Croatian Alphabet: dž, lj and nj Count as One Letter Each

Why does Croatian count dž, lj and nj as one letter each?

Because each stands for one sound, and the alphabet gives it a place of its own. The Slova page (opens in a new tab) of pravopis.hr, run by the Institute of Croatian Language, lists the letters in their fixed order: dž after d, lj after l and nj after n.

The same page splits the script into single letters and the digraphs dž, lj and nj. The Hrvatska enciklopedija calls dž (opens in a new tab) the seventh letter of the Croatian alphabet, lj (opens in a new tab) the seventeenth and nj (opens in a new tab) the twentieth.

A digraph also capitalises in two ways, and pravopis.hr says when to use each one.

Digraph	Its sound	First letter of a sentence	Word in capitals
dž	/d͡ʒ/, as in džep	Dž	DŽ
lj	/ʎ/, as in ljeto	Lj	LJ
nj	/ɲ/, as in konj	Nj	NJ

The sounds are Wiktionary's transcriptions of džep (opens in a new tab), a pocket, ljeto (opens in a new tab), summer, and konj (opens in a new tab), a horse. Each is a single sound in the transcription, not a d plus a ž or an l plus a j.

This changes how you count and how you sort. Ljeto is four letters long. In a Croatian list, every word that starts with plain l comes before the first lj word. We checked it on 2026-10-10 with the Croatian collator in Node.js v26.0.0, which runs ICU 78.3 on CLDR 48 data: lutka sorted before ljeto. English rules put ljeto first, since j comes before u.

When is nj in a Croatian word two letters, not one?

When the n and the j belong to different parts of the word. The Hrvatska enciklopedija's article on dž says the pair can also stand for two sounds, d and ž, and gives nadživjeti (opens in a new tab), to outlive, as the example: the prefix nad- meets the verb živjeti. Wiktionary's entries for injekcija (opens in a new tab) and konjugacija (opens in a new tab) show the same split for nj.

The tell is in the pronunciation. Wiktionary transcribes the real digraph with one symbol and the split pair with two, and its hyphenation breaks the split pair across syllables.

Word	Wiktionary IPA	Syllables	The pair is
konj	/kôɲ/	konj	one letter, nj
konjugacija	/konjuɡǎːt͡sija/	kon‧ju‧ga‧ci‧ja	n and j
injekcija	/injěkt͡sija/	in‧jek‧ci‧ja	n and j
nadživjeti	/nadʒǐːʋjeti/	nad‧ži‧vje‧ti	d and ž

Spelling gives no sign of the difference. The letters on the page are the same, so a reader knows the word, and software has to guess.

Software guesses digraph. In the same 2026-10-10 test, the Croatian collator put konobar before konjugacija and inozemstvo before injekcija. It read each nj as the letter that follows n. Taken letter by letter, j comes before o, so konjugacija and injekcija should come first. The control row is konj, a true digraph: the collator put konobar first, which is right.

For a dictionary or an index that must be exact, check words like these by hand. The collator cannot see where a prefix ends.

Where did Gaj's Croatian letters come from?

From Czech, Polish and older Croatian writers, adopted between 1830 and 1892. The Hrvatska enciklopedija entry on gajica (opens in a new tab), the script named after Ljudevit Gaj, says it was modelled on the spelling of Pavao Ritter Vitezović and on Czech. Mihanović's Horvatska domovina, in the journal Danica in 1835, was the first work printed in it. Each row below covers one or two Croatian alphabet letters, from the encyclopedia's articles on č (opens in a new tab), ć (opens in a new tab), đ (opens in a new tab), š (opens in a new tab), ž (opens in a new tab) and the digraphs.

Letter	Where its form came from	When today's form settled
š, ž	the caron of Czech spelling	in Gaj's Kratka osnova, 1830
č	the Czech caron; Gaj's 1830 c̃ stood for both č and ć	soon after 1830
ć	Polish spelling	about 1835
lj, nj	Gaj proposed one letter with a caron in 1830, then two letters in 1835	from 1838, with a dotted j
dž	first printed in Danica, 1835; Daničić wrote ǵ in 1878	Broz brought dž back in 1892
đ	Gaj used dj and gj from 1835; Daničić crossed the d in 1878	taught in schools from 1892

So the digraphs are a choice, not an accident. Gaj tried single letters with marks first, then went back to two letters. Daničić later tried single letters again for the Academy's dictionary: ǵ for dž, ļ for lj, ń for nj. Only his đ stayed. Since Broz in 1892, dž has been part of the modern Croatian alphabet.

Should you type the one-character lj, or l and then j?

Type l and then j. Unicode does hold single characters for the three digraphs, each in three forms: Dž Lj Nj for the first letter of a word, DŽ LJ NJ for capitals, dž lj nj for small letters. The Unicode names list (opens in a new tab) files them under "Latin digraphs matching Serbian Cyrillic letters", for Gaj's Latin alphabet. It cross-refers lj to the Cyrillic љ, nj to њ and dž to џ. The Latin Extended-B chart (opens in a new tab) prints the same notes.

That header names their job: one Latin character for each Serbian Cyrillic letter. In everyday text they cause trouble. We tested them on 2026-10-10 in Node.js v26.0.0.

Test	ljubav, with U+01C9	ljubav, with l and j
character count	5	6
a search for "lj" finds it	no	yes
after NFKC normalization	ljubav	ljubav

The last row is the way out. NFKC, one of the normalization forms in UAX #15 (opens in a new tab), splits each single-character digraph into its two letters. Run it on pasted text before you search or compare. Text you type as l and then j already has the two characters.

Why does an old Croatian file show Èakovec instead of Čakovec?

The file was saved in the Windows-1250 code page and opened as Windows-1252. Unicode's mapping files for cp1250 (opens in a new tab) and cp1252 (opens in a new tab) show what each byte becomes.

Croatian letter	Byte in cp1250	Same byte read as cp1252
č, Č	E8, C8	è, È
ć, Ć	E6, C6	æ, Æ
đ, Đ	F0, D0	ð, Ð
š, Š	9A, 8A	š, Š
ž, Ž	9E, 8E	ž, Ž

The last two rows are why the damage is easy to miss. Š and ž sit on the same bytes in both code pages, so they survive. A word like šećer (opens in a new tab), sugar, comes out as šeæer: the š is fine and only the ć breaks. We decoded these strings on 2026-10-10 with the cp1250 and cp1252 codecs of Python 3.14. Srećko came out as Sreæko and Đuro as Ðuro.

The same overlap explains the Alt codes in the table above. AccentPad computes them from cp1252, which holds š and ž but not č, ć or đ. So the number pad types two of the Croatian letters with marks, and the other three need another method. To repair a file, reopen it as Windows-1250 and save it as UTF-8. Do not retype the letters by hand.