Because vocabulary acquisition should be as simple as possible - but no simpler.
Foreign language learning researcher Dr. Paul Nation has
long championed the use of extensive reading
for the acquisition of vocabulary in the new context. wordmeat attempts
to supplement this practice by providing a straightforward way to
generate
high-quality practice sentences from your target language to
English, using the magic of
the OpenAI API.
wordmeat defaults to 10 sentences
per word, based on his estimation that learning a new vocabulary word
from high-quality extensive reading requires 7 to 8 meetings. We can't
give you that, you'll have to spend plenty of time actually reading
authentic material to get there -- but we can get you started with
something a little more comprehensive (and, dare we say, fun!) than a
basic dictionary flashcard.
You will need:
- OpenJDK, version 22 or higher; and
- Apache Maven, version 3.8.7 or higher.
If you want to run the test suite, you will also need JUnit, version 5.8.1 or higher.
You may also find IntelliJ IDEA Community Edition or higher useful as well, as this project can be opened directly in IntelliJ and ran using its GUI interface. Here, we will simply show you a command line usage of it.
First clone this repo:
git clone https://github.com/hiAndrewQuinn/wordmeat.git
cd wordmeat/Next, use Maven to build and run the program.
# while cd'ed into wordmeat/
mvn clean installThen you can run the program with
mvn exec:java -Dexec.mainClass="siilikuin.finstem.Main"This will print the default, and ask you (if you haven't already) to
set an OPENAI_API_KEY variable. You can get one of these by going to
https://platform.openai.com/api-keys.
Target Language: Finnish (fi)
Word: sana
The environment variable OPENAI_API_KEY is not set.
Please set the OPENAI_API_KEY environment variable to your OpenAI API key.
Examples:
export OPENAI_API_KEY=your_openai_api_key_here # Unix-based
setx OPENAI_API_KEY "your_openai_api_key_here" # Windows
wordmeat uses
picocli
under the hood for its command-line interface. To actually get the
program to do anything useful, you need to specify at least 2 things:
- A
--word, and - A
--target-language, where the target language in question is given by its IETF primary language subtag.
For example:
mvn exec:java -Dexec.mainClass="siilikuin.finstem.Main" \
-Dexec.args="--target-language es casa"You can see the full command line options by running
mvn exec:java -Dexec.mainClass="siilikuin.finstem.Main" \
-Dexec.args="--help"If you wish to run the test suite as well, you can do so with a simple
mvn testEventually we plan to ship this as a native executable using something like GraalVM. For now, this is the most foolrpoof method to run this program.
Contributions are welcome! Please read CONTRIBUTING.md for details on the code of conduct and the process for submitting pull requests.
This project is licensed under CC0.