Dotless I (ı): a Turkish Vowel Whose Capital Is the Plain I

Is the dotless i a letter of its own or an i that lost its dot?

The dotless i is a letter of its own. People search for it as the i without dot, but the Turkish i comes in two forms, dotted and dotless, and each one has its own capital. Wiktionary (opens in a new tab) writes the sound of ı as /ɯ/. Typing an i in its place gives a different vowel, and often a different word.

Unicode stores it the same way. UnicodeData.txt (opens in a new tab) gives ı, U+0131, no decomposition: it is not built from an i and a mark. Its uppercase is U+0049, the very I that English uses. The dotted capital İ is the one made of parts, an I plus U+0307, the combining dot above.

Letter	Code point	Partner under Turkish rules	Partner under default rules	Stored as
ı	U+0131	I	I when uppercased	one character, no parts
I	U+0049	ı	i	one character
i	U+0069	İ	I	one character
İ	U+0130	i	i plus U+0307 when lowercased	I plus U+0307

The default column comes from UnicodeData.txt. The Turkish column comes from SpecialCasing.txt (opens in a new tab), which keeps separate tr and az lines for these four letters. The trouble sits in the second row. Turkish and English share one capital I but give it different small letters. CaseFolding.txt (opens in a new tab) marks the Turkic mapping of I with its own status, T, for that reason.

Which Turkish words change meaning when ı becomes i?

Short everyday words do. People who type Turkish on a keyboard without ı often write i instead, and the reader gets another word. The meanings and transcriptions below come from Wiktionary's entries for dış (opens in a new tab), diş (opens in a new tab), kır (opens in a new tab), kir (opens in a new tab), kıl (opens in a new tab) and kil (opens in a new tab).

With ı	Meaning	Said	With i	Meaning	Said
dış	outside	[dɯʃ]	diş	tooth	/ˈdiʃ/
kır	countryside, or grey	/ˈkɯɾ/	kir	dirt	[ciɾ]
kıl	body hair		kil	clay	/ˈcil/

The consonant moves too. Look at the k in each pair. Wiktionary writes [c] before i, in kir and kil, and [k] before ı, in kır. So the dot changes how the whole syllable is said, not only the vowel.

Why do Turkish endings switch between ı, i, u and ü?

A Turkish suffix copies the vowels of the word it joins, and the dotless i is one of the vowels it can copy. The Turkish Language Association, TDK, splits the vowels into back ones, a, ı, o and u, and front ones, e, i, ö and ü, in its rule on major vowel harmony (opens in a new tab).

Its rule on minor vowel harmony (opens in a new tab) adds flat and round, and ends with a chart of what may follow each vowel.

Read from that chart, the narrow vowel of an ending has four forms. The dotless i is the one after a and after ı.

Last vowel of the word	Kind, in TDK's terms	Narrow vowel that follows	TDK's example	Question word, from TDK
a	back, flat	ı	alır
ı	back, flat	ı	kılıç	Kaldı mı?
o	back, round	u	omuz
u	back, round	u	uzun	Olur mu?
e	front, flat	i	gelir	Sen de mi geldin?
i	front, flat	i	ilik
ö	front, round	ü	ölçü
ü	front, round	ü	ütü	İnsanlık öldü mü?

The last column is from TDK's page on the question particle (opens in a new tab). The particle is written as a separate word, yet it still takes its vowel from the word before it. Wiktionary's entry for mı (opens in a new tab) gives Arabanı sattın mı?, did you sell your car?

This is why ı is so frequent. It is not tied to a few native words. It comes back in the endings of any word whose last vowel is a or ı. Write kaldi mi for kaldı mı and both words break.

Why does TITLE become tıtle on a computer set to Turkish?

Some lowercase functions read the language of the machine they run on. Oracle's Java documentation (opens in a new tab) gives this case itself: "TITLE".toLowerCase() in a Turkish locale returns tıtle, with a dotless ı. Uppercasing title gives TİTLE. Oracle's advice for identifiers, protocol keys and HTML tags is Locale.ROOT.

Microsoft calls this the Turkish-I problem (opens in a new tab). In its example, a case-blind test of whether Bill starts with BILL returns false under Turkish culture. A check that blocks FILE: links can be slipped past the same way. Its page on case changes (opens in a new tab) adds that ToLower() and ToUpper() with no argument use the current culture.

Pick the call by asking whose text it is. A key, a tag or a file extension needs a call that never changes. A Turkish name shown to people needs Turkish rules on purpose.

Language	Follows the machine's locale	Same on every machine	Turkish rules on purpose
Java	toLowerCase(), toUpperCase()	toLowerCase(Locale.ROOT)	toLowerCase with a Turkish Locale
.NET	ToLower(), ToUpper()	ToLowerInvariant(), ToUpperInvariant()	ToLower with a tr-TR CultureInfo
JavaScript	toLocaleLowerCase() with no argument	toLowerCase(), toUpperCase()	toLocaleLowerCase("tr")
Python	none of the str methods	str.lower(), str.upper(), str.casefold()	none of the str methods

The JavaScript row comes from MDN. Its pages for toLocaleLowerCase (opens in a new tab) and toLocaleUpperCase (opens in a new tab) turn istanbul into İSTANBUL with "TR". The Python row comes from the string methods (opens in a new tab), which take no locale at all.

What happens to the dotless i in slugs, file names and search boxes?

The dotless i stays where you expect it to vanish. A common way to strip accents is to break each letter into parts and drop the marks. That turns İ into I and ş into s, because both are stored as a base letter plus a mark. The dotless ı has no parts, so nothing comes off.

Run in Node.js v26.0.0, with ICU 78.3 and CLDR 48, on 2026-10-10, this method turned Şanlıurfa into Sanlıurfa and Barış into Barıs. An ASCII slug or file name needs ı → i as a rule of its own.

Search fails in a different place. In the same Node.js build we compared each pair below with a collator that ignores case and accents. We ran it with Turkish rules, then English ones. The names come from AccentPad's facts.

Typed into the search box	Text on the page	Turkish rules	English rules
pınar	PINAR	match	no match
pinar	PINAR	no match	match
pinar	Pınar	no match	no match
balıkesir	BALIKESİR	match	no match
nazlı	NAZLI	match	no match

English rules give the capital I to i, so PINAR answers only the spelling without ı. Turkish rules give it to ı. Neither set treats ı and i as one letter, so the third row fails under both. A search box for Turkish names should index both spellings, or fold ı to i on both sides first.

Sorting splits the same way. The same build put ılık, ırmak and ısı before iç and inek under Turkish rules. English rules put them after every i word. The CLDR Turkish collation (opens in a new tab) files ı as its own letter just before i.