Skip to content


Switch branches/tags

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time


Pamparam is a new spaced repetition (SR) memory cards implementation for Emacs.

Spaced repetition is an algorithm for learning and repeating cards. Cards for which you get high scores get scheduled for repetition much further into the future than those with low scores. This means you spend less time repeating things that are easy for you to remember and more time on things which are hard for you to remember.


git clone
  1. Open pamparam/doc/sets/capitals/ and call pamparam-sync.

    You get a message like 35 new cards, 0 updated, 35 total and the file is opened.

  2. Call pamparam-drill.

    You get two buffers. One buffer, named e.g. is your schedule file. It holds 10 cards that you should do today. The other buffer is a card asking you the capital of e.g. Estonia.

  3. Enter Tallinn. (the dot at the end starts the card validation)

    Since the answer is correct, you get a score of 5 (best one). Alternatively, if you reveal the answer with S-TAB and enter Tallinn., you will get only a score of 3. Finally, if you enter a wrong answer, you get a score of 0.

  4. Theoretically, you can stop here with 1/10 cards done. Resume at any later stage with pamparam-drill. Or press n (calls pamparam-drill) to select the next card. Any cards scheduled today but not finished will carry over to tomorrow.

Assuming you finished all 10 cards, what’s next?

  • (optional) call pamparam-commit to commit your progress.

    It’s not really required, but it’s neat to track everything, and it’s a nice restore point if you ever mess up your repo.

  • if you feel like doing more cards, call pamparam-pull.

    I don’t recommend doing more than 50 cards in a day, because you’ll be spending a lot of time on repetition in the future.

  • come back tomorrow for a new set of cards to repeat and learn.

How this works


After pamparam-sync and pamparam-drill, you get the following tree:

└── capitals.pam
    ├── cards
    │   ├── 00
    │   │   └──
    │   ├── 01
    │   │   └──
    │       ... (more cards) ...

With this example, let’s get the terminology out of the way:

Master file

This is an Org file that has a heading tagged with :cards: that contains the definition of all your cards.

Card file
e.g. capitals.pam/cards/00/

This is an Org file that encapsulates a single card. Besides the front (the question) and the body (the answer), it also contains the metadata. The metadata contains: dates and scores of previous learning sessions with this card and the card’s accumulated ease factor.

Schedule file
e.g. capitals.pam/

This is an Org file that stores the schedule for a particular day as TODO items with links to card files. Each item’s state starts at TODO and ends at DONE which you finish the corresponding card. An intermediate state REVIEW happens if you get a score lower than 5 for the first time. When doing the card for the second time, REVIEW becomes DONE with a score of 3 or more.

Pile file

This is your cards pile. Very similar to a schedule file, it holds all cards that are not yet scheduled. There are two ways to interact with your pile: add to it by calling pamparam-sync from the master file or remove from it by calling pamparam-pull.


A Git repository to store all your cards files, schedules, and the pile. Note that the master file isn’t here. This is intentional, since any of your existing Org can become a master file simply by adding a :cards: tag to one of the outlines. See pamparam-alist for a way to connect a master file to a repo that’s not in the same directory.

Master file example

A master file is an Org-mode file with things you want to learn.

One simple example is provided in

Here’s another, slightly more elaborate, example, which I use for learning Dutch:

* Cards            :cards:
** comic story
het stripverhal
*** The Adventures of Tintin is a world famous Belgian comic strip.
De Avonturen van Kuifje is een wereldbekend Belgisch stripverhaal.

** singer
de zanger
de zangeres
*** The singer is only known in Belgium.
De zangeres is alleen in België bekend.

The format of the master file is fairly straightforward:

  1. There needs to be one or more card sources - first level outlines marked with the :cards: tag. A separate card file will be created for every second or third level child of each card source. In the example above, four cards will be created.
  2. For each second or third level outline, the heading name is the question and the heading body is the answer. I usually put e.g. a noun or a verb into the second level, and a more elaborate example of using that noun or a verb into the child third level. I also like to organize the words by generation rules and thematically, so that e.g. honest will be close to modest and very close to honesty.

    The only hard and enforced requirement is that all heading names must be unique.

  3. If a word has many correct possibilities (like de zanger and de zangeres both mean singer), I put each on its own line. This allows to enter either synonym during validation.

The master file is a great summary of the info that you have available. It’s easy to search and organize.

One more option is to put all cards as level one headings. In that case, instead of tagging them with :cards:, you can add to the top of the file:

#+PROPERTY: pamparam t

Card file example

A card file looks like this:

* m
#+STARTUP: content
** scores
| <2017-04-23> | 3 |   |
** stats
(setq ease-factor 2.360000)
(setq intervals '(1))
* Slovenia

The first heading holds all the metadata, like:

  • all times and scores when you did a card
  • your wrong answers, if any
  • an estimate of the card’s ease

The second heading’s name is the card’s front, the question. The second heading’s body is the anwer, it starts out hidden.

Card scoring

5perfect answer, body hidden
4wrong answer, pamparam-card-redo was called, followed by a perfect answer
3perfect answer, body revealed
0wrong answer

On the first try, you can get either a 5 or a 3 or a 0. Unless you get a 5, you have to REVIEW the card today.

You can use pamparam-card-redo if you make a mechanical typo and get a 0, even though you knew the card. If you manage to correct the typo, you get a 4. You can make use of C-y to yank your previous answer.

In the REVIEW stage, entering the answer with body revealed is acceptable to move it to DONE. Still, you might want to try to keep the body hidden.

Commands and key bindings

Certain commands are applicable only in certain types of files. There are 3 types of files, all of which use org-mode: master, card and schedule.

Card file

Global bindings:


Local bindings (only active if your point is at a heading start), in order of importance:


Master file


Anywhere in the repo



While it’s possible to have multiple repos, currently I have only a single one. In my case, it’s not convenient to keep the repo dutch.pam in the same directory as the master file So I use this setting:

(setq pamparam-alist
         . "/home/oleh/Dropbox/source/site-lisp/git/dutch.pam")))

pamparam-drill doesn’t know where your repos are located. It can only determine if the current buffer’s file belongs to a repo or not. In case it does, the current repo is used. Otherwise, the default repo is used which is pointed to by pamparam-path.

By default, pamparam-path points to the repo of the provided example master file. Here’s my custom setting:

(setq pamparam-path "/home/oleh/Dropbox/source/site-lisp/git/dutch.pam")

Finally, you can have all key bindings in one place with a hydra:

(global-set-key (kbd "C-c m") 'hydra-pam/body)


Simple and fast flashcards for Emacs






No releases published


No packages published