Glossika method

[Solarin]

Hey fabio, interesting method! Do you do any of the dictation or recording stuff? I haven't been doing that, not sure if it's worth the time. 

[hedwards]

@Solarin, dictation is always a good idea. But, if you're working with material that you've memorized verbatim, that would be a waste of time. You should do the dictation prior to memorizing the entire sentence, but after you've memorized most of the words.

[fabiothebest]

I didn't do dictation and recording yet, I'd like to try though.

[imron]

I haven't been doing that, not sure if it's worth the time.

It's absolutely worth the time, and you should absolutely be doing this even for things you have already memorised.  Memorising is one thing, being able to speak something fluently without stumbling or umming and ahhhing is a completely different skill.

 

Recording yourself is vital, otherwise you'll think you're saying everything correctly because your mind will self-correct itself.  Only by recording and listening back will you find out how bad you actually are, and what points you need to focus on to correct those mistakes.

[OneEye]

Agreed. It's a painful process, but well worth the effort and self-loathing.

[hedwards]

@imron, I agree, but I'd like to point out that for the beginning beginners, it's a good idea to run the sentences through Praat to make sure that the tones are what they're supposed to be. Or at least approximations. Or better still post the video somewhere that you can get real feed back.

 

But, yes, even just doing a recording for yourself is going to lead to much greater accountability than if you just say things out loud.

 

It's definitely something I need to go back and do, I've probably been too focused on the reading and writing as of late.

[etm001]

Hi,

 

I recently purchased the Glossika Mandarin Pronunciation and Fluency Basic 1-3 materials. The file names and organization (or lack there of) for some of the material is not obvious unless you read through the manuals first.

 

Mandarin Pronunciation

 

• Pronunciation Manual: there are a 37 audio files that accompany the pronunciation manual.
  • "GMS-ZH-PRON-C1" and "GMS-ZH-PRON-C2": accompany the tables in the "Consonant" section.
  • GMS-ZH-PRON-V01 to "GMS-ZH-PRON-V35": accompany the tables in the "Core Vowels" section.
• 2 and 3 Tone Training: there are 16 audio files for "2 Tone Training" and 64 for "3 Tone Training"
  • 2 Tone Training: the files end in "TNN" where "NN" represents a tone number, e.g.,  "GMS-ZH-PRON-T11" is the file for "1st tone + 1st tone" training.
  • 3 Tone Training: the files end in "TNNN" where "NNN" represents a tone number, e.g., "GMS-ZH-PRON-T111".

Note: in all three of the manuals there are "Click here to listen" links. These links do not work. Based on a brief exchange that I had with Mike, it appears that the links were included inadvertently, as the accompanying audio files are currently sold separately (perhaps in the future the audio will be integrated into the PDF).

 

Fluency Basic 1-3

 

The file names make more sense than those for the pronunciation package.

• GSR: all files end in "DAYNNN" where "NNN" is a number, e.g., "GLOSSIKA-GSR-BASIC1-ENZH-DAY001"
• GMS: all files end in "NNX" where "NN" is a number and "X" is a letter. Each file contains 50 sentences. The number "X" is a sequence number that tracks the order of the transcripts in the PDF file.  The letter "N" denotes the language(s) spoken in the file. For example:
  • GLOSSIKA-BASIC1-ENZH-01A: the first 50 sentences. English (1x), Chinese (2x)
  • GLOSSIKA-BASIC1-ENZH-01B: ................................... English (1x), a pause in the audio, Chinese (1x)
  • GLOSSIKA-BASIC1-ENZH-01C:.....................................Chinese (1x)

My biggest gripes so far:

• The PDFs are not organized into sections by topic, and thus no table of contents. (Perhaps as part of this method sentences are deliberately not organized by topic, I don't really know).
• You can't search the PDF, which makes it difficult to quickly find the sentence to which you are listening.
• The GSM file names don't reflect the sentences they contain. You can figure it out with simple math (i.e., if I'm listening to file #2 then I know it contains sentences 51-100 (50 sentences per file x 2), but it would have made more sense to reflect this information in the file name:
  • GLOSSIKA-BASIC1-ENZH-01_50A

Finally, in regards to the confirmation email that is sent after purchase:

• As noted earlier in this thread, the full download URL is displayed in the body text of the email, which is ugly and confusing (especially if you have to download multiple files).
• The confirmation email and download links are not immediately generated after purchase. 
• My confirmation email was routed to my spam folder.

Despite the above - which are mostly small annoyances - I'm still excited about the material and hope it proves beneficial in the long run.

[simc]

I brought the "expression" level business sentences. One annoying feature of these is there seem to be a lot of sentences where in English they say some dollar amount but in Mandarin they have done some sort of USD to Taiwan dollar conversion so number is 30x what it was in English.

 

Am I supposed to do the arithmetic in my head, or am I supposed to memorize what various quantities of USD are worth in Taiwan dollars? I don't see what possible logic made them decide to do it that way.

[imron]

I think the logic is that it teaches you the natural/common term to use in the environment you'll be using it in. That is, the sentences are designed to help develop a feel for the language rather than something to use as an exact translation.

After all, you won't be buying sofas in Taiwan using USD.

So, I can understand the logic behind it, but I also think it's not a great way of going about it, and I suspect it was done to make it easy to generate courses for opposite language pairs just by swapping the audio rather than recording different audio for each way.

[crt32]

So I bought the Basic 1 package to give it a go.

 

I still don't quite understand how I'm supposed to use the GSR files. The "manual" just says to listen to them and repeat as many sentences as possible.

However, am I supposed to pause the audio? Speak along the Chinese audio? Do I record myself?

Also the audio is really fast and makes pausing at the right moment really difficult. It also makes it difficult to compare my recording with the Chinese audio.

 

I also tried the GMS method, but it took me more than one hour to learn 10-15 new sentences and reviewing was not fun at all, as I (due to the lack of "real" spaced repetition) reviewed the easy sentences as much as the more difficult ones.

 

Finally, I cut up all the GMS files into single sentence audio files (English and Chinese separately), and cut up the PDF (which unfortunately contains images as opposed to plain text) into images each containing one sentence.

Then I put everything into Anki, with English audio on the front side and Chinese audio as well as the pictures with the sentence on the back,

Now I'm getting through 40 new sentences plus reviewing old ones in about one hour total, using Anki's voice recorder to record myself.

This method works great for me, although cutting up all the material was a bit annoying.

[wibr]

Just to make this clear, the pdfs are vector images not pixel images! So the quality is the same but you can't export/select any text.

 

I use the GSR files at the moment, but I already know most of the words they use in this level so I use it for fluency in pronunciation and to internalize sentence structures. I just listen to it, occasionally pause it to repeat the sentence and at the end of the file where new sentences are introduced I pause after every sentence to repeat it and maybe look it up in the files.

 

@crt32 Did you cut it up manually or did you write a program? Right now my method works ok for me but I am thinking about writing a small script to cut it up and then try to OCR it, I would do it right away if I could share the files but since they are copyrighted I would do that just for me which is not worth the effort right now...

[crt32]

@wibr: I used pdftk to split the pdf into single pages (pdftk burst), then I used IrfanView's batch processing to crop out three different areas on each of those pages.

For the audio, I used Audacity's Sound Finder. I also wrote some python scripts to generate a tsv file for Anki and rename the files appropriately.

Since I'm not sure if I'm going to buy another set of sentences, I haven't fully automated everything yet.

[tysond]

I am thinking about writing a small script to cut it up and then try to OCR

 

 

 

OCR?  Haven't yet found a reliable way to batch OCR Chinese text.   Always rely on sources that are already converted.   Anyone with success in this space?

 

I use Pleco has good OCR for basic use (like taking a photo of a page of a textbook).  But I'm at the point that it's limiting returns because taking a photo takes long enough that I can actually understand 95% of the sentence and type it myself anyway.  But for large collections of text I haven't found a good OCR solution.

[wibr]

I have used finereader and in my experience the accuracy is the same for Chinese and English, around 99.5%. OCR is never 100% unfortunately so you still have to check everything by hand...

[hedwards]

@tysonD, if you're looking to use Pleco, you should convert the PDF into a series of images that you can feed to Pleco. The main issue with doing that is when there's text that's been underlined or annotated making a few of the characters unrecognizable to the program even if you can identify them yourself.

[Snuggles]

Does anybody have an email for the people over at Glossika? I tried purchasing but it didn't go through and I didn't see contact info on the website.

 

Also I have seen posts elsewhere that suggest this company may not be above board.  People claim to have had trouble with them. Anybody have any experience with that?

 

https://www.forumosa.com/taiwan/viewtopic.php?f=8&t=5937

 

In all due fairness...these posts are OLD...2002..so they might not count for much.

[roddy]

"I have seen posts elsewhere"

Link us up, Snuggles...

[fabiothebest]

What do you mean by "it didn't go through"? I purchased their products and although I had to wait max 24 hours in order to get download links, I didn't have problems. Did you have a problem with the payment? Or didn't you receive the confirmation email? If you didn't see the email I suggest you to check your spam folder. It's true that the email isn't written on the website, except here (accidentally), anyway you can try to contact the owner at this email: glossika@gmail.com . I successfully did it.

[Snuggles]

It was a paypal error and I did get it to go through. Now I am waiting for the download links. I hope the materials are good.

 

They really need to add contact information to that website. 

 

Update*

 

I got the downloads links within the stated 24 hours.  The materials I have downloaded at first look (only a quick glance) look nice.  I didn't mean earlier to claim that Glossika is a bad company as I had no experience with them.  So far everything seems good and problem free.

[Snuggles]

I have started going through the Glossika materials.  One thing I think people should note is the choice of font.  You can see the Glossika font (on the right) compared to the dictionary.  If you didn't already know the characters perhaps this could get confusing.

 

Otherwise the materials seem pretty good.  It has been already said that the accent isn't standard northern Mandarin.

 

If anybody has any questions I would be happy to answer them.