Jump to content
Learn Chinese in China

The long plateau

  • entries
  • comments
  • views

About this blog

I've been following Publius' and Imron's advice: Transcribe by hand to improve listening ability. Here I'll keep track of the results.

Entries in this blog


90-day transcribing challenge: Final update

This morning I finished day 90. I used two types of content: 


1) Clearly spoken stuff:

Slow Chinese, HSK5 recordings, and a magazine podcast for natives. Sometimes I prepared subtitles beforehand using WorkAudioBook, and during the transcribing session I thus was able to do corrections immediately after each line. This lead to time "wasted" doing the corrections, but also stopped me from repeating the same mistake again. Other times, I did not prepare subtitles, and just used WorkAudioBook for automatic segmentation, and did the corrections after finishing each session. This, of course, can cause an accumulation of errors in repeated words, but also means I could write more in a session, as I was not distracted with corrections. So... The left column in the data is not very consistent in how it was done, and even less with the material used. In day 52 I forgot to start the pomodoro clock, so I got an outlier score. I'm leaving it out of the monthly averages.


2) A TV drama called Great Marriage.

I downloaded both mp4 video and srt subtitles from YouTube and used them with the fantastic Lingual Media Player, which can automatically stop after each subtitle line and makes it easy to toggle subtitles. In 90 days I only reached episode 8 of a 40+ episode drama, and that's watching long parts without transcribing! So, with this abundance of ready-made material, the right column is consistent both in source and in execution.


During the first 75 days, I did 2 pomodiri (50 minutes) per day for each column. But two weeks ago I signed up for December's HSK5, so, to make time for vocab study and practice tests, during the last 15 days I only did 1 pomodoro (25 min) per day for each column. So, in order to "normalize" the scores with the previous days, I added a *2 in the formula.


You can also notice that around day 32 I also started to seriously attack my Pleco SRS backlog. The number here is how many pending cards I have each morning. 


My observations:


Clearly spoken stuff

You'll notice that during the last month my average score actually dropped for "clear stuff". Maybe in part because I switched exclusively to a magazine podcast for natives in day 60. I must add that, although this podcast is for natives, the magazine is a Chinese translation of the English original, and the podcast is actually just read from the magazine, so it's not at all like 原来是这样 or any similar 100% native, conversational podcasts.


TV drama

In the graph, you'll also notice that, after a fantastic increase in comprehension from the fist month to the second month, the're no such big increase for the third month. Maybe I'm hitting "diminishing returns" with this particular drama. Still, I've learned a lot!



As mentioned, I'm attacking HSK5 on December, just as a personal challenge, not for scholarships or anything. My cousins, who are Chinese teachers at the local Confucius institute, passed this exam two years ago and then went on to get their Master's degrees in China, but my current level is nowhere near what theirs was two years ago! My current level fits perfectly the B1 description given by the Europeans. Still, after measuring myself with a couple of old HSK5 papers, I discovered I can pass, even if they completely discard my two essays. So in part I'm taking the test to prove a friend of mine that HSK is actually just B1... So I signed up for a test preparation class at the local Confucius. Nobody else signed up for level 5, so I accepted being put with level 4 test takers. My teacher can't speak Spanish, which helps.



So yes, this helps. The data shows it. I believe this has mostly given me confidence with my handwriting, as, before this, I only wrote individual words. This will certainly come in handy during the HSK5 writing part, because the only option available in my country is the paper test. During my attempts with past papers, I found this part to be the most relaxing. I can finish it in half the time. Of course, with awful grammar! (My teacher will help me with my writing). I haven't really done any traditional study of grammar after an introductory course back in 2012. It's been mostly input, input, and more input, particularly after I finally took Chinese seriously in 2015 and started with Heisig's Remembering Simplified Hanzi. Of course, I've checked difficult to understand points with Pleco and the Chinese Grammar Wiki along the way.


So, what will my listening practice be now? I'll be attacking every single HSK5 past paper I can find, so that will be it, for the most part. I'll also keep watching the drama with LaMP, but without transcribing it. I might transcribe dubbed videos of talks, however, just to keep writing.


Thank you for reading! Suggestions are welcome. I'm attaching the raw data, the monthly averages and a sample of my "day 90" handwriting. Now my focus will switch to reading speed, as it's currently my weakest point. I'll soon write another post about it.






The 90-day transcribing challenge: First complete month

Back in March, I started to transcribe as a listening exercise, but life happened, and I had to stop. Now I'm doing this again, this time with these rules:


1) I'm doing it for 2 hours every day:

  • For one hour, I work with slow, clearly spoken material (Slow Chinese)
  • For one hour I struggle with a Chinese TV drama (Great Marriage).

I might rethink this mix, but I stuck with it for the first month.


2) I use WorkAudioBook for Slow Chinese and Lingual Media Player for the drama. The drama has .srt subtitles and I prepare subtitles for Slow Chinese days before using each episode for this listening exercise. I only listen to one subtitle line at a time. Each line is 1-7 seconds long.


3) I can use anything to get the transcription done: Pleco, Google, Baidu, etc, but NOT voice recognition software or looking at the original transcript itself. That would be cheating.


4) I work with paper, mechanical pencil and an eraser in paper with 25 squares per line, with separations to write down corrections and omissions. This makes it easier to compute total characters and count errors/omissions.


5) After each session, I count total characters originally written and subtract double the errors. This is my score. When I reach 1,000 points in one day, I'll go out and celebrate in my favorite pizza restaurant.  I subtract errors twice because this way I force myself to try and strike a balance between speed and accuracy: I can't go too fast because my accuracy might fall below 50% and then I get negative scores; I can't linger for too long on difficult parts because then I won't be making more points in easier parts. There's only so much you can do with Pleco and Google when you just can't make out the exact sounds people are saying, so you have to choose when to move on. 


So far, I've seen progress. With both Slow Chinese and the drama, the proverbs, quotes and fixed expressions are the most difficult. Of course, the drama is way more difficult than Slow Chinese, so now I'm skipping the parts where several people are talking at the same time/quarreling/shouting/drunk/etc. I find that my handwriting is becoming faster, but sloppier. And I'm learning to understand both a slight Zhejiang accent (Slow Chinese) and Northern accent (Great Marriage), and how sounds are slurred together/omitted/whispered/etc.


I want to sit for the HSK in order to have a clear goal to work for, so I'm considering switching from transcribing Great Marriage to transcribing past recordings of the HSK listening section, to really get used to them. Of course, during the exam you won't have the luxury of replaying, but still, this might be a useful exercise and a break from all the quarreling found in a TV drama. To be honest, I find it disheartening that even pausing and repeating and using Pleco and all that stuff, I can only get around 70% of the drama dialogue right. It can feel like a chore, so I guess I'll take a stroll through the HSK forest and then come back to the TV drama when I have more vocabulary. What do you think?


I'm also attaching a sample of my handwriting. Above the line, it's the drama. Below the line, it's Slow Chinese.


To clarify, the "stopped" information doesn't reflect how much I transcribe in each session, because I sometimes keep watching/listening after finishing the transcribing session. It's just a bookmark.





Publius' 90-day transcribing challenge: First week

I've been following Publius' and Imron's advice: Transcribe by hand to improve listening ability. Here I'll keep track of the results.

Here's the original thread: https://www.chinese-forums.com/forums/topic/53708-transcribing-mandarin-as-a-learning-method/#comment-412781


To ensure consistency in how I keep this record, I've created my own arbitrary "rules":


-Before starting, I watch aprox. 4 minutes of the video I'll be transcribing, without subtitles. This time is not counted. This is the "step 2" we talked about in the forum thread.

-I transcribe using LAMP player, which allows me to quickly toggle subtitles on/off, pause after each line, and repeat a line, all with keyboard shortcuts.

-I work for 2 25-minute sessions per day, separated by a 5-minute break

-If the break starts when I'm mid-sentence, I leave it there and continue after the break.

-If, during the second 25-minute session, the time's up and I'm mid-sentence, I finish the sentence.

-I'm writing by hand on paper, with a mechanical pencil and an eraser. I designed my own page w/25 8mm squares per row, and left enough separation between rows for corrections. I was previously working with 7.5mm squares, but it felt cramped. I'm attaching the PDF in case you want to use it. It has 2 pages to make it easier to print it double-sided in duplex printers.

-I don't consult any dictionary when I'm writing down each line.

-After each subtitle line, I show the subtitles (toggle with Q in LAMP player) and make corrections. Only now I do consult the dictionary for new words, words I got wrong and to clarify stroke order if I'm in doubt. This correction is done after each subtitle line for immediate feedback, and included in the 25 minutes. This way I got a second chance to get a word right if it appears often within the session.

-I'm sticking only to this particular type of video: Documentary/interview style, clearly dubbed to Mandarin from the original English. Subject matter is very familiar to me. I did watch it in Spanish days before starting to transcribe it in Chinese, so I know the general idea, but I already forgot the specifics.


As you can see, my stats go up and down wildly from one day to the next. I guess I need to complete the first 30 days and average the results to get a better picture of my current ability. I can then compare that result to the next 30 days, and the final 30 days.

first week.PNG


  • Create New...