yet all the words with the distinct leading features to their proper positions in the parts 1-4. It might happen, that the parts 1-4 will include some 3,500 words finally.

[Pall]

And the last, but not the least. After a series is learnt for recognition with a good result (90%), one more set of fleshcards should be made, which is designed for memorising the sceleton. Yes, it was the beginning of the process described above that firstly fleshcards were reviewed all the time for that untill their sceleton could be at last recognised at a check after several days. But now it's not for the final purpose of recognition of the words in characters, but almost the opposite. Of course, we could use the initial set of the fleshcards for that, but (1) it's not good to mix up these two purposes, and (2) the additional set contains two times more cards. While learning the sceleton we need to vary the translation, at which we look on the face side, because otherwise it's percepted mechanically, and we remember the words in our language, to which a sceleton corresponds, but not their meaning. So, if it's written on a card that to the translation 'independent; independence; to stand alone' the sceleton for 独立 corresponds, we dublicate this cards with the same translation in the opposite order: 'to stand alone; independence; independent'. On the reverse on both fleshcards there is the sceleton at the top and the word in characters at the bottom, but the last is only for reference. We learn only the sceleton and don't even look at the characters, otherwise the varied outline of them would overload our mind. We know how they look well enough at this stage, it's not necessary to overload our memory once more. The sceleton should be written down, not only reproduced in sound. Learning the sceleton is for those who (1) want better results with recognition of words in characters, and (2) want to be able to use these words in their speach and writing (typing). Thus, if we know the sceleton, we are able to perform three language activities: to read, to write (to type) and to speak, hearing function would benefit from that, too.
 

[Pall]

And yet, I think that when it comes to studying parts 5 and 6, even if the words are taught by levels of HSK, at the first opportunity the fleshcards should be distributed into packs according to the selected 18 radicals in part 5 and five groups of the elements in part 6. In any case, this will allow better to learn and remember the words.

[Pall]

As to learning the sceleton in the very beginning of the process and in the very end of it, it also consists in this. At the very beginning we look first at the translation on the face side, screening with a finger or something the word in characters, then write down the sceleton or say the transcription, imagining the indicators, and finally cast a glance at the word in characters. When it's done in the very end, we don't look at the characters at all, exept cases, when you feel you have no idea how it looks even approximately. This operation doesn't take much time because the words are remembered very well, but it fixes the knowledge perfectly.

[Pall]

I'd like to summerise it and make more clear.
First. Instead of learning to recognise words in characters directly we do it (1) at the more generalised level of the 'sceleton', at which distinct but simple indications of characters are used in combination with transcription, what saves capacity of our mind, and (2) with the help of 'roundabout' processes. Even the sceleton is not learnt to be recognised directly. Firstly during several days we just learn to remember what sceleton corresponds to a certain translation, recognised fleshcards aren't put aside for the next day, they're reviewed again and again. In several days we check if we can recognise sceleton. If not, the unrecognised fleshcards again aren't learnt directly for that, they go to the initial pack. Some fleshcards may keep at this stage longer than others, at the same time new fleshcards are added. However, if we noticed that some fleshcards aren't learnt too long, we can dublicate them, writing the translation in the opposite order. Usually that is enough to learn them in a day or two. When sceleton is recognised, the fleshcards go to the horizontal pack. At this operation dublicating fleshcards are removed and thrown to the bin.
Second. When almost all the fleshcards of a section are in the horizontal pack we check how well we can recognise the words in characters. As a rule, the result is 80-90%. Then we just learn the remainder in the traditional direct way. 
Third. Making one more pack of fleshcards of a bit different apperance for each learnt section, dublicating each card, is optional. If someone wants to master only reading, he can skip that. However, it's worth to do for him, too, because reviewing even two times more cards in the direction from translation to the sceleton, even if it's to be written down, is a way quicker than reviewing cards of the initial pack for recognition of the words in characters, and the effect is almost the same. If you have a day without much duty normally you can review in this way up to 500 words (1,000 fleshcards). I doubt that it's possible if reviewing recognition of words in the traditional way because of very big load on our brain's capacity to analyze diversed character outline. So, despite the opposite direction of reviewing and it's partial nature (only for the sceleton) it results in better remembering the words due to more frequent reviewing.
 

[Pall]

I'm still making corrections   When almost all fleshcards have moved to the horizontal pack, and we're checking how well we can recognise the words in characters and some cards cannot be recognised properly, we should not learn them in the traditional way at that stage. It's much better to put them back to the vertical pile and to process with the cards remaining there again (maybe adding dublicating fleshcards). Only when it's come to the situation when only few cards of a series cannot be recognised, we switch to the traditional learning of the remainder. And later, when we review the old packs in a month or two we do the same regardless the number of unrecognised cards. 
At the same time, if the proposed additional twice as big pack is made for the same series, which is reviewed more frequently (in the direction from translation to sceleton with writing down the latter), and some fleshcards cannot be recognised, we should make additional cards of the first type for the same words with the opposite order of the translation and add them to the vertical pack of the initial pile. 
OK, I don't bother you any more till the very year end, and we'll see what results will be actually in the half a year term.

[Pall]

There can be a better strategy. Firstly I came to a conclusion that learning to recognize words in characters should be only after one can recognise words' sceleton. I started to do this. Initially words of a section are learnt to recognize their sceleton, learnt words going to a vertival pack behind the initial horizontal one. Then the words in the vertical pack are learnt to be recognized being written in characters, recognized words going to one more horizontal pack behind the two. However, later I thought that it would be much more reasonable first of all to learn the whole HSK list in the first way, to be able to recognize the sceleton, and only after that to begin to learn words in characters on a large scale, combining it with reading. There are such reasons to do so: (1) learning words in mass  produces an effect rezembling the effect of 'big data', you can feel the language regularity, its patterns, (2) if words are learnt to the stage of sceleton recognition, it's very likly that you'll be able to recognise most of them in a text in characters, (3) reading provides more certainty regarding word use. Also, if the parallel set of fleshcards is learnt ment to remember the sceleton as one see the translation, words could be learnt not only through reading and retyping, but also through composing some texts like dairy notes, etc.