Can someone explain the sounds all the diacritics make.

[xain1112]

I've always wanted to make a language with diacritics, but I only know the tilde, umlaut, and stress accents.

[Rhetorica]

That, sadly, is completely dependent on the language's phonology. For example, most languages use ¨ as diaresis, not umlaut—its sole purpose is to say "hey, I'm not a diphthong!"

You should probably start with the relevant Wikipedia article if you want an exhaustive list, and then ask for specifics here afterwards if you're still confused.

[hwhatting]

That, sadly, is completely dependent on the language's phonology orthography.

Hope you don't mind. For the rest, good advice.

[Aurora Rossa]

Stress accents (by which I assume you mean acute accent marks) indicate the position of stress, usually on irregularly accented words in languages with otherwise regular accent. The orthography of Spanish illustrates the usage of accent marks particularly well, so you could start by looking at that. Orthographies for tonal languages almost always use the acute accent to indicate high or rising tone on vowels.

Tildes usually indicate nasalization when used on vowels, though some orthographies (most notable the polytonic orthography of ancient Greek) use the tilde for certain contour tones. The tilde also occurs on the letter <ñ> where it indicates a palatal nasal sound somewhat like the "ny" in "canyon".

Umlauts indicate originally back vowels (like /a o u/) which have shifted to the front of the mouth while retaining their original rounding. The classic illustration for this comes from German, although letters with umlauts occur in the orthographies of neighboring languages like Swedish and Finnish. The double-dot diacritic has numerous other uses besides indicating umlaut.

[Nortaneous]

http://en.wikipedia.org/wiki/Acute_accent - stress or sometimes length on vowels, palatalization on consonants
http://en.wikipedia.org/wiki/Grave_accent - quality change (usually lower) on vowels
http://en.wikipedia.org/wiki/Breve - quality change (usually centralized) on vowels, can be used to mark short length on vowels where long is default; g-breve is /ɣ/
http://en.wikipedia.org/wiki/Caron - tone on vowels, postalveolarity/palatalization on vowels
http://en.wikipedia.org/wiki/Cedilla - it depends
http://en.wikipedia.org/wiki/Circumflex - it depends
http://en.wikipedia.org/wiki/Diaeresis_(diacritic) - quality change (usually reversed front/back) on vowels, don't fucking use it on consonants
http://en.wikipedia.org/wiki/Macron - length or tone on vowels
http://en.wikipedia.org/wiki/Ogonek - nasalization on vowels

and rhetorica, i would bet the majority of languages that use the diaeresis use it to mark quality changes

[WeepingElf]

http://en.wikipedia.org/wiki/Caron - tone on vowels, postalveolarity/palatalization on vowels

The second "vowels" must be "consonants", I guess.

[Rhetorica]

That, sadly, is completely dependent on the language's phonology orthography.

Hope you don't mind. For the rest, good advice.

Blarg! That's more accurate. Although I kinda meant phonology, too, in the sense that the orthography has to accommodate the needs of the phonology... but that assumes a perfect orthography, I guess.

[linguoboy]

Umlauts indicate originally back vowels (like /a o u/) which have shifted to the front of the mouth while retaining their original rounding. The classic illustration for this comes from German

I would actually go further and say that the only real illustration of this is German. Other Germanic languages--and other languages generally (e.g. Finnish, Turkish, Hungarian, Hanyu Pinyin), most of which as far as I know have not gone through a historical umlauting process--borrowed the convention from German. Even there, they are widely used unetymologically (e.g. analogical umlaut plurals, foreign borrowings). Swedish is even worse in this respect, as least as regards the use of ä.

[Thry]

It varies widely, some languages I know:

Spanish: we use ´ to mark stress when it's not considered "natural", and ¨ to mark ü when it's pronounced to contrast with the digraphs gu (guerra vs bilingüe, /"ge.ra/ vs /bi."liN.gwe/). ñ is n+tilde but is considered a separate letter, it's value is /J/, the palatal nasal.
Catalan: uses ü like Spanish (also with qü, for analogous reasons, cf. que vs. pasqües /k@/ vs /"pas.kw@s/), and ´ and ` for stress accents (also marking stress patters that are not considered "natural stress"), using ´ for close vowels (ú, í, ó, é) and ` for open vowels (à, ò, è). Also has the cedilla in the letter ç (again considered a separate letter, to separate the /s/ sound from the /k/ sound; i.e. caçar /k@."sa/).
Portuguese: uses ~ to denote nasalization of a vowel or diphthong (ã, õe, ão, ãe), and ^, ` and ´ for stress. Whereas Catalan uses "é" for close e and "è" for open e, Portuguese uses "ê" for close e and "é" for open e (i.e. mês /meS/ vs pé /pE/). ` is used in Portuguese solely on the letter à to denote a phenomenon referred to as crase (crasis), which is the opening of a due to a contraction of the preposition a and a determiner or article beginning with a-, such as à (to the) or àquela (to that...). â sometimes denotes a nasalized, close a: trânsito. The rules are much less straightforward than in Spanish or Catalan, but still follow the "natural stress" principle, only that with some exceptions. They get more chaotic regarding diphthongs and hiatuses and just are more erratic and numerous. Aside from the fact that there was just a reform in 2012.

This "natural stress" system principle is easy: when the native speaker reads some words naturally, the stress is mentally parsed on a certain syllable (and this is all because of convention and specific to the language; in fact it changes). For instance, to a Spanish speaker, "amo" is naturally read as ["a.mo], so "amó" needs to be marked as an exception [a."mo]. Consequently, the "rules of graphical accentuation" are just rules on what isn't considered a natural stress. Historically, some accentuation rules have been modified for diacritic economy. Since proparoxytone (stressed antepenultimate or third-to-last syllable) words are rare, for instance, they are all marked graphically: a word is never considered proparoxytone if it is not marked. sabana is [sa."Ba.na] (savannah), sábana is ["sa.Ba.na] (blanket).
An example on how these vary according to language: whereas Spanish considers words ending in -u and -i to be naturally paroxytone (stressed in the second-to-last syllable), like Mari and Edu (both nicknames, pronounced ["ma.4i] and ["e.Du]); Portuguese does the opposite, so táxi needs an accent mark to be ["ta.ksi] whereas peru is [p@."4u]. In both languages [ko."mi]/[ku."mi] mean "I ate", but they're spelled comí and comi respectively.

French: the diacritics, much like English ´ and ç (which in fact are from French, mostly) denote fossils from an old orthography more than any functionality, with notable exceptions (i.e. é /e/ vs e/è /E/, /@/). For instance: Middle French coste evolved into côte and the ^ is a remnant of that -s-. In addition, the marks sometimes serve a real diacritical purpose, to tell words appart: du vs dû ("of the" vs past participle of "must"). This last function is also present in Spanish, Portuguese and Catalan. ¨ breaks vowel combos (Noël, haïs...), stating that they are not digraphs or diphthongs.

Italian: from what I gather, they are scarce and dispensable [some people replace them for an apostrophe ' informally], and mostly appear in oxytone (stressed in the last syllable) and tonic words (un pò, più, cantò, lassù...)

German: uses the umlaut, ¨, to manifest a change in the quality of a vowel (the historical reason is a sound change known as i-mutation, which is also responsible for things like English goose-geese, mouse-mice or man-men): ä is /E/, ö is /2/ and ü is /y/.

[ObsequiousNewt]

In my Proto-Elmin orthography, I use macrons (ē) and circumflexes (ê) to indicate medium and high tone, respectively, but in one of its yet-unnamed descendants (although the orthography is liable to change), I use a macron to represent long vowels. In another of my conlangs (Kíndai), I use an acute accent (é) to mark weak stress and a circumflex to mark strong stress, a diaeresis to mark non-diphthongs, and a trema to mark word-final ⟨e⟩ so it is pronounced (thank you, Quenya.)

Romanian is one of my favorite examples of diversity in diacritics: a breve on the a (ă) to represent the schwa, a circumflex on â or î to represent /ɨ/, and a comma below ș or ț to represent /ʃ/ and /t͡s/ respectively. Some of the old letters are even better: a breve on ĭ to represent /j/ or /ʲ/, a breve on ŭ to represent /w/ or a lack of palatalization (like Russian ъ), a breve on ĕ which was identical to ă, an acute on é or ó which now corresponds to "ea" (usually) or "oa", and, last but not least, a comma below d̦, which would become /z/.

Oh, and the general term for two dots over a vowel (most well known as an umlaut or a diaeresis) is "trema".

[Particles the Greek]

Don't forget that different diacritics can be used for the same thing in different languages, like ogonek and tilde for nasalisation, and that sometimes a diacritic can have unexpected uses, like the circumflex in Turkish.

[Hallow XIII]

or the Ogonek in Lithuanian, where it represents long vowels

wait stop there are circumflexes in Turkish???

[vec]

In Icelandic acute accents refer to previously long vowels that have now become dipthongs.

á used to be /a:/ but is now /au/
é used to be /e:/ but is now /ɛi/
í used to be /i:/ but is now /i/ whereas i has become /ı/
ó used to be /o:/ but is now /ɔu/
ú used to be /u:/ but is now /u/ whereas u has become /ʊ/ and then later /ʏ/ which is its modern pronounciation
ý used to be /y:/ but then y and ý derounded and merged with i and í, then underwent the same change as i and í (ý is now /i/ and y is /ı/)

The usage therefore seems arbitrary if you don't know the phonological history.

[finlay]

Isn't é /je/?

[vec]

Yes, sorry about that.

In Icelandic acute accents refer to previously long vowels that have now become dipthongs.

á used to be /a:/ but is now /au/
é used to be /e:/ but is now /jɛ/
í used to be /i:/ but is now /i/ whereas i has become /ı/
ó used to be /o:/ but is now /ɔu/
ú used to be /u:/ but is now /u/ whereas u has become /ʊ/ and then later /ʏ/ which is its modern pronounciation
ý used to be /y:/ but then y and ý derounded and merged with i and í, then underwent the same change as i and í (ý is now /i/ and y is /ı/)

The usage therefore seems arbitrary if you don't know the phonological history.

[linguoboy]

wait stop there are circumflexes in Turkish???

They're only found on originally Arabic borrowings and used with something like the frequency of umlauts in Pinyin (i.e. rarely nowadays), but yes. They indicate vowel length as well as palatalisation of the preceding consonant before back vowels.

[Pole, the]

http://en.wikipedia.org/wiki/Diaeresis_(diacritic) - quality change (usually reversed front/back) on vowels, don't fucking use it on consonants

Ẅḧäẗ ẗḧë ḧë.., ẅḧÿ?

[Particles the Greek]

http://en.wikipedia.org/wiki/Diaeresis_(diacritic) - quality change (usually reversed front/back) on vowels, don't fucking use it on consonants

Ẅḧäẗ ẗḧë ḧë.., ẅḧÿ?

Ĵuŝt don't ŝtart on ĉirĉumflexed ĉonŝonantŝ - avert yoŭr ĝaẑe noŵ!

[Rhetorica]

http://en.wikipedia.org/wiki/Diaeresis_(diacritic) - quality change (usually reversed front/back) on vowels, don't fucking use it on consonants

n̈ is used in Malagasy, Jacaltec, and Cape Verdean Creole for /ŋ/. (In Lilitika I wimped out and co-opted ñ so the whole language would fit in Latin-1.) It's also the only consonant known to go to 11.

As for the grave accent, I think it's worth emphasising that oxytone stress was a major historical usage—it's basically a fancy acute accent. (Polytonic) Greek still does this, like Italian, although when it was invented it was first used to mark secondary pitch (like secondary stress, except... with pitch.)

[linguoboy]

n̈ is used in Malagasy, Jacaltec, and Cape Verdean Creole for /ŋ/.

I don't recall ever having seen this in Malagasy texts, but looking the relevant section of the Wikipedia article I see that it's only used in those dialects where /ŋ/ is a phonemic, a group which does not include Merina (the basis for the standard language).

[ol bofosh]

English also makes use of diacritics, but, as with it's orthography, not so regularly, and often they're not used at all. Lone words and foreign names sometimes retain their diacritics ()

I didn't learn them so usually avoid using them. Coördination, coöperation. façade. blessèd are possible, though may not be well understoond by most (especially the first two)

[linguoboy]

I didn't learn them so usually avoid using them. Coördination, coöperation. façade. blessèd are possible, though may not be well understoond by most (especially the first two)

That's traditional New Yorker house style, any reasonably well-read American would recognise it. YMMV, Commonwealthers.

[Drydic]

The diareses are far less often used than the graves, acutes, and cedillas.

[Imralu]

wait stop there are circumflexes in Turkish???

They're only found on originally Arabic borrowings and used with something like the frequency of umlauts in Pinyin (i.e. rarely nowadays), but yes. They indicate vowel length as well as palatalisation of the preceding consonant before back vowels.

They're generally not written today, unfortunately. They're present in some dictionaries, but they're not written in print publication anymore, so you just have to know if hala is /haːlaː/ "still" (formerly written hâlâ) or /haɫa/ "paternal aunt".

[Salmoneus]

The diareses are far less often used than the graves, acutes, and cedillas.

Except that they are still often used in the words <noël> and <naïve>.

What words are graves commonly used in? Must be hiding from me at the moment.

EDIT: oh, of course, in poetry to indicate long form inflections.