Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/eumiro/lumipallo

Snowball effect in language learning
https://github.com/eumiro/lumipallo

language-learning natural-language sentences source-language target-language

Last synced: about 1 month ago
JSON representation

Snowball effect in language learning

Awesome Lists containing this project

README 

        
.. image:: https://github.com/eumiro/lumipallo/workflows/CI/badge.svg

  :target: https://github.com/eumiro/lumipallo/actions?query=workflow%3ACI



.. image:: https://img.shields.io/pypi/v/lumipallo.svg

  :target: https://pypi.org/project/lumipallo/



.. image:: https://img.shields.io/pypi/pyversions/lumipallo.svg

  :target: https://pypi.org/project/lumipallo/



.. image:: https://img.shields.io/github/license/eumiro/lumipallo

  :target: https://github.com/eumiro/lumipallo/



lumipallo: snowball effect in language learning

===============================================



**lumipallo** is the Finnish word for **snowball** (lumi = snow, pallo = ball).

Learning a foreign language is like a snowball. Start with a tiny amount

of snow and by rolling add one snowflake after another to get a big snowball.



How it works

------------



The smallest meaningful unit of a language is a sentence consisting of words.

If you see a sentence in a foreign language, you understand somewhere

between 0 and 100 percent of those words. If you understand nothing,

you'll be overwhelmed by all those new words. If you understand all

of them, you probably won't learn much. The best learning effect

is to get a sentence with one or two words that are new or almost new to you.

One unknown word in a familiar context is easier to understand

than seeing it isolated in a dictionary.



So where to get the right sentences?



**lumipallo** uses the extensive sentences database from Tatoeba and keeps

track of the words you know. Each time it tries to find a sentence with

as few and as frequent words as possible and asks you about those

new words. It will provide you with a translation, but feel free to

check for the word in a dictionary, compare grammar tables, or just do

anything with the sentence. Your brain has to deal with stuff in order

to learn something new.



Example

-------



Let's say you speak English and want to learn German. Here comes the first

German sentence with an English translation:



    Hallo Tom.



    Hello Tom.



Then it asks for both words: *Hallo* and *Tom*.



*Tom* is a name and *hallo* is *hello*. That was easy.



    Ich heiße Tom.



    My name is Tom.



We know *Tom* from the first sentence. Now you have to learn two new words:

*Ich* and *heiße*. *Ich* is *I* and *heiße* is the first person singular

for *call themselves*. You understand the context and if you see these

words often enough, your brain will recognize them.



    Ich heiße nicht Tom.



    My name is not Tom.



If in the previous two sentences you confirmed that you understand all words,

there is only one new word in this sentence: *nicht*. That's a negating *not*.



    Mein Bruder heißt Tom.



    My brother's name is Tom.



There are three new words here. *Mein* is *my*, *Bruder* is *brother*,

*heißt* looks like *heiße*, but since you are speaking about a third

person, the form is slightly different.



The sentence *Mein Bruder heißt nicht Tom.* (*My brother's name is not Tom.*)

would not appear now, just because you already know all words

from that sentence, but might appear in a future version of *lumipallo*

just in order to show you more content.



That's one important point: knowing a word is not ``False/True``.

It's usually somewhere in between, and your brain needs to see a word

often and in different contexts to understand it fully.



Prototype

---------



The project is in the alpha stage, features may appear/disappear quickly.



Install it:



.. code::



    pip install lumipallo



Start it with:



.. code::



    lumipallo



In this first prototype there is little you can do, but this is just

to show the principle and get in touch with people interested in trying

something new.



Your source language is English (``eng``),

your target language is German (``deu``).

It has a list of 13 somehow related sentences with 15 different words

(different forms of the same word are different words).

Each session starts from zero and there is no load/save functionality.



It shows you a sentence in your target language, then in the source language.

Then it asks for every new word in the sentence.

Answer ``y`` if you know the word, ``n`` otherwise.

It should show new sentences with minimal number of new words

and these words should be the most popular (within the list of course).

When you “learn“ all 15 words, it's over.



A save/load functionality is the an essential part of the project,

because it is your own database of known words. There should

be a possibility to load your own sentences/texts to get even more

content.



Contributions

-------------



Yes. I'm looking forward to any ideas, questions, shared experiences or

code/texts contributions.