Taiwanese tones and making Skritter more difficult

I’ve got a couple of issues with Skritter that, if resolved, would be really helpful for me.

  1. It’s killing my tones. I was unaware until maybe just a couple of years ago how many characters are pronounced with different tones in the mainland and here in Taiwan. I don’t know how many it comes to, but I’ve been working with a set of about 5500 characters, and I think there must easily be several hundred that are different according to region. Skritter doesn’t accommodate this difference, and doesn’t allow me to personally edit my lists to reflect the correct tones for me. I’ve been ignoring it up to now, but after about 7 moths of concerted work with Skritter, now that I’ve started using my Anki decks again, I find that my tones have really suffered. I suppose I can individually add a note to each card in the definition field to reflect what the tone should be for me here, but that only works if I happen to have the correct tone memorized already. Most of them, I don’t know, and wouldn’t recognize as they occur in my studies.

  2. The graphic stroke-by-stroke interface of Skritter is really short-cutting my memory. It’s way too easy to figure out what the next stroke should be, or if there’s another radical that needs to be included that I’d forgotten about because the placement of the other strokes shows lots of room at the side or at the bottom. Sometimes Skritter stops and tells me I’ve correctly completed the character when I’d intended to make another stroke or two. It’s a very helpful learning tool as it is, but there’s a definite ceiling imposed on its ability to help cement the characters fully into my memory because it provides so many crutches as I’m writing. If I had to draw an entire character first, and then submit it for analysis (like using the character lookup function in Pleco) and grading, that would force me to write it 100% correctly, with no help or feedback along the way. I realize that’s a major structural change for a program that’s still in the process of settling into a major update, but maybe it’s something to keep in mind for future versions.


A few suggestions:

  1. Regarding awareness of Taiwan pronunciations, I use the Zhongwen Chinese Popup Dictionary in Chrome and it does a good job pointing out pronunciation differences for Taiwan. Of course, you would need to know that there is a difference or else check every word as you come upon it, which might be a bit tedious.
  2. Regarding the stroke by stroke interface being “too easy”, are you using the raw squigs option? That may solve the problem.
  3. Maybe there is a way for you to take your 5,500 words and dump it into some translator to generate a list of definitions such as MDBG.net. From there you could do a search for every instance of “Taiwan pr.” Maybe you could use Excel or something to create a list of only those words where the Taiwan pronunciation is listed as a variant. I realize this does not do anything for Skritter, but at least it would give you a reference for doing the annotations you mentioned.
1 Like

Thanks Podster. Regarding point 1) yes, that’s pretty much the problem. I’d have to check each character individually as they came up, which makes an easy useful educational app into something unwieldy and tedious. 2) I’ve been meaning to look into what a “squig” is. That might solve the problem. 3) I hope it doesn’t come to that, but you may be right.

Thanks again.

happy to help. If you are using Skritter 2.0 web (beta) version, click to gear to access settings and check the box that says “raw squigs” to give it a try. I couldn’t find the setting in 1.0 but it used to be there. Also, all of the “more info” links lead to a blog post about the new Android beta. Whoops!

1 Like

One other comment about finding out that one has written a character correctly (surprise!) when one was planning to add a few more strokes. (“画蛇添足”)
Yes, Skritter will mark you as correct as soon as you finish the last necessary stroke, but if you intended to keep going you can always mark it wrong manually. I try to be disciplined about that, though not sure I am 100% honest when I am in a hurry.

This is what I do if it marks as complete when I was still going on iPhone - tap back then tap and drag the green tick to a red cross to mark wrong then forward to next word again - takes a second or so

I’m careful to mark my mistakes manually when I get credit for not really knowing something. Was just trying to come up with several examples of different ways the app allows me more wiggle room than I need. It’s actually helpful on the first pass to get the extra help like where the various strokes and portions of the character are located on the screen to job my memory, but after I get through this batch, I’ll probably go back through and try the raw squigs option.

There is also a “teaching mode” that you can select that will walk you through the strokes the first time a character is presented and basically let you trace the character while prompting you on the correct stroke order. You can try this and also experiment with having the raw squigs on or off. Remember, you can always hit the erase button and write the word again. I recommend doing this if you got scored as correct or almost correct but the character was particularly mangled looking. :slight_smile:

What I mean is, if you have a complex character with radicals or components in say 3 out of 4 quarters, let’s say you have a tall one on the left and 2 small ones on the right, like with the character 撼. After I’ve written the 咸 portion, there’s a clear space underneath where I’m expected to write something else, and actually I can see there will be empty space by the second stroke of 咸. That gives a hint that can’t be ignored, no matter how hard you try not to rely on it.

As far as stroke order goes, for the most part the biggest problem I have is with inconsistencies between the app’s preferred order and how I’ve learned it from other sources–or those characters where you can throw in practically any old order you like, and it still gets accepted.

Raq squigs completely takes care of this, where you would finish writing the character, and the prompt would still be expecting more strokes and not complete. At that point you would know the character wasn’t correct. Once you reveal the character it’s automatically marked wrong, so you could then reveal the character, and then erase what you’ve written to start over. With raw squigs disabled, you could still manually mark the character wrong once you’ve written the 咸 and see it’s not formatted to how you were expecting. You are in control!

The preferred (or correct) order has been constantly tweaked over the years if in need of adjustment. The chances for stroke order errors as time progresses becomes exponentially lower-- so while there still may be some stroke order errors here and there, chances our what Skritter has listed is correct. For traditional characters we abide by the Taiwan MoE. You can check on stroke order here: http://stroke-order.learningweb.moe.edu.tw/character.do. Regarding the characters you can throw in different orders, those are ones with stroke order variances built in to them and have more than one accepted stroke order. (You could use the teaching mode to view the preferred stroke order in this case).

I find characters with problems here and there on a nearly daily basis. As an example, you can look up just about anything using 婁. I don’t have much else to offer because I haven’t been taking extensive notes. I made a handful of observations in the Bugs forum, and I’ve probably found at least that many more since then. But as you mentioned there, nobody can fix it anyway right now, so I haven’t been making a point of noting more. I’ve checked with the website you linked for some of these characters to confirm errors, but the interface is unwieldy and slow, and doesn’t even work on every browser apparently.

Anyway, thanks for the feedback. I’ve tried raw squigs and it is mostly what I was asking for originally.

1 Like

I know this is a few months old, but anyway, I did a search on www.mdbg.net for “Taiwan pr” and got 319 hits

it shows the China/Chinese pronunciation with the Taiwan/Chinese pronunciation in the definition. Not just tones, but sometimes completely different pronunciation. Since we can’t edit the pronunciation on Skritter we’d need to create an Anki deck or something using the pronunciation provided in the definition and study it separately.

You need to use the advanced search and put Taiwan pr in the English exact phrase search box.

1 Like

On Taiwanese tonal differences: I’ve taken to just editing the definition and writing the pinyin for the different pronunciation in there. This also works as there are frequently differences in usage or meaning between the two, so you can add that in at the same time. It’s not ideal, because obviously you’re not actually studying that difference directly, but for the most part the differences are very obvious or follow a pattern.

e.g. majority of final neutral tones being replaced with the full tone.

Not a long-term solution, but good enough to keep moving at speed through your vocab!

1 Like

Russell, thanks for the tip. I used to spend lots of time at mdbg before stumbling across Pleco and eventually shifting the bulk of my char/vocab studies to phone apps. It’s crazy how many pronunciation discrepancies exist between the mainland and Taiwan, given the relatively short span of time in which they’ve been developing separately (although I guess it’s a lot more complicated than that, but anyway).

I do something like what Parkemoon suggests, I just put a tag at the top of the definition showing the alternate pronunciation when I find one. That’s one of my key gripes about the beta app, unless they’ve fixed it: last I knew, you couldn’t alter the definition field, and it’s essential that I be able to do that. Far too many default defs just say “place name” or “family name,” and generally if you dive into the dictionaries you can find a definition that may not be in current use but still helps give a cognitive handle to characters while also helping understand the nuances of meaning and associations they (might) lend to other compounds.

1 Like