Tone changes gives error in tone practice

mikelove

皇帝
Staff member
It's that I don't want to change this behavior twice, basically. It's a disruptive enough behavior change that I'd rather not change it now and then turn around in a little while and change it again. Not related to what the *correct* behavior is, but rather that there's a certain behavior people currently expect (which is also the way it's been behaving for more than a decade) and I don't want to suddenly change it in a minor update and then change it again when we revisit the problem more deliberately.

What you're describing as far as "what you hear is what you get" only works if you're doing tone practice as a listening test, which is not how everybody uses that feature. For someone who's reading tones on the screen and learning them, they're going to want the tones they saw on the screen to be marked as correct, not the tones they hear. We could of course simply accept both answers, but then people might be annoyed that an answer that didn't match what they heard was nevertheless marked correct because it matched the way the card was written.

So I totally agree that it would be better if it marked you correct on a listening test based on what you heard, but it's not a change we're likely to make in a minor bug-fix update, because it probably requires a new setting and a bunch of other changes and I'd rather not do it haphazardly now + then revisit it again later.
 

Aslaksen

秀才
I'm sorry, but I still don't understand why this is problematic. You can configure the tone test as either showing characters or not, and then you can determine what the correct answer is based on that choice. To me it makes perfect sense that 23 is the right answer if you only listen, and 33 is the correct answer if you see the characters. I don't see how anybody could be unhappy with that arrangement. The people who listen will be happy, and the people who look at the characters will just do what they are used to. I can't imagine that it would be hard to program that the correct answer depends on the type of the test. So you accept both answers, depending on the setting of your test, but of course not both answers in all cases.
 

mikelove

皇帝
Staff member
It's not obvious what the 'type of the test' is, though; what about people who have the characters displayed on the screen and the audio auto-played? Which one of those they're relying on is far from obvious; heck, some people might have their devices muted all the time and not even realize they've got the audio set to auto-play.

Again, I'm not saying we're not going to do this, I'm just explaining why we haven't done it yet - why it's something we're saving for our next major update and not addressing in a minor bug fix.
 

Aslaksen

秀才
I'm sorry, but I really do not understand what you mean. Why would anybody do a tone test with a muted device? If you do a tone test, you are presumably trying to test your ability to detect tones. I believe, but you might have data that shows otherwise, that most people would want to do that with the audio only option. However, I totally understand that some people might want to also see the characters. My suggestion is to set 23 as the right answer if the characters are not shown, and 33 as the right answer if characters are shown. My definition of type of test is whether characters are shown in the tone test or not. What is not obvious about that? I don't see why you consider that to be a major change, and not a minor bug fix.
 

mikelove

皇帝
Staff member
Even if I agreed with that rubric, now the app is suddenly changing its behavior in one situation and not another - somebody was previously getting a correct answer on 23 but then they change their test to start seeing characters but now they're not and they don't understand that that's the reason why. For somebody who didn't know this was coming it could feel like a bug. Doing this in a major update, in which it could potentially be accompanied by a new setting but at the very least when people would know it was happening + know to expect changes like this, seems like a safer approach.

I'm sorry if you disagree with that, but honestly, it's been working this way for more than a decade and I can count the number of complaints we've had about it on one hand; even just based on the coding effort (which is not massive but would probably be a good day or so to implement + test properly, since the code that automatically generates sandhi for audio is in a whole different part of the app) I'd be reluctant to devote the resources to it when we already have a fix in the works for our next major update.

For now, my best suggestion would be that you go into any cards affected, convert them to custom cards and manually edit the tones to match the tones you'd hear playing audio - that should get Pleco to accept your responses correctly.
 

Aslaksen

秀才
I'm sorry, but I find this thread incredibly frustrating. I honestly cannot tell if you understand the issue or not, like when you talk about people doing a tone test muted! Your new example does not make any sense to me either. If you first just listen, and then switch to listen and seeing, then how do you know that what you see and hear is the same as what you just heard earlier? Maybe the first one was a 23 word? Do you understand that? You do realize that I'm using random words, and not a list I have selected myself?

As for your suggestion, it again does not make me feel that you understand my issue. It again sounds to me like you believe that I'm tone testing a specific group of cards. But the whole point is that I do not want to know which cards I'm using, so I'm just using random cards. I may be wrong, but I believe that is the default setting for the tone test profile.

I've been a big fan of Pleco in the past, but I must say that this thread is really making me very frustrated and much less supportive of it.
 

Shun

状元
Hi Aslaksen,

I'll try to explain the problem in one or two sentences: Pleco is unable to tell from the recording you hear whether there's a tone change in it or not. That's why Pleco cannot ask it of you, because it doesn't know it. It's only a sound recording. I hear that it will, however, support it in a clean way with the next major release, by adding an extra field for tone sandhis.

Hope this helps,

Shun
 

Aslaksen

秀才
Hi Aslaksen,

I'll try to explain the problem in one or two sentences: Pleco is unable to tell from the recording you hear whether there's a tone change in it or not. That's why Pleco cannot ask it of you, because it doesn't know it. It's only a sound recording. I hear that it will, however, support it in a clean way with the next major release, by adding an extra field for tone sandhis.

Hope this helps,

Shun
Yes, of course! But I don't care whether there is a tone change in it or not. The point of the tone test is to test whether I can detect what I hear. Pleco knows what sound it is playing, so surely it can mark it based on that.

Let me try one last time to explain the problem. I select tone profile and audio only. That plays some sounds based on some random cards, and those sounds are sometimes based on tone changes. What I am asking for, is that when I do that, the correct answer should be based on the heard sounds, not the unseen characters.

This change should only apply when people do this kind of tone test with audio only and random characters. Not when testing your own flashcards. Not when the characters are shown. Not in any other situation.

I get that you guys are planning some deep stuff with tone changes, and I'm sure that will be great for some people in some situations, but all I'm asking for is audio only tone test with random characters is graded on the audio, and not the underlying characters that I don't know.
 

Shun

状元
Point taken, well said! It clearly is an unsatisfactory state of affairs for you right now. But is the glass half full or half empty?

As with everything and like now, when you recognize that you were on a path you can't take, for reasons beyond your control, it's often best to look for an alternative route. For example, in this case your goal is to practice tone recognition, not necessarily to use Pleco for this purpose. The most effective way for that is, in my view, a textbook with audio recordings like the New Practical Chinese Reader, volume 1, Exercise book, which has numerous well-made tone recognition exercises right at the beginning. Most textbooks start with this type of exercise, in fact.

Feel free to ask in the Chinese Language part of the forums if you need further guidance about studying a particular aspect.

Cheers,

Shun
 

Aslaksen

秀才
Yes, the app Pinyin Trainer by trainchinese does exactly what I want. I started this thread because I was hoping that this was something that Pleco would be willing to fix. I've been very impressed by other features in Pleco. Unfortunately, I haven't been too impressed by the replies to this request. :-(
 

Aslaksen

秀才
I appreciate that you were trying to help! I think the problem was that you didn't quite understand what it was I was looking for. I'm sorry if I came across as rude and unappreciative of your efforts.
 

Shun

状元
You're welcome!

I think the problem was that you didn't quite understand what it was I was looking for.

Yeah exactly, I thought too quickly that I had fully appreciated what you were getting at. I may have understood the technical issue, but not your exact point of view.

I'm sorry if I came across as rude and unappreciative of your efforts.

You didn't—I didn't perceive it that way.

Have a nice day,

Shun
 

mikelove

皇帝
Staff member
Sorry, upon reflection I did not handle this very well and I regret that you came away with a poor impression from it. I think in retrospect I should have simply stated that this was a good idea to fix and that it was on our to-do list; those are both true statements, but I got preemptively defensive about exactly when and how it ought to be fixed and started an unnecessary fight about it.

So, to sum up and simplify: this is a good idea, fixing it is on our to-do list, there's nothing really I ought to be pushing back on about it and I'm sorry we got into an argument about it. I can't promise exactly when it will be fixed, but a fix is coming. And next time somebody makes a suggestion like this, I'll leave it at that, and not offer an elaborate justification about why we're fixing it at such-and-such hypothetical future time and not an earlier hypothetical future time.
 

Aslaksen

秀才
Mike, I really appreciate your reply!!! It totally matches my sentiments, and I feel completely heard and understood. It is a very classy reply and is in fact exactly what I was expecting from the maker of this fantastic app!

I love Pleco, and I'm looking forward to seeing it become even greater in future versions!
 
By the way, is it possible that there are mistakes in the tones? For example 尽快 jin3kuai4 is 3-4 (but the first character can also be pronounced 4. The male voice is ok, but it seems to me that the female is saying 4-4.
 

mikelove

皇帝
Staff member
It's possible, yes. We realized in retrospect that we probably should have been firmer with the people making the voice recordings about following the tones as written even if that wasn't the way they would naturally pronounce that word. But nowadays TTS has gotten good enough that interest in recorded audio is pretty limited and I'm not sure if it's worth the effort to go back and re-record stuff.
 
Top