If you're reading the phonology section of the Wikipedia article on Greenlandic, it starts with some lists - vowels, consonants, syllable shapes - but then it starts in on several paragraphs full of rules and sound changes. This is what was missing from the forum post! But maybe new makers of languages are just visible examples of a certain conception we have about the way we speak. It's easy to notice the fact that languages all contain distinct "sounds" (which is to say, phones), but words of course aren't just lawless noise sequences. In early infancy, the rules our mother tongue applies to our native language is irreversibly wired into our brain, so anyone can be forgiven for being blind to them, but the truth is that every word in every language has been run through a phonological filter - allophony.
The concept of allophony - or more broadly, synchronic sound change - is essentially linguists' way of explaining the gaps in a language. Let me give an example: Greenlandic words often have doubled consonant sounds in the middle of its words: [nː], [pː], [sː], [qː], etc. There are some conspicuous abscences though - [v], [l], and [ʁ], phones which all definitely show up in other contexts, are never doubled. On the other hand, we see [fː], [ɬː], and [χː], consonants which never occur un-doubled. Suspicious... "aha!" says the clever linguist. "The never-double set are all voiced, and the always-double set are all voiceless. In Greenlandic, voiced consonants are always unvoiced next to other consonants. Those never-double consonants do get doubled after all - they just lose their voicing when they do."
And that's it. This is an example of an allophonic rule, which converts phonemes (abstract, underlying sound categories unique to a language) to phones (real noises). So in this case we could say that Greenlandic has a phoneme /v/ that gets spit out as the phone [f] next to another consonant and [v] anywhere else. [f] and [v] are both called allophones of /v/. Take two notes here: First, that phonemes are written in slashes // and phones is square brackets . That's just convention. Second, that the phoneme is abstract, so any representation works. It's more helpful to pick a symbol that represents a common phone of that phoneme, or the phone of its pronunciation when it's not obviously being affected by assimilation, but we probably could have gotten away with representing that phoneme as /f/.
So what type of allophony is most common? Not just anything could work - a rule changing /s/ into [w] between vowels wouldn't make any sense at all. Allophony is the product of the universal human tendency to work less hard at pronunciation if there's no risk of miscommunication, so usually allophony is assimilation - sounds becoming more similar to their surroundings - so that our speech organs don't have to work as hard switching orientation. For example, it's super duper common, almost universal, for /n/ to become [ŋ] before [k]. [n] is an alveolar sound, but [k] is velar and so is [ŋ]. If it's about to be needed at the soft velum for a [k], the tongue probably doesn't want to make the marathon run to the alveolar ridge first, and settles for a [ŋ] as close enough. This kind of process is the engine for most allophony. Speech sounds are all defined by a combined set of features - [k] is a voiceless, velar, stop consonant - and the simplest type of assimilation just switches one of the features of a phoneme to that of an adjacent one.
Some really common allophony rules:
- Assimilating the place of articulation of a nasal consonant to a following consonant (i.e., /np/ > [mp])
- Turning voiceless sounds voiced between voiced sounds (i.e., /asma/ > [azma])
- Turning stops into fricatives between vowels (i.e., /ada/ > /aða/)
- Devoicing sounds at the ends of words (i.e., /bad/ > [bat])
- Pulling unstressed vowels close to the center (i.e., /iˈvana/ > [ɪˈvanə])
- Pulling consonants pronounced with the tongue toward palatal position in the vicinity of front vowels (i.e., /ki/ > [ci] or [t͡ʃi])
Allophony occurs in essentially every language, and the specific rules each languages uses contribute a lot to the way it sounds. Allophony is also the engine of historical accent change, but maybe that's a story for another post.
Index Diachronica (an encyclopedia of sound changes that actually happened)