Mandarin Sentences

Mandarin Sentences was created as part of the Trainslate Suite because of the lack of reading material suitable to the continually progressing level of a person learning Mandarin. Books, or newspapers sooner or later confront a language novice with unfamiliar characters, grammar, and vocabulary. A computer on the other hand can provide a translation and pronunciation hint in an instant.

Have a look at the video, to see how the user is presented one sentence after another. The words are usually obscured, while listening comprehension is exercised. In case of a new word, or ambiguity the current word can be looked up in the dictionary.

All the resources are gathered from the Internet:

With the collection of sentences from the largest online Chinese newspaper, and encyclopedia at hand, Mandarin Sentences aims to present a sequence of sentences that are most suitable to strengthen and advance the users level of skill. The ordering of characters are influenced by the course Integrated Chinese. At the begin of each session, the user selects his/her current personal character count. The software then displays sentences conprising exclusively of characters below the threshold. If a character or words is unfamiliar, has slipped the mind, or the pronounciation is unclear, the user marks the character or word. The software ensures that the variety of sentences, words, and characters makes the reading experience enjoyable and entertaining.

Trainslate (Java 1.6) * Trainslate.zip 45 MB
Trainslate (latest source, Java 1.7) Trainslate_src.zip 170 kB
Pinyin audio of Google Translate (mp3) ** pinyin.zip 4.3 MB
* Credits: Wee Lih Lee helped me with the grammar decomposition
** audio files of 1243 different syllables and tones

With the advance of technology education is undergoing revolution. TED talks

The following presentation ...

Look, Your Worshipfulness, let's get one thing straight.
I take orders from just one person: me.
Han Solo

Monitoring progress


I have used Trainlsate in 41 consecutive sessions over a period of 44 days. A total of 49 hours of EEG was recorded. The vizualization of the complete EEG data is 4 GB. On this website, we show the EEG for two words as examples that were learned with different success rates:

建设 (jiàn shè), learning to recognize the word is quite sucessful. The pitfall happens between two sessions.

进行 (jìn xíng), failure to reliably learn the word despite frequent exposures

Artifacts

The long term recordings from a single subject also consitutes a good foundation to average eye blinks or characteristics of the amplifier hardware and electrode cap.


Das letzte Hemd hat keine Taschen.

Implementation details

The use of electroencephalogram in combination with Mandarin Sentences is optional. In order to monitor the voltage fluctuations across the scalp, we employ the NuAmps 40 channel EEG by NeuroScan Labs. The electroencephalogram and the language software are synchronized via 'trigger' signals through the lpt port. The Utility Package For Java by Aegidius Plüss provides the class LptPort to carry out the parallel port communication. Note: the files inpout32.dll (69120 bytes version) and jlptport.dll need to be made accessible to your java sdk/jre to run the ensemble on Windows 7. Moreover, in our experience, two computers are required in parallel to ensure flawless operation: One computer to read-out the EEG, for instance with Acquire 4.5 by NeuroScan, and a second computer to present the GUI of Mandarin Sentences and playback the audio.

NuClient (client to Neuroscan Acquire 4.5) Nuclient.zip 250 kB
Political language is designed
to make lies sound truthful and murder respectable,
and to give the appearance of solidity to pure wind.
Eric Arthur Blair