I with Accent: í ì î ï ī, the Lost Dot and the Turkish Trap

Why does an i with an accent lose its dot?

The accent takes the dot's place. The Unicode Standard (opens in a new tab) says a dotted i followed by a common mark above "loses the dot in rendering". So an i with accent mark shows one mark, not two. Letters that behave this way carry the Soft_Dotted property in PropList.txt (opens in a new tab). Plain i and j head that list.

The dot is not stored anywhere. UnicodeData.txt (opens in a new tab) breaks the acute i, í, into a plain i, U+0069, and a combining acute, U+0301. Strip the accent and the dot comes back.

What you see	How it is stored	What it breaks down into	Dot on screen
í	U+00ED	i + U+0301 combining acute	No, the acute replaces it
ï	U+00EF	i + U+0308 combining diaeresis	No, the two dots replace it
ı́	U+0131 + U+0301, no single character	dotless ı + acute	No, but this is not í
i̇́	U+0069 + U+0307 + U+0301	i + dot above + acute	Yes, the Lithuanian dictionary form

Two rows look like í and are not. The core spec warns that i plus a diaeresis is not ı plus a diaeresis. Normalization keeps them apart as well. We checked in Node.js v26, with ICU 78.3 and CLDR 48 data. There i + U+0301 composes into í, and ı + U+0301 stays as two code points.

The last row is the exception. The same chapter says some Baltic dictionaries keep the dot under a top accent. SpecialCasing.txt (opens in a new tab) builds this into Lithuanian lowercasing: Í becomes i̇́ and Ì becomes i̇̀, each with an explicit dot.

What happens to i and İ when Turkish text changes case?

Turkish and Azerbaijani split the letter in two. Dotted i pairs with capital İ, and dotless ı pairs with capital I. Wiktionary (opens in a new tab) gives ı the sound /ɯ/ and lists it just before i in the Turkish alphabet. The core spec (opens in a new tab) calls the resulting case mapping "language-specific", and SpecialCasing.txt keeps separate rules for tr and az.

Here is what JavaScript returns in Node.js v26 under default and Turkish rules.

Input	Lowercase, default	Lowercase, Turkish	Uppercase, default	Uppercase, Turkish
I	i	ı	I	I
i	i	i	I	İ
İ	i + U+0307, two code points	i	İ	İ
ı	ı	ı	I	I
IZMIR	izmir	ızmır	IZMIR	IZMIR
Diyarbakır	diyarbakır	diyarbakır	DIYARBAKIR	DİYARBAKIR

Each rule set breaks a different word. Typed in capitals as IZMIR, the city lowercases to ızmır under Turkish rules. Every capital I there is dotless. An English uppercase turns Diyarbakır into DIYARBAKIR and loses its dotted i. Both real spellings, İzmir and Diyarbakır, are in AccentPad's place facts.

The default lowercase of İ keeps the dot as a separate mark. CaseFolding.txt (opens in a new tab) agrees: full folding turns İ into i + U+0307, and only the Turkic entry gives a plain i. Code that lowercases and then slices by index counts one character too many.

Why can a Turkish search miss Ísland or BENOÎT?

Because Turkish rules file every accented capital I under dotless I. The CLDR Turkish collation chart (opens in a new tab) lists Í, Ì, Î, Ï and Ī right after ı and I. Their lowercase forms come after i and İ. So under Turkish rules, Í and í stop being one letter in two cases.

We ran a search in Node.js v26 that ignores case and accents. It used Turkish rules, then English ones. The words come from AccentPad's facts.

Search for	Text	Turkish rules	English rules
ísland	ÍSLAND	not found	found
benoît	BENOÎT	not found	found
naïve	NAÏVE	not found	found
izmir	İZMİR	found	found
izmir	IZMIR	not found	found
diyarbakir	Diyarbakır	not found	not found

For Turkish text these results are right: read the Turkish way, IZMIR is not the city. They go wrong when a site applies the visitor's locale to French or Icelandic text. Compare with the rules of the text's language. Dotless ı is a different base letter under both rule sets. So add the plain spelling as a search term for Turkish names.

When does Romanian write î instead of â?

At the edges of a word. Romanian writes one vowel, /ɨ/, with two letters. Wiktionary's usage note (opens in a new tab) puts î at the start and the end of a word and â inside it. A prefix does not move the letter. The IPA in the table comes from the entries for înger (opens in a new tab), coborî (opens in a new tab) and român (opens in a new tab); neînțeles (opens in a new tab) shows the prefix case.

Word	Meaning	Where the vowel sits	Letter	Wiktionary IPA
înger	angel	first letter	î	/ˈɨn.d͡ʒer/
coborî	to descend	last letter	î	/ko.boˈrɨ/
român	Romanian	inside the word	â	/roˈmɨn/
neînțeles	misunderstood	after the prefix ne-	î

Older texts break the rule: Wiktionary lists romîn as an obsolete form of român. In a dictionary, î is a letter, not an accented i. The CLDR Romanian collation (opens in a new tab) has its own rule for it, and a Romanian sort puts îmi and înger after inel.

Can French drop the circumflex from î?

Yes, in most words, since the 1990 spelling reform. The Académie française (opens in a new tab) says the circumflex on i and u becomes optional. It stays on verb endings and where it keeps two words apart. The Quebec language office, the OQLF (opens in a new tab), accepts both spellings. It keeps the accent in passé simple and subjunctive endings. It also stays in forms of croître that would look like croire.

The words below are the most frequent French î words in AccentPad's data, with their spellings from the OQLF list.

Traditional	Reformed (OQLF)	What the circumflex records
île	ile	a lost s: the word was once isle
maître	maitre	not a sound: WikiPron reads it /mɛtʁ/
dîner	diner	not a sound: WikiPron reads it /dine/
boîte	boite
connaître, il connaît	connaitre, il connait
il paraît	il parait
s'il vous plaît	s'il vous plait

This i accent mark once had a job. Wiktionary (opens in a new tab) says it shows that an s beside it has gone, as in île from isle. Both spellings are correct. Pick one and keep it across a document.

What does the mark on an accented i tell the reader?

It depends on the language. Spanish uses the acute for a stressed i. Italian puts the grave on a stressed final i (opens in a new tab). Elsewhere the mark makes a long vowel with its own letter. Wiktionary (opens in a new tab) lists í in the Czech, Slovak, Hungarian and Icelandic alphabets, with a long [iː] in the last three. Latvian ī (opens in a new tab) is [iː] too.

The sort column shows where Íris lands beside Ida and Ivan in a sort under each language's rules.

Language	Letter	What the mark does	Sorted	Examples in AccentPad's facts
Spanish	í	marks the stressed i, /i/	with i	sí, aquí, día, Joaquín
Italian	ì	marks the i of a stressed final syllable	with i	sì, così, lunedì, Forlì
Czech	í	a letter of its own, after i	with i	Jiří, Vladimír, Antonín
Hungarian	í	a letter of its own, long [iː]	with i	Tímea
Icelandic	í	a letter of its own, [iː]	after every i	Vigdís
Latvian	ī	long [iː]	with i	Kristīne, Arvīds
French	ï	the vowel is said on its own	with i	maïs, égoïste, Anaïs

Icelandic is the one that moves the name. Its CLDR collation (opens in a new tab) files í as a separate letter, so Íris comes after Ivan. In Icelandic, í is also a whole word, the preposition in.

Which words change when the i loses its accent?

Short ones, mostly. Here the accent on i is what tells two words apart. Drop it and you write a different word. The meanings come from Wiktionary's entries for sí (opens in a new tab), si (opens in a new tab), mí (opens in a new tab), mi (opens in a new tab), sì (opens in a new tab), lì (opens in a new tab), li (opens in a new tab), maïs (opens in a new tab) and mais (opens in a new tab).

With the accent	Meaning	Without	Meaning
sí (Spanish)	yes	si	if
mí (Spanish)	me, after a preposition	mi	my
sì (Italian)	yes	si	oneself
lì (Italian)	there	li	them
maïs (French)	corn, /ma.is/	mais	but, /mɛ/

To write the i in Spanish with accent, use í. Both sí and mí rank among the most frequent Spanish words in AccentPad's subtitle data. The French row follows the Académie's rule: the tréma stops two letters from being read as one sound. English keeps that ï in naïve (opens in a new tab), a French borrowing also spelled naive.