By accent mark
An accent mark is a small sign added to a base letter. Each mark below shows its combining code point, a few of the letters that carry it and a link to the full page with the dead keys for Windows, Mac and Linux.
A stroke rising to the right, the most widely used Latin diacritic of all. It marks stress in Spanish and Portuguese, a closed e in French, vowel length in Czech, Slovak, Hungarian and Irish, and the sắc tone in Vietnamese.
A stroke falling to the right, the mirror image of the acute. It separates French homophones such as à and a, marks the open vowel in Italian and Catalan, indicates crasis in Portuguese, and carries the huyền tone in Vietnamese. It is not the backtick used in code, even though both live on the same key.
A little roof over the vowel. In French it usually marks a lost s (hôtel from hostel, forêt from forest); in Portuguese it marks a closed stressed vowel against the acute's open one; Welsh calls it to bach, Esperanto ĉapelo and Afrikaans kappie, and Turkish uses it optionally to tell kâr from kar.
A wavy line over the letter. Spanish uses it on the single letter ñ, Portuguese on the nasal vowels ã and õ, Estonian on õ (a distinct vowel, not a nasal), and Vietnamese for the ngã tone. Careful with the Spanish terminology: in Spanish the word tilde normally means the acute accent, and this mark alone is the virgulilla.
Two dots above the vowel, one glyph doing two different jobs. As a German umlaut it marks a changed vowel (Mann to Männer, Buch to Bücher); as a diaeresis it tells you to pronounce two vowels separately (naïf, Noël, coëfficiënt, pingüino). Unicode encodes both identically, so only the language decides which name is right. and ae, oe, ue are acceptable substitutes only for the German umlaut.
A small hook hanging under the letter, curling to the left. On the c it forces the /s/ sound before a, o and u: français, garçon, coração, força. Turkish uses it on ç and ş, Catalan calls it ce trencada, and the Latvian letters ģ ķ ļ ņ are named with cedilla in Unicode although they are drawn with a comma.
A small angular v over the letter. the mirror image of the circumflex, and not to be confused with the rounded breve. It gives č, š and ž across Czech, Slovak, Croatian, Slovene, Latvian and Lithuanian, plus the uniquely Czech ř. On ď, ť and ľ it is printed as an apostrophe-like stroke for want of vertical space.
A small rounded cup over the letter. Romanian calls it căciulă and uses it for the schwa ă; Turkish ğ (yumuşak ge) only lengthens the preceding vowel and never starts a word; Vietnamese ă and Esperanto ŭ round out the set. Its curve is what separates it from the angular caron.
A straight bar over the vowel marking length. Latvian calls it garumzīme and uses ā ē ī ū, Lithuanian has only ū, and outside Europe it is the Māori tohutō and the Hawaiian kahakō. It is a letterform, not an overline or an underscore.
A little tail hanging under the vowel and curling to the right. the Polish word ogonek means exactly that. Polish ą and ę are nasal vowels (ą is nasal o, not nasal a), while Lithuanian ą ę į ų, called nosinė, were once nasal and now simply mark length. Do not confuse it with the cedilla, which curls the other way and sits under consonants.
A small circle above the letter. Danish, Norwegian and Swedish å is a full letter of the alphabet, sorted after z, and replaced the old spelling aa in 1917 and 1948; Czech ů, called kroužek, writes a long u inside words such as dům and stůl. The ring is not the degree sign and not the masculine ordinal º.
A line drawn straight through the letter. It gives Danish and Norwegian ø, Polish ł (pronounced like English w), Croatian, Serbian, Bosnian and Vietnamese đ, and Maltese ħ. Unlike the accents above letters, these are all separate letters of their alphabets and the precomposed characters do not decompose into base plus mark.
A single dot placed over the letter. Polish calls it kropka and uses it for ż, distinct from ź; Lithuanian ė is a long e; Maltese has ċ, ġ and ż; Turkish keeps the dot on the capital İ, which is a different letter from plain I. Old Irish type used it for lenition, where modern spelling writes bh, ch, dh and so on.
Two parallel acute strokes, used in the Latin script by Hungarian alone, on ő and ű. They are the long counterparts of ö and ü. kő, tűz, gyűrű. and Hungarian calls the mark hosszú ékezet or kettős ékezet; typographers know it as the hungarumlaut. Writing ö or ü instead is a spelling mistake, not a variant.
A small hook attached to the top right of the letter, used by Vietnamese alone to make ơ and ư. Vietnamese calls it dấu móc or râu, and the two letters are independent vowels of the alphabet, quite separate from o and u. cơm, mới, nước, người. They can still carry any of the five tone marks on top.
A comma sitting below the letter, detached from it. this is the mark the Romanian Academy prescribes for ș and ț, code points U+0218 to U+021B. The cedilla forms ş and ţ that older Windows versions produced are wrong for Romanian; Unicode states plainly that the cedilla is preferred in Turkish and the comma below in Romanian. Latvian ģ ķ ļ ņ are drawn with a comma too, but they use the cedilla code points.
A small question-mark-like hook above the vowel, used by Vietnamese alone for the hỏi tone. mả, hỏi, nhỏ, biển, phở. It is one of the five Vietnamese tone marks and can sit on top of a vowel that already carries a breve, circumflex or horn, which is why the alphabet has so many precomposed letters.
A single dot under the vowel, the Vietnamese nặng tone. mạ, mẹ, học, một, bạn, lạnh. Outside Vietnamese the same mark appears in Yoruba, Igbo and in IAST transliteration of Sanskrit, where it marks retroflex consonants such as ṭ, ḍ, ṇ and ṣ.
ligature (æ, œ, ß)
6 letters
Two letters fused into one shape. There is no combining mark for it, so the ligature exists only as a character of its own: French œ is mandatory in cœur, sœur and œuf, æ appears in ex æquo, and German ß comes from the long s plus s or z. In Danish, Norwegian and Icelandic, æ is not a ligature at all any more but a full letter of the alphabet.