Building a dictionary for a language with none

I grew up speaking Aklanon, but I’m far from fluent.

Whenever my friends and I chatted in Aklanon, I’d occasionally run into a word I didn’t know or wasn’t completely sure I understood. Sometimes we’d even disagree about what a word actually meant or how it should be used.

So I’d search for it. Most of the time, nothing useful came up. If I was lucky, I’d ask my parents.

That experience made me realize how little digital infrastructure existed for my own language.

So I built Aklish.

Aklish is a web application where people can contribute, vote on, and discover Aklanon-English translations. Underneath that, it’s really an attempt to build the pieces that a language needs to exist online: a searchable dictionary, quality control through voting, a spell checker, and learning tools that encourage people to keep using the language.

Why I built it

Aklish started as my Senior High School Research Capstone Project, but the motivation came long before that.

I wanted a place where I could reliably look up Aklanon words without relying on scattered sources or asking ancient family members. More importantly, I wanted something that could continue growing as more people contributed.

Instead of building a static dictionary, I decided to build a crowdsourced one.

That meant users could submit translations, vote on existing ones, and gradually improve the quality of the dictionary together.

Building the dictionary

The biggest challenge wasn’t writing code. It was getting the words.

There wasn’t a digital Aklanon dictionary I could simply import, so I manually encoded more than 5,000 dictionary entries from an old printed dictionary. Looking back, modern OCR models might have made that easier, but at the time the scanned pages were messy enough that manual encoding seemed like the only realistic option.

Those entries became the foundation that everything else in Aklish was built on.

Letting the community improve it

A dictionary alone wasn’t enough.

Language changes over time, and many words have multiple meanings depending on context. Instead of making every translation fixed, Aklish lets users contribute their own translations while the community decides which ones are most useful.

Users can submit translations from either language, vote them up or down, and build reputation through their contributions.

The reputation system was inspired by Stack Overflow. Voting helps surface better translations while rewarding people who consistently contribute high-quality entries.

Helping people contribute better translations

One feature I’m particularly proud of is the spell checker.

Because contributors were writing in both Aklanon and English, I wanted to catch obvious mistakes before they became part of the dictionary.

The spell checker uses SymSpellPy for spelling suggestions together with LemmInflect and AklStemmer to recognize inflected English and Aklanon words.

Building this feature also led to another project entirely. While implementing the spell checker, I realized dictionary lookups weren’t enough because inflected Aklanon words often don’t appear in dictionaries. That problem eventually became AklStemmer.

Making it usable

Since the project depended on community participation, I wanted it to feel approachable instead of looking like a research tool.

Users can browse and search the dictionary, create accounts, contribute translations, and use the application comfortably on desktop or mobile.

Looking back now, it’s funny seeing how I built the interface almost entirely with Django templates and Bootstrap. This was before I had learned React or modern frontend frameworks.

Encouraging people to learn

I also added a few small games to make the platform more engaging.

There was a Wordle-inspired game and a synonym-antonym matching game. The idea was that repeated exposure to vocabulary would help people become more familiar with Aklanon while making the platform feel less like a dictionary.

What I learned

Aklish taught me how much work goes into building a real web application.

Before this project, I had never built something with authentication, user-generated content, moderation, search, reputation systems, deployment, and thousands of records. Building every part of the application myself gave me a much better understanding of full-stack development than any classroom exercise could.

It also changed how I think about community platforms.

I expected people would be excited to contribute translations. Instead, many users spent more time playing the games than adding new words. This showed me that building a community is much harder than building the software itself. People need reasons to come back, and those reasons aren’t always the ones you expect.

Even so, I’m still proud of what it became.

Aklish started because I wanted a better way to look up words in my own language. Along the way, it grew into a platform that brought together a dictionary, spell checker, crowdsourced translations, and language-learning tools.

It also became the foundation for Kulasisi, where I’m now applying the same ideas to many Philippine languages instead of just one.