Javanese is not a hard language to pronounce, but it has three contrasts an English speaker will not hear at first and will therefore not produce: a dental versus retroflex stop series, a stiff versus slack voice distinction where English has voicing, and a vowel rule that turns most word-final -a into [ɔ]. Get those three and you will be understood everywhere; miss them and you will be understood slowly.

This is the pronunciation guide. The grammar, vocabulary in context and the CEFR-graded practice readers for Javanese live in its companion language guide, which shares this deck — see the link at the end. The two are deliberately not duplicated: this page is about the sound system, and it is graded so you can work through it alongside the language guide's CEFR bands.

1. Six vowels, and why they move

Javanese is usually analysed with six vowel phonemes. The complication is not the inventory but the allophony: several vowels change quality predictably with syllable structure, and the orthography does not always show it.¹

The vowels and their main allophones (Surakarta standard)

The single most audible rule

In the standard dialect of Surakarta, /a/ is pronounced [ɔ] in a word-final open syllable — and in any open penultimate syllable that precedes such an [ɔ].¹ The colour spreads backwards one syllable. This is the rule that makes Javanese sound Javanese.

Final -a is [ɔ], and it drags the penult with it

Some older spellings write this vowel å to make the rule visible. Modern spelling does not, so you have to apply it yourself. A learner who reads final -a as English “ah” is audibly a beginner from the first word.

Drill. Say these deck words with the rule applied, then check: sega [səgɔ] ‘rice’, lima [limɔ] ‘five’, mata [mɔtɔ] ‘eye’, tuwa [tuwɔ] ‘old’, lunga [luŋɔ] ‘to go away’, apa [ɔpɔ] ‘what’.

2. Dental versus retroflex

Apart from Madurese, Javanese is the only language of western Indonesia to contrast a dental and a retroflex stop series.¹ In the modern Latin orthography they are written t/th and d/dh.

The dental/retroflex contrast

The trap. th and dh are not the English sounds of thin and this. There is no fricative anywhere in Javanese th. English t and d are alveolar — made between the two Javanese positions — so an untrained English speaker produces something that is neither, and the contrast collapses.

Minimal-pair practice. Work these until the difference is automatic. Put your tongue on your teeth for the plain series, and curl it back for the h series.

Dental versus retroflex — say each pair aloud

3. The “voiced” stops are not voiced

This is the subtlest thing on the page and the one that most changes how you sound. The Javanese phonemes written b d dh j g are not phonetically voiced. They are voiceless, with breathy voice on the following vowel. The relevant contrast is described as stiff voice versus slack voice rather than as voicing.¹

Stiff versus slack voice

Kenstowicz's acoustic study is the reference treatment of exactly this contrast, and it is why this guide will not tell you to “pronounce b as in English”.² Practically: aim the breathiness at the vowel after the consonant. bapak ‘father’ is not English “bapak” — the a is where the voice quality lives.

English speakers also need to stop aspirating the stiff series. English initial /p t k/ come with a puff of air; Javanese ones do not. Hold a sheet of paper in front of your mouth: it should not move on pitu ‘seven’ or kutha ‘city’.

4. Syllables, clusters and stress

A Javanese syllable is (C)(S)V(C), where S is a sonorant — /j r l w/ or a nasal. About 85% of Javanese morphemes are disyllabic, and words longer than three syllables are broken into disyllabic groups for pronunciation.¹

Syllable shapes

Consonant clusters between vowels are generally a nasal plus a homorganic stop — [mp], [mb], [ɲtʃ] and so on — or [ŋs], optionally with /l r j/ added at the end. A syllable before such a cluster counts as open for the vowel rule, which is why tampa ‘accept’ is [tɔmpɔ] while tanpa ‘without’ is [tanpɔ].¹ That pair is worth memorising: it shows the vowel rule and the cluster rule interacting.

Stress is the good news. Views differ — some describe weak stress on the penultimate syllable unless it contains a schwa, in which case it moves to the final; others describe word-final stress. Either way Javanese stress is not contrastive, so there is no stress pattern you can get wrong at the level of meaning.¹

5. From spelling to sound

The Latin orthography descends from a Dutch-based system introduced in 1926 and revised in 1972–73; this guide's deck follows the 2006 Pedoman Umum spelling standard.¹³ Most letters are predictable. These are the ones that are not.

Letters that need care

Word-initial ng- deserves a sentence of its own. English has /ŋ/ only at the end of a syllable, so ngombé ‘to drink’ and ngerti ‘to understand’ feel impossible at first. Practise by saying “singer”, holding the /ŋ/, then dropping the “si-”.

6. What English speakers get wrong

• Reading final -a as [a]. It is [ɔ], and it colours the syllable before it.
• Pronouncing th and dh as English “th”. They are retroflex stops. padha contains no fricative.
• Aspirating p t k. Javanese stiff-voice stops have no puff of air.
• Voicing b d g j like English. They are slack-voiced: the breathiness belongs to the following vowel.
• Reading bare e as /e/. It is schwa; only é and è are mid front vowels.
• Avoiding initial ng-. It is a normal onset in Javanese.
• Stressing heavily. Javanese stress is weak and non-contrastive; English-style stress-timing distorts the vowels.
• Applying the [ɔ] rule to Banyumasan. The western ngapak dialects say a real [a] — where Solo says åpå, Banyumas says apa. Neither is wrong.

7. Sources

• Wikipedia — Javanese language — Vowel inventory and allophony, the final-a rule, the dental/retroflex contrast, stiff vs slack voice, the syllable canon, stress, and orthography history.
• Kenstowicz, “Phonetic Correlates of the Javanese Voicing Contrast” (2021) — NUSA 70: 1–37. The acoustic study behind the stiff-voice / slack-voice description.
• Javanese for English speakers — the companion language guide — Grammar, the ngoko/krama register system, aksara Jawa, CEFR-graded readers, and the same 667-word deck.