#ditto for Goidelic

[ukfrislandembassy]

Another way that 'phoneme inventories' mislead conlangers: seemingly systematic palatalisation contasts.

So you're looking at the consonant system of e.g. Russian, and you think to yourself, 'Man, that's a lot of contrasts, and most consonants come in a palatalised/unpalatalised opposition, I want to make a language like that'.

So you go about thinking how you're going to create this systematic contrast, and you think 'well, I want to generate it from vowels as per usual for language, but I also want to have a vowel inventory afterwards.' In most instances I've seen (and tried myself), this ends up with a symmetrical set of front-rounded vowels which palatalise preceding consonants and then retract, complemented by a set of back-unrounded vowels that front.

Here's the thing though. That particular kind of sound change is barely attested. It is seemingly attested in Nenets, but after that the trail runs dry. The category error here seems to be the assumption that 'X is a phoneme, therefore we need some way of historically deriving the fact that it must be able to stand by itself as a phoneme, right?'

But that's where the category error is, because actually the palatalised consonants in a language like Russian generally aren't evenly distributed across their various contexts.

Let's start with a simple example. Standard Japanese doesn't have *che, *she or *je, because those vowels didn't trigger palatalisation in the Japanese varieties from which the standard derives, and furthermore palatal consonants before vowels other than *i are mostly found in loanwords from Chinese. Thus we can observe that palatal consonants in Japanese have a different distribution to non-palatalised consonants when it comes to which vowels they can precede.

A similar point can be made about palatalisation in Slavic, because it's again skewed by which vowel follows. The only reason that Russian even has palatal consonants before /o/ is because of a funky Russian-only rule which backs and rounds *e before hard (coda) consonants (in stressed syllables only!). For most other Slavic languages a syllable of the *sho type is rare to non-existant. And even in those languages where it does occur, it's not like these kinds of syllables are equally as common as palatal consonants before front vowels (where non-palatalised consonats are almost non-existant). The only context in Russian where palatalised and non-palatalised consonants are on even close to an even footing is word-finally, and only because Russian actually bothers to preserve contrastive palatalisation there (unlike many other Slavic languages).