A learning tool · the Nazief & Adriani algorithm
Watch an Indonesian word peeled back to its root.
Search engines and language software strip affixes first, so that “mempelajari”, “pelajaran” and “belajar” are all recognised as the same word: ajar. Normally that happens behind a closed door — word in, answer out. This tool opens it one layer at a time, including the times it gets the answer wrong.
It looks like this
One affixed word, peeled from the outside in until only the root is left.
affixed word
root (kata dasar)
ajar
Grey is an affix that has come off. Red is the final answer. On the trace page every one of those cuts gets its own line, with the rule that made it.
Why this needs opening
When it is wrong, the wrongness is invisible
A bad result is still a plausible-looking string. Without seeing which rule fired and which dictionary lookup succeeded, a wrong answer and a right one are equally convincing.
The answer depends on the dictionary, not only the rules
The algorithm checks a list of roots at nearly every step. Swap the list and the answer changes. Here that list is visible and you can edit it — and the fastest way to understand it is to break it on purpose.
More than one answer is often correct
“Beruang” reads as ber- + uang, or as the root beruang. No amount of rule refinement settles it. The algorithm picks one and says nothing; here they are all shown at once.