Outlier Semantic Components Poster: Review

[Tianjin42]

Recently, OneEye offered some members of the forum his company's new Semantic Components Poster PDF. Here, I will discuss the posters.

 

 

Above is a picture of the posters on my wall. 

 

General

The posters are A1 size (594mm x 841mm). In total there are 100 components. I had no problems with the files and printed them off at the university print shop. The printing cost more than I was anticipating - £9 per page. Perhaps I have been spoilt by the cheap print shops back over in Beijing. 

 

Aesthetic 

The posters themselves are aesthetically beautiful. The characters are smart and stylish, and the overall layout is easy to use and well-organised. I won't go into the precise details of everything that is included in each cell but each contains everything you need to know (such as stroke order, sound, traditional equivalent, variant and more) as well as occasional notes in red offering additional information where required.

 

It could be argued that the cells are slightly busy; at a short glance it may not be clear at the edges of the cell where the border is between the neighboring cells and the examples of characters using that particular component, but we really are picking bones in eggs here. 

 

I actually feel that, from an aesthetic point of view, the posters resemble something of a periodic table of elements for Chinese learners. In this sense I think the posters work well as an elegant reminder and would be perfect in a learning environment. 

 

Applications 

I think it is important to understand that the posters are part of the system that the dictionary will expand upon, rather than a solution in themselves. They would be useful for newer learners but imagine referencing the information will be quicker on the dictionary once it is released. 

 

I gather that there already has been talk of a textbook and my own feeling is that the method (looking at the form, meaning and sound in this way) suggests that it could be used to eventually build something resembling a reformed logical Heisig method (where characters would be divided up and learnt according to the reasons the components are included - be it meaning, corruption etc). I realise that this in itself would be a massive project but I think that for new learners, it could offer a really interesting new roadmap. 

 

In terms of the posters themselves, issues like character corruption can't really be included which is why it is best to see them as part of the whole solution. Looking at some of the additional information in red has certainly triggered further interest from me and I intend to upgrade from the basic dictionary to the more advanced one to get a clear sense of where we believe each character/ component came from according to recent research. 

 

Overall, it is a very smart piece of work and I feel it works best as a beautiful reminder on the wall of what you are learning. If the dictionary follows as expected then for most learners the posters will also serve as a first step to a more nuanced understanding of how the components of characters work. 

[OneEye]

Really useful feedback, thanks. I love the "periodic table" comparison!

 

We do plan on releasing Sound Component Posters as well, but it will be after the dictionary is released. Corrupted/Empty Components would be tough to cover on a poster, but the dictionary will cover those pretty thoroughly.

 

We're hoping someone will work out a mnemonic system using our character explanations—mnemonics rooted in a sound understanding of character structure would be more powerful than either approach alone. I'm not sure if we'll have room to cover that in-depth in the workbook we're doing with John Pasden and Olle Linge (though I'm sure we'll touch on it), but maybe in a future product.

 

Thanks again for the review!

[li3wei1]

As promised my review of the Outlier components posters. I'm working with the Simplified version.

 

First, it's a good idea to get this information laid out in this way: poster-sized, visually appealing, clear and easy to use. I've seen similar lists in the backs of books, or spread out over the course of a book, or online, but this would make a good reference for the wall of a classroom or study. I think the choice of which information to include and which not to include is good. I would agree that some thin horizontal lines separating the cells would be good. Also, putting both the 'variant' and 'traditional' items in the lower right corners can be confusing, and means you have to put in those v's and t's to label them. Maybe you could find a different place for one of them.

 

Quibbles:

The order in which the components appear is not obvious. It's not stroke number, or pronunciation, nor does it seem to be thematic. Different forms of the 'same' component, such as 水and its three-dot version, which would be treated as one in a list of radicals, are listed next to each other, which is good, but similar-looking components that could easily be confused, such as 月and月 (moon and meat), are not. 行 and 彳 are next to each other, but 厂，广，and 疒 are not, and they appear in order of decreasing stroke number. So if you're looking for a particular component, you just have to scan the whole chart.

 

While I can see the justification for including the two versions of 水as different components, it is harder to see that justification for some of the other pairs. I count 10 such pairs, including 人 and its 'squeezed' version, the 单人旁. The 'squeezed' versions of 土，火，女 and others all appear as 'abbreviated forms', why not 人? Somewhere between these two ends of the spectrum, you have 示/礻，衣/衤，near the look-alike end, and 火/灬 and 犬/犭at the look-different end. You have to draw the line somewhere, and I'm not sure where I would draw it. If you are aiming at learners who aren't aware of these connections, then including all variants as different components might make sense. On the other hand, as they do become aware of these connections, the chart will appear cluttered and redundant, and with some consolidation, you could have fit so many more components onto it.

 

A similar inconsistency can be seen in 匕，which has two meanings and two pronunciations jammed into one cell, while 月has two cells. There may be a reason for this, connected to traditional characters, perhaps, but it would not be apparent to the learner without considerable explanation.

 

Is there any reason why 土and女appear as characters with abbreviated forms, but 米and 足appear as components with full forms?

 

I note that 攵 is included, but not 夂. There may be some academic argument that 夂is not a true component or something, but to me, this is the sort of confusion that posters like this should be sorting out. At least a note in the section on 攵 would be useful, as you do this with 日 (listed) and 曰 (not).

 

Finally, what you have as the 'full form' of 工 should actually be in grey, because it's the abbreviated form.

 

I must admit that for all my criticism, I don't know that I would have made any better decisions. It's a messy language, and any attempt to force it into an orderly and systematic structure is going to run into difficulties. Some of my objections may come from the fact that I was raised on the traditional set of radicals, and I'm having trouble wrapping my head around the new paradigm. As paper dictionaries go the way of the typewriter, so will radicals, and maybe functional components will make more sense to those who will start learning Chinese in 2020.

[Gharial]

So it's essentially a chart of about half the items otherwise known as radicals, but presented in a quite jumbled order regarding number and type of stroke? How will the actual (paper) dictionary be indexed? Presumably it's going to provide a conventional radical-based index as at least one of the look-up options (like Harbaugh say provides a traditional Kangxi-based one in addition to his genealogical tree system - not that the Kangxi system isn't itself a bit of a jumble, and would obviously be ill-suited to simplified characters!)? If so, then it seems a missed opportunity to not present what's on the posters in an ordering similar to just such an index, as it will help people navigate between works and appreciate things more.

 

Sorry for the questions but I find it frustrating when reference resources appear to lack any organizing principle. It reduces the poster to little more than something to simply brighten up a classroom wall and look pretty. I know your aim isn't to produce a conventional dictionary, but even so, I wouldn't see the harm in at least the more conventional aspects of it being organized in, well, more conventional, familiar, and tighter ways. But hey, it's only a by-product for or sideshow to the main event, and done and dusted now eh, so perhaps it's no longer that important!

[Shelley]

So I took the DIY route and printed out my poster (simplified version) on my standard A4 printer. I chose print as large publication and it printed it out on 9 A4 sheets per poster.

Then I got busy with my scissors, glue and tape and got down to some old fashion cut and paste   The end result is not bad, it is usable, not the prettiest thing but ok. I might do the first part again as I learnt from some of my mistakes and part 2 is much better.

 

My attempts at printing it out aside, the content is fabulous. I can also view it in its original PDF form on my pc and that is very useful for studying.

 

I do not feel qualified to criticise how correct or not it is, but I like it. Having read a lot of the Outlier posts and some of the detailed blogs and investigated the website, I am prepared to accept it as a valid and useful tool for learning chinese.

 

I have not had time yet to study it in depth, but am inclined to agree with others who have likened it to a periodic table of chinese characters.

 

I like the layout and the subtle use of colour, it could have all too easily been overdone, garish and confusing. If this is an example of the style of the forthcoming dictionary, I think it will be very usable.

 

I am sure this will be a formidable tool in my box of chinese learning materials. I intend to check out the blog post and the video series as recommended in the poster guide.

 

As someone who has always believed in an etymological method of learning characters and as an anti "Remembering Chinese Characters by Heisig" person, this fits into my learning style beautifully.

 

Here are some photos so you can see my cut and paste attempt.

[lechuan]

 

 

As someone who has always believed in an etymological method of learning characters and as an anti "Remembering Chinese Characters by Heisig" person, this fits into my learning style beautifully.

 

@Shelley, curious to hear more why you don't like Heisig. All of Heisig's primary keywords are based on real meanings (the ones I checked on the chart have the same keyword meaning as you'd find in Heisig).

[laowhiner]

I'm currently going through Heisig (up to 1328 as of today), so I have a few thoughts on why someone might not like the method. 

 

There are times when he breaks up characters into their wrong components or makes up a definition for a component  that doesn't align with the accepted etymology or definition. That's why I reference wiktionary and several other sites that purport to have character etymologies; if I can learn the actual meaning of a component, why wouldn't I? Thus, I can understand why someone who is interested in getting the etymology right would be annoyed with Heisig.

 

While I think the etymology is cool and would rather know the actual meaning behind each component, I am OK being a bit sloppy because I don't think the etymologies will ever play a significant role in my life in my future learning -- long as I "know" the characters to my standard, I am happy. For instance, I've seen a few posts from Outlier showing how characters become corrupted (I believe this when a component comes to resemble another one even though they have different origins). This is pretty cool stuff... but I am happy to pretend these two identical looking components are actually the same one and then use it for character building. Others will want to be more accurate for a variety of reasons. 

 

That said, to repeat myself, if you could learn the characters in an order like Heisig but ALSO learn the etymologies and proper definitions for each character component, there would be no reason not to. 

[OneEye]

Before I respond to the comments here, let me ask: would people be interested in having this data in Pleco? Sort of a mini-dictionary of semantic components as both a primer for the dictionary itself and as a convenient way to index/study/make flash cards out of the poster content?

 

 

@li3wei1,

 

 

The order in which the components appear is not obvious. It's not stroke number, or pronunciation, nor does it seem to be thematic. Different forms of the 'same' component, such as 水and its three-dot version, which would be treated as one in a list of radicals, are listed next to each other, which is good, but similar-looking components that could easily be confused, such as 月and月 (moon and meat), are not. 行 and 彳 are next to each other, but 厂，广，and 疒 are not, and they appear in order of decreasing stroke number. So if you're looking for a particular component, you just have to scan the whole chart.

 

While I can see the justification for including the two versions of 水as different components, it is harder to see that justification for some of the other pairs. I count 10 such pairs, including 人 and its 'squeezed' version, the 单人旁. The 'squeezed' versions of 土，火，女 and others all appear as 'abbreviated forms', why not 人? Somewhere between these two ends of the spectrum, you have 示/礻，衣/衤，near the look-alike end, and 火/灬 and 犬/犭at the look-different end. You have to draw the line somewhere, and I'm not sure where I would draw it. If you are aiming at learners who aren't aware of these connections, then including all variants as different components might make sense. On the other hand, as they do become aware of these connections, the chart will appear cluttered and redundant, and with some consolidation, you could have fit so many more components onto it.

 

A similar inconsistency can be seen in 匕，which has two meanings and two pronunciations jammed into one cell, while 月has two cells. There may be a reason for this, connected to traditional characters, perhaps, but it would not be apparent to the learner without considerable explanation.

 

Is there any reason why 土and女appear as characters with abbreviated forms, but 米and 足appear as components with full forms?

 

I note that 攵 is included, but not 夂. There may be some academic argument that 夂is not a true component or something, but to me, this is the sort of confusion that posters like this should be sorting out. At least a note in the section on 攵 would be useful, as you do this with 日 (listed) and 曰 (not).

 

 

Here’s our criteria for deciding whether to split variants into separate entries or combine them into one. If the component/variant is used as a semantic component in 6 or more characters out of the top 3500 most common characters, then we made a separate entry. If not, we combined them. That should give you an idea of how common the variant is. That’s also why, for example, 匕 has one entry and 月 has two—not enough examples/not common enough (note: on the traditional poster and in the ROC standard, “meat” and “moon” are actually written differently, so they don’t appear to be two frames containing the same component).

 

For combined frames (like 米, 工, 竹, etc.), we chose the version of the component that shows up more frequently. If it’s the full form but there’s also a very common abbreviation, then we put the abbreviation in gray in the top corner. If the more common form is the abbreviation, then we put the full form in red in the top corner.

 

Hopefully that will clear things up. This is really useful feedback, and lets us know what sorts of explanations people need to accompany the product.

 

As for 夂, there just weren’t enough characters using it as a semantic component. Perhaps a note would be a good idea though (something for v 1.1).

 

 

Finally, what you have as the 'full form' of 工 should actually be in grey, because it's the abbreviated form.

 

 

Thanks for pointing that out. We’ll fix it in v 1.1.

 

 

@Gharial,

 

 

So it's essentially a chart of about half the items otherwise known as radicals, but presented in a quite jumbled order regarding number and type of stroke?

 

 

Well, no. We don't really talk about radicals at all because they're useless for discussing character etymology. They're simply a way of indexing characters in a dictionary. More here.

 

 

How will the actual (paper) dictionary be indexed? Presumably it's going to provide a conventional radical-based index as at least one of the look-up options (like Harbaugh say provides a traditional Kangxi-based one in addition to his genealogical tree system - not that the Kangxi system isn't itself a bit of a jumble, and would obviously be ill-suited to simplified characters!)? If so, then it seems a missed opportunity to not present what's on the posters in an ordering similar to just such an index, as it will help people navigate between works and appreciate things more.

 

 

We don't have any official plans to make a print dictionary. It's a Pleco dictionary. We'd like to do a print dictionary eventually, and we're looking into it, but it will likely be arranged by pinyin. Sure, there will be a radical index, but I doubt the dictionary itself will be arranged by the (unwieldy) traditional radical system.

 

 

Sorry for the questions but I find it frustrating when what ostensibly are reference resources appear to lack any organizing principle. It reduces the poster to little more than something to simply brighten up a classroom wall and look pretty. I know your aim isn't to produce a conventional dictionary, but even so, I wouldn't see the harm in at least the more conventional aspects of it being organized in, well, more conventional, familiar, and tighter ways. But hey, it's only a by-product for or sideshow to the main event, and done and dusted now eh, so perhaps it's no longer that important!

 

 

If our dictionary were already finished, then yes, we'd be able to organize the poster to fit in with the dictionary somehow.

 

As it stands, any order we could come up with was problematic. We plan on working out a better order in the future (these are v 1.0, and we'll be updating them), and we're actually working with some academics on developing a sort of "ideal learning order" for both the characters and the components, but that's going to take a lot of time, and a completed database and dictionary. It's simply not feasible at this stage—maybe toward the end of this year if all goes well.

 

They're all very common components though. Components 1-50 are more common than 51-100, which is why some of the components in 51-100 have fewer example characters—there weren't enough to fill the space. But they're all worth learning because they're all pretty frequent. Also, it serves as a handy reference—if the example character

 

One of the main purposes of releasing the posters early is to get people started on our system of learning characters before the dictionary comes out. It will also serve as a complement to the dictionary when it comes out.

 

I'm going to compile a FAQ from these questions and others. Really valuable feedback, thanks everyone!

[Gharial]

Oh, sorry, I was under the impression that a printed version had been planned first and foremost or at least simultaneously, but that may've been a while before Pleco's involvement was announced or made clear. Hard to remember LOL.

 

Radicals may well be "completely useless" for talking about character etymology, but 98% of the forms on the posters are commonly used as radicals*, and I doubt there would be a simpler, better or clearer way of arranging them here and for the current purposes than stroke count and recursive stroke order. I'd decide on a radical indexing system pronto then (why not simply use Pleco's more or less as is?), and refer the poster user on to that for the "why"** of that particular ordering, problem solved. Holding out for an ideal learning order won't help people become better at looking things up, and they'll be needing to refer to the supplementary workbook or whatever it was you mentioned for that ideal learning order anyway, right? Then, think of the excitable but keen and reasonably intelligent~"independent" learner who asks 'Hey, aren't most of these (used as) radicals too?', do you a) simply answer or footnote 'Yes!' and direct them onwards as before, or do you b) look pained, mutter something about 'all this talk of radicals' being simplistic and backwards and retrograde, and henceforth direct them to a somewhat long-winded blogpost, which they'll then be tested on?

 

I wrote similarly detailed feedback to li3wei1's here, for a site called IIRC Chinesehacks when they produced a simplified radicals poster v4 or thereabouts, but now they've switched to Disqus commenting my and other's comments there have disappeared. Shame as it might've still been useful to direct people to.

 

Personally I can't see the point in a primer minidictionary of your poster's content, just incorporate that (somewhat scant) info into the actual dictionary's entries and if need be (and I think there's a need) into a supplementary guide, like that provided in the ABC dictionaries, to the radicals as used in whatever eventual index. (As I have only the free version of Pleco I can't be sure how much of this type of info is made available in the paid version, if it's already included it will obviously save some work, at least as far as the use as radicals and their ordering is concerned).

 

 

*Regarding the link you posted, which I'd read before anyhow LOL, I think most people can grasp soon enough the distinction between use as a radical and use as a non-radical (if only because in the latter case, another radical will be doing the radical work), though it's perfectly excusable to talk of at least the form if not the meaning of a radical when decomposing characters, learning stroke order etc etc.'First it's the donkey's brow radical, then a wormwood stump component, and finally a rat's ass'.

 

**See from the words 'supplementary guide' onwards in the 4th paragraph above.

[Shelley]

Before I respond to the comments here, let me ask: would people be interested in having this data in Pleco? Sort of a mini-dictionary of semantic components as both a primer for the dictionary itself and as a convenient way to index/study/make flash cards out of the poster content?

 

Absolutely, I was going to put the PDF of the posters on my tablet to be able to reference them in just that way so, yes I think it would be very useful.

 

@lechuan I don't disagree with his choice of primary keywords (I haven't had time to cross reference them, but I will take your word for it) it's his little stories to aid remembering them that really drive me nuts. Some of them seem far fetched and tenuous at best and some have what I consider unnecessary references to biblical events and other unrelated things. The reason I don't like biblical references is that when these characters came into being, the bible hadn't been written yet, so it really annoys me to have this sort of thing forced on to the chinese language.

 

I have always been resistant to this style of learning, I feel as if I will spend my time learning these stories when i could just learn this character (A) means this (XXX). learning the component parts is my way of reminding myself. I can look at a character and say ah there is the foot component, this must have something to do with walking (simplified example) and use this type of thing as my aid de memoir.

Yes there are corrupt components and like a lot of rules and systems there are always exceptions, but you learn to live with these.

 

I am studying simplified characters because I believe this will be more useful to me, but I will study the traditional version of a character when I first encounter a new character that has both, because I think some of the etymology has been lost during simplification and familiarising myself with the traditional is one more aid.

 

Some characters lend themselves very well to a "story" but that is just its meaning, things like 家, pig under a roof = house,family makes sense when you know most chinese families kept a pig under the roof of the house. Of course after some time this aid is not needed and it is just house, family.

 

I appeared to have gone on a bit, sorry   but I wanted to get my point across clearly.

 

 

 

[Shelley]

If these are v1 as you mentioned in your post #8 then maybe you could have a look at reorganising the page breaks, when I printed it out a 9 sheets of A4 there was only a very few places where the text flowed over the page on to the next and had to be very carefully aligned.

 

With a bit of fore thought these could be avoided making a DIY printing job much easier.

 

Just a thought

[li3wei1]

I think many of the decisions you've made are arbitrary, and unfortunately there are no better ways to make them. You seem to be ordering them at least partly by frequency, i.e. the 1-50 ones are more commonly encountered than the 51-100 ones, but this is interrupted by the pairing of similar ones. Also, is frequency measured by the number of characters in the top 3500 that use them, or the frequency of those characters? So, if two components A and B appear in six characters each, but A's characters are in the top 100 most frequent, and B's are mostly past 2000, A would be a more frequently encountered component. In any case, frequency isn't going to help the student find the component on the chart. For that, I think you need either stroke number or some sort of grouping by visual similarity (for instance, there are some that are essentially rectangular, or vertical-horizontal: 口，土，田，山， and some that involve sloping strokes in both directions: 人，女，犬，示，欠）.

 

I think it would be helpful to find a better way of grouping together sets, i.e. stand-alone character, smushed component form, and possibly alternate versions i.e. grouping together 火with 灬 and 言 with 讠 (pardon the underlining, I don't know how to type parts of characters). I think for a learner, learning that one meaning (fire) is associated with several different versions of essentially the same pictograph, is no harder than learning that sometimes this component appears on the left, sometimes on the bottom, and sometimes in the lower right. And sometimes it appears twice. What is the advantage in learning it as two (or more) separate components? One problem with this approach is that it runs into problems with the above ordering principles, i.e variant forms don't necessarily have the same number of strokes, or might slip into different families of similar-looking shapes. 

 

Traditional radical tables have these problems, and in thousands of years no one has found a way around them. I wish, when they thought of simplifying the characters, they had tackled this, rather than just create a bunch of additional variant forms. But no one asks for my advice . . .

[Gharial]

Pleco uses what is essentially the ABC dictionaries' Comprehensive Radical Chart, that is, the Kangxi radicals and their simplified equivalents all in not only stroke-count but also zha2 札-type stroke order, and shorn of any redirecting numbers clutter (as the app will do that work for you). All very logical, easy and useful. I'd've thought it would be a simple and straightforward matter to adopt (adapt?) that ordering for a poster for a dictionary destined for the Pleco stables, but obviously not! (If one is insisting they shouldn't be called or even referred on to as radicals in case it misleads or "isn't as useful" for anyone). I'd add an eyeroll emoticon but that might be overdoing it a tad.

[OneEye]

@Gharial,

 

We made a very deliberate choice not to arrange them according to any radical list. Had we done so, it wouldn't matter how many times we say "It's not a list of radicals," people would still think it was (speaking from experience). We're not arranging it by radical order, and that's not a choice we're going to back down from. If people want a poster about radicals, there are several out there on the market to choose from, but this isn't one of them.

 

If you have a suggestion for ordering the components in a more useful way that will not reinforce the notion of radicals in people's minds, we'd love to hear about it.

 

But really, the ordering isn't very important from a learning perspective. If a student doesn't know all of the components on these posters, then they should learn them all. Changing the order isn't going to make a significant impact on their learning because this is all very basic stuff.

 

 

@li3wei1,

 

 

 

Also, is frequency measured by the number of characters in the top 3500 that use them, or the frequency of those characters?

 

For now, just the number of characters. In hindsight, we could have had a more rigorous way of analyzing this stuff had we waited until after our database was fully built (and thus, the dictionary finished too), not to mention the process of making the posters would have been much easier and less prone to error. But we promised the posters much sooner than the dictionary not realizing how much time it would take to do it this way.

 

Originally, the components were simply ordered according to frequency. Then I grouped them thematically: 土, 水, 阝/阜, 山, 艸, 日, 月, 木, 火 and so on are all "nature," for example, and they're in order of frequency within the category. But then it made sense to move 阝/邑 next to 阝/阜. And so on.

 

There's still largely a thematic arrangement, it's just interrupted, as you said.

 

S01-20 nature (except S04)

S21-37 people/the body

S38-45 tools, manmade things

S46-50 animals

 

So those are the most common 50 components. Some other components belong to those categories too, but they're less common so they're on the second poster.

 

Would it help to have the themes listed on the poster itself? Maybe at the bottom as part of the legend?

[Gharial]

If you have a suggestion for ordering the components in a more useful way that will not reinforce the notion of radicals in people's minds, we'd love to hear about it.

 

How about settling on an English keyword or phrase in caps for each and arranging the poster by them alphabetically. At least then they'd be easier to find and the user would be reinforcing the semantics in the process each time. Can't get much quicker and clearer than a single (and ostensibly non-radical!) sort that English speakers and learners are already familiar with.

[OneEye]

That's actually a really good idea. I think we might implement that in a future version. It might take a while before we have the time to rearrange everything, but it's a good idea. Or maybe if our next intern is better with Adobe Illustrator than we are and can do it more quickly, we'll implement it sooner. 

 

Thanks!

[Gharial]

Heh, you're welcome, OneEye! No worries if it isn't implemented though LOL.

[li3wei1]

Sorry I didn't spot the thematic grouping. If this is pursued, I would throw out the frequency, and instead go for sub-grouping, i.e. the nature group would have basic elements first (earth fire, air, water, wood, stone), then landforms, vegetation,agricultural products, and astronomical terms. Human body would start with 人,女, 子 and 儿 then start with the head and work down to the feet.

 

How many semantic components are there in total? Would you eventually want to get them all onto a set of posters?

[OneEye]

There are a few hundred. I don't have the exact number yet, sorry—that's something we won't know until we've finished doing all the research for the dictionary. We'd definitely be up for doing more posters if there's a demand for them, but there won't be nearly as many example characters once you get past the 100 most common.

 

We do plan to make some Sound Components Posters later too. Those will be really cool, but they'll have to wait until after the dictionary is released too.

[Gharial]

There'd just better not be 214 of them!!