prokem

English · Bahasa Indonesia

Converts Indonesian and Javanese text into four secret languages, and back.

What this is

Prokem is the Indonesian word for an argot — a way of speaking built to be opaque to outsiders. The word is an example of the thing it names: it comes from préman, "street tough", chopped up and rebuilt around an inserted -ok-, the same pattern that turned bapak into bokap and gave a generation of Jakarta slang its shape.

Across Java the most durable of these argots work by walikan — "reversal", from the Javanese walik, to turn something over. A word is put through one fixed transformation and comes out unreadable to anyone who does not know the rule, and perfectly ordinary to anyone who does. Each city settled on a different rule, so the transformation doubles as a badge of where you are from: Yogyakarta swaps letters across the twenty-letter hanacaraka alphabet, Malang reads words back to front, Semarang pairs the same alphabet a different way.

These are not ciphers invented for this package. They are spoken, painted on shopfronts, and used daily — by traders, by football supporters, by anyone who wants the person beside them to understand and the person behind them not to.

This package implements four of them, in both directions.

npm install @khaeransori/prokem

import { semarang, jogja, malang, unang } from '@khaeransori/prokem'

semarang.encode('mangan')      // 'kahath'
semarang.decode('kahath')      // 'mangan'
jogja.encode('mangan')         // 'daladh'
malang.encode('malang')        // 'ngalam'
unang.encode('hancur')         // 'uncar hanung'
unang.decode('uncar hanung')   // 'hancur'

Every encode and decode takes arbitrary text: punctuation, digits, spacing and capitalisation are preserved, and only runs of letters are converted.

semarang.encode('Mangan sik, 2 menit!')   // 'Kahath jim, 2 kethit!'

Import a single dialect when bundle size matters:

import { encode } from '@khaeransori/prokem/jogja'

Drive a picker with the dialects record:

import { dialects } from '@khaeransori/prokem'
dialects[name].encode(text)   // name: 'semarang' | 'jogja' | 'malang' | 'unang'

CLI

$ npx prokem semarang "mangan bapak"
kahath calam
$ echo ngalam | npx prokem malang --decode
malang

Browser

<script src="https://cdn.jsdelivr.net/npm/@khaeransori/prokem/dist/prokem.js"></script>
<script>prokem.semarang.encode('mangan')</script>

The dialects

Walikan Semarang (boso gali)

The first ten hanacaraka letters are swapped with the last ten in reverse order. Vowels stay, and the formula is symmetric.

ha  na  ca  ra  ka  da  ta  sa  wa  la
nga tha ba  ga  ma  nya ya  ja  dha pa

A word-final consonant whose swap is awkward to pronounce (t, s, h, ng) is left alone, so mas becomes kas rather than "kaj", and sikat becomes jimat rather than "jimay". Homorganic nasal clusters (mb, nd, ndh, nj, ngg) count as one sound and lose the nasal before swapping, so ombe becomes ngoce.

Attested vocabulary, applied before the formula, is exported as LEXICON:

| Javanese | prokem | meaning | |---|---|---| | mangan | kahath | makan | | ombe | ngoce | minum | | turu | yugu | tidur | | lunga | puha | pergi | | mas | kas | mas | | bapak | calam | bapak | | wedok | dhenyom | perempuan | | enak | ngetham | enak | | apik | ngalim | bagus | | iso | ngijo | bisa | | ireng | ngigeng | hitam | | rokok | gomom | rokok | | kopi | moli | kopi | | sik | jim | dulu, sebentar | | rak | gam | tidak | | ono | ngotho | ada | | sikat | jimat | sikat, ambil | | jalan | sapath | jalan | | loro | pogo | dua | | seket | jemet | lima puluh | | sepuluh | jelupuh | sepuluh |

rak and ono are separate entries that combine as a phrase: rak ono → gam ngotho, "tidak ada".

This code appeared among the gali of Semarang in the 1970s and 80s around Terminal Terboyo, Pasar Johar and Pelabuhan Tanjung Emas, and spread into the city's everyday slang. The rules follow Samidjan (2013) as quoted in Khoiriyah (2018), Bahasa Prokem Semarang atau Basa Walikan dalam Komunikasi di Kota Semarang, skripsi UNNES.

Walikan Jogja

The same twenty letters, paired in parallel instead of reversed: h↔p, n↔dh, c↔j, r↔y, k↔ny. matamu becomes dagadu, mangan becomes daladh.

Yogyakarta's version is the one outsiders are most likely to have met without knowing it: the local clothing label Dagadu Djokdja takes its name from dagadu, which is simply matamu, "your eyes", run through this table.

Walikan Malang (ngalam)

The aksara of a word are read back to front, so Malang becomes Ngalam and sam becomes mas. Digraphs (ng, ny, th, dh) move as single units.

Ngalam is what the city calls itself, and the argot — also known as osob kiwalan — is commonly traced to fighters during the independence struggle, who needed a way of speaking that an infiltrator could not follow. It stayed, and is now ordinary among arek Malang.

Bahasa Unang

Each word becomes two, by the formula U(x) (b)n(c)ng, where x is the last syllable with every vowel changed to a (the consonant immediately before that syllable joins x), b is the word without its last syllable, and c is the original vowel. A one-syllable word has no b. The suffix -nya is detached first and reattached at the end.

| Indonesian | Unang | |---|---| | hancur | uncar hanung | | lari | ura laning | | siapa | upa sianang | | sebentar | untar sebenang | | bel | ubal neng | | sepedanya | uda sepenangnya |

Decoding is pattern-based, not a lookup, so it can occasionally match and transform an ordinary two-word phrase that was never Unang at all — for example unang.decode('untuk menang') returns 'mentak'.

Round-trip fidelity

Decoding recovers the original for every attested word and for the vast majority of formula-generated ones. Four cases are lossy by construction and are asserted as known exceptions in the test suite:

• Semarang inserts an h between adjacent vowels, and it cannot be told apart from a real one afterwards: amalia → ngakapinga → amaliha.
• Semarang's nasal collapse discards the nasal: ngombe → hoce → ngobe.
• In hanacaraka an initial a and an initial ha are the same aksara, so decoding cannot distinguish them.
• A word ending in y or j encodes to one ending in t or s (y→t, j→s under the Semarang cipher), and the final-consonant rule then refuses to swap either back, since both are in KEEP_FINAL: boy → cot → bot.

Credits

Walikan Jogja and Malang are produced by libil (MIT, © 2014 Didiet Noor), wrapped here so output stays identical to the reference implementation. The decode direction, which libil does not implement, is this package's own.

MIT