I would firstly like to offer a brief apology to anyone who bored by my “does what it says on the tin” approach to post titling. Guess I’m just not feeling too imaginative at the moment?
Anyway, today is host to another “throwing out half-baked ideas without bothering to try them first” session. The problem? Active recall is necessary for meaningful L2 communication, but creating SRS cards to test active recall takes longer than creating passive recall ones. The solution(s)? Hmm…
Scenario-Based Testing (SBT)
This is something I started to dabble with a long time ago – your SRS card has a scenario in L2 on the question side and the answer is anything you like, so long as you’re fairly sure it’s grammatical. Par example:
Q: “There’s a fat-ass roach in the bathroom”
A: “Holy crap! killitkillitkillitkillitkillit!”
- No corrective feedback (i.e. if it’s wrong no-one’s gonna tell you)
- Although the scenario is repeated at optimal intervals, your answers are unlikely to be the same each time, thereby partially defeating the point of spaced repetition
- Even though writing the questions in the first place in L2 is good practice, you need to check with a native first
- The testable material isn’t easily controlled (sometimes the production words will unavoidably crop up in the question)
- Takes a long time to set up per card for a non-native speaker.
Grammar-Guided Scenario-Based (GGSB)
This is broadly the same as the above card model, except that each card includes EITHER an unrelated grammar example sentence OR the name of that grammar point on the question side as well. The challenge is to create an answer to the question that follows that grammatical pattern. For example:
Q: “There’s a fat-ass roach in the bathroom”/present continuous
A: “I’m killing it as fast as I can!”
- Harder to get wrong than SBT
- Less variety in answers possible than SBT so more worthwhile to apply spaced repetition
- Still takes ages to set up each card
- Testable material still isn’t easily controlled
Word-Based Testing (WBT)
Two variants on this:
- Be presented with L1 and translate into L2 (waste of time with languages as dissimilar and English and Chinese/Japanese so we’ll ignore it)
- Let the question-side contain one or two words (e.g. dog and loudly) and come up with a sentence orally using those words.
- Very quick to set up
- Usefully tests vocabulary recognition and sentence production
- No corrective feedback
- Unfocussed – answers each time are likely to be very different
Grammar Guided Word-Based (GGWB)
In this model, the question side includes a word or two (e.g. fish and swim) and a grammatical pattern/label. The challenge is to create a sentence using said word(s) and grammatical pattern.
- Useful practice of converting words into sentences
- Hard to screw up grammar if there’s an example or twain to compare with sitting right there
- Good way of focussing on particular grammar points
- No corrective feedback on collocative language
- Reasonably fast to set up (esp. if you reuse the grammar example sentence(s) a few times)
And they all lived happily ever after…
So there you have it. One man’s 30-minute private brainstorming session.
There’s an underlying problem in each of these that there’s no obvious consistent way of scoring cards. Nonetheless, I think I’m going to go away and have a crack at the last one (Grammar Guided Word-Based) since I have no heart (冇心機) to make lots of cloze cards at the moment. Other suggestions/directions/tales of pre-extant attempts are as ever, most welcome. 😀
P.S. I am well aware that you could ditch the SRS for practicing active recall and just talk/write to people, but that’s not really the point of this experiment…