Laravel localization in Indonesian Bahasa Indonesia

Locale codes

Set the base locale in config/app.php. For regional variants, Indonesian commonly uses: id_ID

// config/app.php
'locale' => 'id',
'fallback_locale' => 'en',

// Or switch at runtime
App::setLocale('id');

Plural rules: what Indonesian actually needs

CLDR defines 1 cardinal category for Indonesian. The sample numbers below were computed by ICU for this exact locale:

Laravel's trans_choice() maps numbers to 1 pipe-separated form for this locale:

// lang/id/messages.php
'items' => ':count'

// Usage
trans_choice('messages.items', $count, ['count' => $count]);

This language has no grammatical plural, so a single form covers every number.

Try any number live in the pluralization tester.

Localized dates with Carbon

Real output for Indonesian (id locale), generated by Carbon for March 21, 2026:

$date = now()->locale('id');

$date->translatedFormat('l, j F Y');
// "Sabtu, 21 Maret 2026"

$date->isoFormat('LLLL');
// "Sabtu, 21 Maret 2026 pukul 14.30"

$date->isoFormat('L');
// "21/03/2026"

$date->subDays(3)->diffForHumans();
// "3 bulan yang lalu"

What to watch out for in Indonesian

• No plural inflection (plurality is optional reduplication: buku-buku) and no verb tense conjugation.
• No grammatical gender; the polite second person Anda is capitalized and standard in UI copy.
• Decimal comma with dot as thousands separator (1.234,56).

Grammatically the simplest language you will localise

Indonesian removes most of the difficulties that make other locales expensive. There is no grammatical gender, no verb conjugation for person or number, no grammatical tense, no cases, no articles, and no obligatory plural. Nouns and verbs keep one form regardless of context.

For software this means placeholders behave predictably. A noun dropped into a sentence needs no case ending, no article agreement and no gender concord, so the class of bugs that dominates German, Czech, Finnish and Korean localisation simply does not arise.

That does not make sentence assembly a good idea — word order still matters, and a fragment-based template still produces stilted copy — but it does mean Indonesian is unusually forgiving, and a reasonable first non-European locale to add. It also means machine translation performs better here than in most languages, since there is far less grammatical agreement for a model to get wrong — though vocabulary and register still need a human pass before shipping.

The language is written in the Latin alphabet with no diacritics, so encoding, sorting, case conversion and search all behave exactly as they do in English. There is no collation configuration to get right and no multi-byte trap in case transformations.

No plurals, and reduplication instead

Indonesian has a single CLDR plural category, so trans_choice() takes one form that serves every number. Plurality is either left to context or marked by repeating the noun — buku is book or books, and buku-buku emphasises plurality explicitly.

One form for every count

// lang/id/messages.php
'files' => ':count berkas',

// 0  -> 0 berkas
// 1  -> 1 berkas
// 42 -> 42 berkas

Crucially, reduplication is not used when a number is present. Tiga buku-buku is wrong; it is tiga buku. A translator working from an English source with singular and plural forms may supply a reduplicated noun for the second, producing text that is incorrect wherever a count appears.

Reduplication also has meanings other than plurality — it can indicate variety, repetition or intensity — so it is not a mechanical plural marker and should never be generated in code.

The absence of plural machinery has a quiet benefit worth noting: counting strings need only one translation, so the combinatorial growth that makes Arabic or Czech expensive simply does not occur. A product with several hundred counting messages saves a substantial amount of translation work here.

Affixation changes words considerably

What Indonesian lacks in inflection it makes up for in derivation. Prefixes and suffixes combine with a root to build related words, and the resulting forms can look quite different from the root they came from.

Note that the prefix can alter the first letter of the root: me- plus simpan gives menyimpan, not mesimpan. These sound-change rules are regular but not obvious, and they mean prefix-based search matching will fail to connect related words.

For search this is the main practical concern. A user typing simpan expects to find menyimpan and penyimpanan, and substring matching will not connect them because the stem is altered. Elasticsearch and most search engines ship an Indonesian stemmer; enabling it makes a substantial difference.

Affixation also drives text expansion. Indonesian runs roughly twenty to thirty per cent longer than English, largely because a single English word often becomes an affixed form plus a helper word. Buttons and labels need room accordingly.

Formality, and how much English to keep

Indonesian has several pronouns for addressing the user, distinguished by formality rather than by grammar. The choice does not change verb forms, so it is a matter of tone rather than of structure — which makes it considerably cheaper to revisit than the equivalent decision in German or Korean.

Anda is the safe default and is conventionally capitalised in the middle of a sentence. Consumer products aimed at younger users increasingly prefer kamu, which reads as warm rather than casual, but Anda is never wrong.

As in Hindi and Dutch, deciding how much English to keep matters more than it first appears. Indonesian technical vocabulary borrows heavily, and some official Indonesian coinages are less familiar to users than the English originals — unduh for download and gawai for device are correct but not universally used, while email, upload and login are entirely standard.

Give translators a glossary settling these cases. Mixing a Sanskritised official term on one screen with the borrowed English word on the next is the most common quality complaint about Indonesian interfaces, and it is entirely preventable.

Formatting, and the Malay question

Indonesian and Malay are closely related and mutually intelligible in large part, which regularly tempts teams into shipping one to both markets. Vocabulary and spelling differ enough that users notice, and the two have separate locale codes for good reason.

Some of these are not merely unfamiliar but actively confusing: kereta means car in Malay and train in Indonesian. Treat the two as separate locales rather than aliasing one to the other, and keep their translation memories separate as well, since a shared memory will propagate each variety of vocabulary into the other over time.

Numbers use the European convention: dots for thousands and a comma for decimals. That creates the familiar numeric-input hazard, where floatval('1234,56') returns 1234. The rupiah has no subunit in practice, so amounts carry no decimals, and values are large — a modest price runs to six or seven digits, so currency fields need more width than a dollar design assumes.

Month names are capitalised in Indonesian, unlike most European languages, so 21 Maret 2026 is correct. This is one of the few places where the English convention happens to match. Day names are capitalised too, and dates are written day-first, so a date such as Sabtu, 21 Maret 2026 mixes an English-looking capitalisation with a European ordering — a combination that trips up formatters assembled by hand rather than driven by the locale.

Getting Indonesian into the application

Indonesian asks very little of your infrastructure, which is worth stating plainly because the effort saved can go into the things that do matter. The Latin alphabet without diacritics means default collation sorts correctly, case conversion is safe with ordinary string functions, and there is no encoding consideration beyond using UTF-8 as you would anyway.

Set the full locale nonetheless. Number formatting, currency and date output all depend on it, and the European-style separators are the one place Indonesian diverges from English conventions in a way that produces real bugs. Parse numeric input explicitly rather than relying on floatval(), which will silently discard everything after a decimal comma.

Invest the saved effort in search. Affixation is where Indonesian genuinely differs from English, and a LIKE '%term%' query will miss most morphological relatives of the word a user typed. Enabling the Indonesian stemmer in whichever search engine you use is a small configuration change with a disproportionate effect on how usable the product feels.

Invest the rest in the glossary. Because Indonesian borrows English technical vocabulary inconsistently across the industry, terminology drift is the dominant quality problem rather than grammar. A short list of settled terms — which words stay in English, which use the Indonesian coinage — given to every translator prevents an interface that speaks two dialects of its own vocabulary.

Install the Laravel-Lang Indonesian files for framework strings. There is less linguistic complexity to get right than in most locales, but the messages are already translated and reviewed, and writing your own guarantees a second vocabulary to keep consistent with the first.

What ships broken most often

• Reduplicated nouns after numbers. 3 buku-buku instead of 3 buku.
• Malay shipped as Indonesian. Or the reverse — mutually intelligible is not the same as interchangeable.
• Inconsistent borrowing. Official Indonesian coinages on one screen and English on the next.
• Search that cannot match affixed forms. No stemmer, so simpan misses menyimpan.
• Rupiah with decimals. Rp1.234,00 rather than Rp1.234, and fields too narrow for the digits.
• Decimal comma truncating numeric input. The usual silent loss.
• Truncated labels. Affixed forms overflowing buttons sized for English.

Indonesian is one of the cheapest locales to get right and one of the largest markets by user count. The grammar poses almost none of the structural problems that dominate European and East Asian localisation, so the effort goes into vocabulary consistency and layout rather than into string architecture.

It is also a good language to translate first when you want to validate that your localisation pipeline works. Because Indonesian exercises the plumbing — locale resolution, formatting, fallbacks, coverage checks — without adding grammatical complexity on top, a problem that appears in Indonesian is almost certainly a problem in your infrastructure rather than in the translation.

Framework strings already translated

Laravel's own validation, auth and pagination strings are maintained in Indonesian by the open-source Laravel-Lang project (MIT). Install them, then manage your app's own strings live:

composer require laravel-lang/common --dev
php artisan lang:add id
php artisan lang:update