OmegaWiki and language exercises

From Meta, a Wikimedia project coordination wiki

At the moment early brainstorming on this topic. When we have some ideas, the following step would be a anouncement on the mailing lists

Some of the programming could be done via sponsoring. See language exercises extension programming

Feel free to add aspects to the lists or on discussion page or contact me on #wiktionary

to be combined/checked with this[edit]


Automatic generation of exercises[edit]

  • Exercises with correction in users mind
    • Masking certain words in texts from Wikipedia... in the language you want to learn. The learner browses arbitrary texts of his intrest in the foreign language and he can guess masked words from the context.
      • Example of random-masking of vocabulary wikipedia articles mokup with external programming The article is dynamicaly taken via the url parameter from wikipedia. Additionaly there might be lookups to wiktionary and posibilities to copy and create phrases with masked words to a users >exercise watchlist<
      • There might be special wordlists in the user account like
        • >my current words< (the red checkbox symbol in the example above)
        • >word I know (now)< (the green symbol)
        • >not interesting/ mask never in the future< (the trash symbol)

  • Exercises false/correct feedback

automatic generation of such exercises from OmegaWiki might possible for some content types e.g.:

    • give example phrase with gap and infinitive as hint behind the gap. If in the example phrase is a inflected form of the lemma, the hint could be taken via associated inflections on that lemma. This might be useful too with other relations:
      • inflected forms of that word like person or time
      • choose from list what's a synonym or antonym to a given word
    • Take the lemma as gap from idiomatic expression or proverb sections
      • Essere di /casa/ (essere frequentatore abituale di un certo ambiente)
      • Mettere su /casa/ (crearsi una famiglia)
      • Giocare in /casa/ (agire in un ambiente conosciuto)

Editing exercises by authors[edit]

Some exercises might be derived from texts (perhaps from wikipedia texts about the country of the language to learn). There are some XML-based standards for interchange but additionaly there could be a syntax for human authors like wiki-syntax. This exercise pages might usually fit better in wikibooks than in wiktionary. Ideas in wiki definition syntax for language exercises.

Types of language exercises[edit]

  • Fill gap with Text hint this could be something like this
    • same language
      • not inflected form given
      • Antonym etc given (automatic creation?)
      • audio given (automatic creation)
  • Fill gap with visual hint
  • Fill gap with audio hint

Grouping questions[edit]

(different preselected groups, perhaps have this outside via link parameter collection e.g. in wikibooks Perhaps this could be generated from a language Wikibook Template. Example on [1]

Saving user interaction data[edit]

(most extensive way would be storing all entries by the users for diagnostic purposes..., perhaps not on wiktionary but external gateway e.g. at universities between unser and wiktionary)

  • selected questions by one user
  • which questions next for a user
  • interaction like vobabulary box, repetition of former wrong tries

Using categories[edit]

We have begun work on categories that could be useful in creating language excercises. Categories may encompass words on certain subjects, such as basketball or music, or they may contain words with certain characteristics of interest, such as heteronyms or idioms. Either sort of category could serve as a convenient way to group words for language exercises. We also have various pages and categories containing basic word lists that might interest beginners.

Using audio[edit]

Wiktionaries are not paper, and as such, have begun to contain audio pronunciation. Sound is crucial to languages, especially for beginners. Even in languages that are strictly phonetic, the phonetics must first be learned. Thus, efforts have begun to incorporate audio pronunciations into some of the Wiktionaries. Language exercises would also benefit greatly from embedded audio pronunciations of words and sentences.

Currently, Ogg Vorbis is the audio format of choice, because the format is free. For language exercises, especially, it would help to streamline the process to play these files, so that users do not need to save the files and launch an external application to play them. We may wish to devote some attention to streamlining the upload process, as well, to encourage more contributions.

Training of pronunciation[edit]

When people learn a language, they need training with their pronunciation. On the Internet this is possible because of tools like Skype. You can have one on one and three to one ratio of pupil to teacher. This may need to be payed service, but even so it is essential that enough training is given at the start to make the pronunciation "bearable".

Points to start with[edit]

dicologos childrens dictionaries[edit]

(see for further information Wiktionary_and_dicologos)

  • exercise types from this perhaps (development of exercise extension...)
    • image with choose/type/see word (images already there)
    • image then sound
    • translation with choose/type/see on click word
    • something like "superlearning" seeing the picture, hearing the pronounciation, some music...
  • contents
    • Uni Bamberg LangLab could work on Farsi and German with this
  • how to combine this with material on wiktionaries. There might be several lists of basic vobulary?

derive exercises from available synonym + antonym + definition[edit]

  • exercise types
    • definition with choose (list of antonyms and synonyms)/type/see on click
  • content from
    • dicologos (More info about this to come...)
    • UltimateWiktionary (to come)
    • relations in current wiktionaries (?)