Jump to content
Learn Chinese in China


Phonetics&radical based approach to learn characters, words and grammar

Recommended Posts


I was advised to clearly show what was included finally in the method I was talking about. In fact, over the past year since I started working in this direction, my approach has changed to a large extent. Even its title has to be revised. However, some basics remain the same. 


What was the reason making me learn characters ahead? When characters are learnt as components of words traditionally,  memorising 3,000-4,000 characters takes 4-6 years, as I guess. Characters are the main handicap in advancing in Chinese as quickly as it's possible in the majority of alphabet languages. At the same time, one has to know those 3,000-4,000 characters (or 10,000 words, as Imron mentioned) that he can read books and especially newspapers. And this is my goal, too.


There are two main problems in learning characters apart from words in advance. First, it's difficult because there is no context. Second, while one is learning characters, he does not advance in the language as to words, word use and grammar. I can say that the first problem is solved now. I'm approaching the meridian in learning 4,500 characters set that I took as the 'must'  language base. It would take me less time if I didn't have to develop the system. I think, it would be not more than 5-6 months under normal conditions. Although in my particular case it was as that: in September 2019 I new about 300 characters (250 words). The second problem is partly solved due to introduction of special semi-anchor writing.


However, it's not the main measure to judge about the efficiency of the approach. The most important question remains to be answered: to what extent does that accelerate learning the language as a whole? Well, it can be proved only if I manage to pass official exam for some advanced grade in reasonable time span. 



Now what the approach is in details. It was right desision to group characters according to their phonetics, since there are only 400 syllables without accounting for tones. Also, it helps to learn phonetics as well. Although, I did a mistake at the beginning. Characters are very diverse, and only one basis for their groupping is not enough whether it is phonetics, character structure, origin, or whatever else. There should be two contrasting bases itersecting each other. So, I came eventually to intersection of phonetics and radicals. 


I should probably present the process in stages as I see it now to be more concise. 


First stage. Phonetics (pinyin, pronunciation) and HSK1 (words and grammar, not only characters) are learnt as an elementary basis.


Second stage. For each syllable without accounting for tone a character is selected as a representative of the whole syllable, about 400 'anchore' characters. All other characters are to be learnt by using these characters as the key stone. To learn anchore characters I composed seven mnemo-poems according to 声母  vs    韵母 correspondence. So there are B-P-M-F, D-T-N-L, Z-C-S, ZH-CH-SH-R, J-Q-X-Y, syllables written with 'y' at the beginning not having any  声母  , but they're represented by the same  韵母   as those with J, Q or X in the beginning, G-K-H poems and one poem for all remaining syllables without  声母  . In the poems (short verses in fact, not poems) in each line the meaning of all syllables of the same   韵母  are used while  声母  go one after another by lines. For example, in the B-P-M-F poem the meaning of the 拔, 怕,马, 法  characters is used in the first line,      白,  拍 , 卖   in the second ,  班 , 盘 , 慢 , 饭  in the third, etc. They represent syllables ba, pa, ma, fa (bá, pà,  mǎ, fǎ) ,  bai, pai, mai (bái, pāi, mài - syllable fai doesn't exist), ban, pan, man, fan (bān, pán, màn, fàn), correspondingly.  Initially I did another mistake: I took as anchore characters mainly most used ones, but many of them happened to have several variants of pinyin, whereas anchore characters could best serve as anchores only if their sounding didn't vary, or did a little. So I have to recompose all poems now, the first two are ready. Of course, one can learn those 400 anchore characters by other means, however they must be learnt very very confidently, otherwise they woun't serve as anchors. Poems are written in poetry metric, if possible, with rhyme for certain lines, stanzas are of certain length. All that helps to remember them very well and to find quickly each meaning of anchore characters and recollect their pinyin. For that poems are accompanied by tables, where used anchore characters are put in ceratin cells according to their syllable and tone.


Characters have a number of meanings sometimes, but for these purposes only one, two as maximum, are selected to indentify the character, it also enables to memorise it better. However learning the meaning of characters is not only for their identification, it'll be helpful in learning words, too.


Third stage. The set of 4,500 characters is subdivided in the following parts: 1st - the most frequent radical characters of 'hand on the left', 'tree', 'water on the left' and 'mouth' (some 800 characters); 2nd -  characters of 11 not so frequent radicals, each represented by 70-150 characters (about 1,000), 3rd - characters of 27 not so rare radicals, represented by 30-70 characters each (totally more than 1,100), 4th -  ones of the most rare 140 radicals (totally about 1,200-1,300), 5th - unique characters and those with not very distinct radicals (only 150). It's necessary to learn firstly the 4th and the 5th parts (plus 'mouth' radical characters, perhaps), for learning three others will be in a bit different mode.


An Excel tabel is kept for all characters, where they are distributed according to syllables. It's inportant, that characters relating to the same tone-syllable are spread in space 3X3, making 'macrocharacter'. It helps to remember the characters better and to keep in mind their position in the macrocharacter, which will be needed for the next stage. There can be only seven characters in one 'macrocharacter', the seventh character becomes the first one in the next macrocharacter for the same tone-syllable. The position of characters in the macrocharacter has numbers from 1 to 7, pic. The first character, the 7th, 13th (not 14th, since the 7th one is the 1st for the next macrocharacter), 19th characters are called 'head' characters. It's desirable that head characters are also of only one pinyin variant, though it's not so important. In the Table learnt characters are to be marked by the learner when they are set for learning in their turn,  in complience with the part of the set they represent.


The Excel table is an auxilliary tool, characters are learnt with character cards of special design. On the face of them the character is, on the reverse there is a table 3x3, in which the character is shown in its position in the macrocharacter, also the head character is present in the first place, and in addition in the bottom right corner in different color the anchore character is shown with the tone of the target character before it, pic. There is a space below the table, where a phrase linking the meaning of the three characters must be present. For the top fleshcard the phrase is 'our pheasant was strangled' (sorry), for the bottom one it is 'he let me weed in spite of the cloud', the target words 'strangled' and 'weed' are underlined. The succession of the meanings of the target, head and anchore characters don't matter actually. So the target character is remembered in relation to the anchore character and also (less important) head character. The anchore character with the tone of the target character before it serves instead of pinyin. I came to a conclusion that association between two characters works better than accociation between a character and Latin transcription. In the top card 雉 is the 13th character for the syllable zhì , so it's considered the head character for the given target character, whereas 至 is the anchore character. For that reason anchore character is not shown in the bottom right corner, and only two vertical orange lines on the left reflects that.


The cards are learnt as this (everyone has its own secrets, however I'd like to present my procedure). A pile of fleshcards, which is to be learnt currently, can include up to 60-70 cards. Several times a day characters on them are repeted. Those, which one was able to recognise at the very first repetition (was able to remember both how the character sounds and its meaning) are put to a special file. During next repetitions recognised cards are just put aside. By the evening usually only few cards remain that cannot be recognised. But on the next day all those cards are mixed up again, some new cards are added, and only those cards are considered as learnt that are recognised at the first repetition. The cards that were put to the file are revised once in several days. Those, which were recognised, are put to another file, cards in which are revised once in two-three weeks, and also there is a file to revise cards once in two-three months (I keep several files of each kind according to the part of the character set, otherwise the number of cards would be too big for quick revision at a time). Cards from any file, which were not remembered, are returned to the very beginning of the learning line. i.e. to the pile for today's stydy.


If a character has two or more variants of pinyin, it is present in all corresponding parts of the Excel table and cards, so the principal unit is character-syllable, not character itself.


Fourth stage. The characters of the 3rd, 2nd and the 1st parts are learnt in radical series. It means all characters belonging to the same radicals are kept together in piles and repeted seperately from other characters. In my opinion, it'll help to remember them quicker and better, though I've not experienced it myself yet.


At the same time one can begin to learn words, word use and grammar by using learning materials retyped (or rewritten by hand) in semi-anchore mode. Anchore writing is as below (here old version of anchore characters is used, some could change now). Characters that happen to be either anchore characters or very frequent ones (in this version there is a dozen of such most used characters, but now I'm inclined to use all characters of HSK1 without substitution) appear as they are, whereas others are shown with the use of their anchore character with superscript index consisting of two digits placed before the character (because its necessary to see the right tone at once in order not to use the real tone of the anchor character). The first digit can be 1, 2, 3, 4, 0, and the symbol =. It means the tone of the target character. If its tone corresponds with the tone of the anchor character, the symbol '=' is used. The second digit means the place of the target character in the macrocharacter, which can be from 1 to 7, for the 7th character becomes the next head character. Also in the anchor writing spaces between words are made. 


Real text


The text in the anchor writing

21    11     
你好            哪儿
     2141刀了,21作天     晚上     32雾了?
21作天    晚上     2121喜了。
真的     吗?


Semi-anchor writing means that in the anchore mode only characters of the 3rd, 2nd and the 1st parts are shown, while all the others (those learnt already, HSK1 characters and anchore characters) appear without change. So one can learn  different words, in which one character or both being shown as anchore characters with superscript index, and how they are used in centences. At the same time he continues to learn characters of the 3rd, 2nd and 1st parts . It would be wonderful to have also a conversion program that would present in semi-anchore view any text. It would give an opportunity to read everything even knowing only the 4th and 5th part characters. 


I hope it's helpful to know. Now I'm converting word examples for HSK of different levels to semi-anchore view. As soon as I've finished some sufficient portion of it  I'll upload the examples together with the Excel table, list of anchore characters and lists of all 4,500 characters by the parts. 




Share this post

Link to post
Share on other sites
Site Sponsors:
Pleco for iPhone / Android iPhone & Android Chinese dictionary: camera & hand- writing input, flashcards, audio.
Study Chinese in Kunming 1-1 classes, qualified teachers and unique teaching methods in the Spring City.
Learn Chinese Characters Learn 2289 Chinese Characters in 90 Days with a Unique Flash Card System.
Hacking Chinese Tips and strategies for how to learn Chinese more efficiently
Popup Chinese Translator Understand Chinese inside any Windows application, website or PDF.
Chinese Grammar Wiki All Chinese grammar, organised by level, all in one place.


I forgot to say that it is excitingly interesting to see which characters in outline have what meaning, and from which characters those or other words are composed. I think such interest enhances the memory's ability to remember material.

Share this post

Link to post
Share on other sites

Yet I know what soft and services can be utilised to convert Chinese texts to semi-anchor view. Firstly it's necessary to make spaces between words, that can be done with the help of the Google translator. Although it does it only for pinyin, and it'll be needed to make spaces between characters manually. Also it's not convenient because of the limited volume of text that can be processed at a time in the translator window. At the next step the text with spaces can be processed with the WordPipe, an Australian made program, which is able to replace numerous letters and words for another ones according to the table prepared in advance.


Unfortunately the WordPipe would hardly change colors, which are also needed, see below.

Edited by Pall
additional considerations

Share this post

Link to post
Share on other sites

So from the beginning of the fourth stage one faces the situation when he has learnt characters of the most rare 140 radicals, unique and those characters, which radicals are not detected very well, and he has to learn characters of 41 very well determinable radicals ahead. Since then if he sees a character that doesn't belong to those 41 radicals (it's not difficult to keep in mind, which those radicals are, because radicals like 'a man on the left', 'grass' or 'metal' are easy to see) and that he cannot recognise, this means that he has learnt it most likely. He just looks it up in the Table and, if it's found there, he'll never forget it. If not, it must be a very rare character, which to be either added to the Table or used only for one time to understand the text. 

Share this post

Link to post
Share on other sites

It's not a problem to learn a pile of a hundred fleshcards at the same time. The more cards are learnt the easier new are remembered. So it'll be not too difficult to learn not only the third part, series of characters of not very rare radicals (30-70 characters in each), but also the second part with series of not so frequent radiclas. With the exeption of 'man on the left' and 'grass' these series contain 70-110 characters each only. All that makes up about 2,000 characters. As a result in the end of character learning only 'man on the left', 'grass', 'hand on the left', 'tree' and 'water on the left' radical characters will represent some challenge, but it's not more than 900 characters of the 4,500 most used set. 

Share this post

Link to post
Share on other sites

This mnemonic system you've devised is very unique, and I'm glad that you're sharing your experiences with it here.


Some thoughts:

  • Your goal of constructing these cards for 4000+ characters is commendable, but I would prioritize getting started on the long and vital process of learning words and building your actual vocabulary.
  • Consider migrating this system into an Anki deck. Electronic flashcards are easier to manage and aren't vulnerable to strong breezes.
  • I don't think converting texts to "anchor" characters is the best use of your time. CSL students sometimes use pinyin reading materials too long because it's easier than reading characters. I'd recommend using your system to help you avoid such crutches and read Chinese in its unaltered original form (no spaces, no superscript numbers, "anchor" character substitutions, etc.).
  • Like 1

Share this post

Link to post
Share on other sites
On 9/13/2020 at 2:18 AM, 大块头 said:
  • I would prioritize getting started on the long and vital process of learning words and building your actual vocabulary.
  • Consider migrating this system into an Anki deck. Electronic flashcards are easier to manage and aren't vulnerable to strong breezes.
  • I don't think converting texts to "anchor" characters is the best use of your time. CSL students sometimes use pinyin reading materials too long because it's easier than reading characters. I'd recommend using your system to help you avoid such crutches and read Chinese in its unaltered original form (no spaces, no superscript numbers, "anchor" character substitutions, etc.).

You're right, I should probably just learn the remaining 2,500-2,800 characters as quickly as possible. The learning has accelerated greatly now that  I 've begun to memorise by long series. I expect that learning that number would not take more than six months for me. After that I will switch for learning words. The principal idea is to learn a sufficient number of characters in advance so that I would face only few uknown characters in an average text.


However the idea of anchor writing isn't bad. Of course, retyping or rewriting texts manually in that view would be too time consuming, I agree. But if there was a program that would convert texts to semi-anchore mode it would be quite different. Then one could do the opposite: replacing anchore writing for real characters as soon as he has learnt some radical series. Just look at the beginning of the work that I was going to do. Below are words from HSK6 (I decided to start with HSK6 in order to fix my knowledge of as many characters as possible) with examples of word combinations and sentences. I subdivided the first, second and third parts of the 4,500 character set into seven small sets of 300-500 characters each and gave them distinguishable colors. This kind of writing is a way better than just pinyin. Those who have learnt the forth and fifth parts, characters of 140 most rare radicals and unique ones (plus HSK1 and anchor characters, of course) can read that. The words that consist of the learnt characters are shown in their natural view, in black. The words consisting of unknown characters can be read correctly as if they were written in pinyin, but the colored semi-anchor view contains some valuable information in addition. Orange color means that the radicals of the characters can be 'tree' or 'hand on the left' only, red - 'water on the left', 'man on the left' or 'grass', light blue - some of the eight radicals: 'car', 'sun', shell', 'heart at the bottom', 'stone', 'eye' or 'grains in the field'. And only dark blue means broader choice from twelve radicals (the broadest, but it could be subdivided in two at the expense of the introduction of  one more color). Moreover, some words can have only one character from the not learnt series. Here you can see it at the example of 治好 that is written as   
    =3. It's definately much better than just 'zhìhǎo‘ . Though this example is not very good to show the advantage since it's obvious what character must be instead of 'hǎo' .

As soon as one has learnt those series of dark blue, only about 530 characters in spite of the big number of radicals, he can replace anchor characters with indices for real characters. Further he does it for all colors one after the other. 



23     打;23     骂;23     时间

              那里    23    时间。
    23         了。


24 =6

45          24 =6          24 =6

         24 =6     了。
24 =6             =3好。



同学们                        爱不释手。





As to spaces I still believe that they are usuful for beginners like myself. Chinese write without spaces because they know where words begin and end, whereas beginners don't know. I understand that we should train ourselves to deal with real situations, but switching for text without spaces should  be postponed in order firstly to learn words and their use better.  And in the anchore writing they are particularly necessary.


Also I don't want to change paper cards for electronic ones since the Anki algorithm doesn't meet my needs. I learn by big piles of cards, sorting through the cards one by one. I described my procedure above. Anki doesn't allow this. And it'll be particularly inappropriate for learning by radical series when the whole set of cards of the same radical is kept together all the time. 




Share this post

Link to post
Share on other sites

I started learning the Part 3 and found that those characters can be learnt very quickly. It's the easiest part to learn, because in consists of characters of 26 not so rare radicals (not too many), each series including 30-70 characters (not many again). If using the 'serial' way of learning, the total number of those characters of about 1,000-1,100 can be learnt in a month. 'Serial' way of learming means that all fleshcards of characters representing the same radical are kept together. It allows to compare then while learning. Each day of learning at the very first repitition the cards, which are recognised, are not put to a file, they're just put aside. The cards recognised at later repitions on the same day are put aside as well. On the next day all that batch fleshcards are mixed up and sorted through again in the same manner. Fewer cards remain not recognised by the night. New series are introduced every day or two days. So one can be sorting through several series on the same day, the difference between the different radical series will be that those series that were introduced earlier will be sorted through quicker, only a few cards remaining unrecognised by the night. As soon as all crads of a series are recognised it's put to a file to be revised in a week or later. 


However this mode of learning can hardly be applied so efficiently to larger series of the parts 2 and 1. The characters of them woun't be easily compared to one another. So I decided to learn only the part 3 and then start to apply semi-anchore substitution, but it'll be at a sufficiently  higher level ('improved semi-anchore view'). The situation will be that about 2,700-2,800 characters are learnt and only some 1,700-1,800 remain, a little more than a third of the total set. But the most important improvement is that the remaining characters will be represented by only 15 radical series of 70-200 characters each. It makes it possible that each color correspond to one or two radicals only (three in one case), what improves greatly the effect of the semi-anchore mode. There are 10 distinct bright colors in MS Office, so 'tree', 'water', 'hand on the left', 'grass' 'man on the left' can be attributed by only a single colore each. Also I decided to do the same with the 'mouth' radical, but leaving only 'mouth on the left' characters, the remainder being transfered to the part 5 of unique and not so distinct radical characters (this is for others who might go this way, since I've learnt the 'mouth' radical characters in the very beginning).


Then it can be helpful to concentrate on dealing with HSK materials, up to HSK 6. The bulk of more used characters of the remaining series will be learnt earlier in words and in context. Words written with anchore characters with indices could be learnt in two steps: in anchore view and then in real view. As a result the long series of unlearnt characters of the parts 2 and 1 will become much shorter, and it'll be possible to apply to the remainders the serial way of learning, too. All that will provide the quickest way to learn 4,500 characters in advance with reliable memorisation, while working with HSK learning materials for words, word use and grammar. And after that it'll be possible to read almost any texts with much less effort to remember new words. 

Share this post

Link to post
Share on other sites

Join the conversation

You can post now and select your username and password later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Click here to reply. Select text to quote.

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

  • Create New...