Thanks for the quick and insightful reply! :)

Korean Hangul script is a thing of beauty for sure. I haven't studied it but I know a little about it.

The principle of the IPA setup I'm thinking up is akin to the compose method, but that's not available out of the box in Windows where dead keys take its place. In compose terminology, I'd use Multi+"IPA-key"(something available)+number+letter to achieve the same thing.

That's 4+ key presses for one glyph, but that's okay with the intended usage: One usually only needs to write a bracketful of IPA every now and then and that makes switching to another layout or compiling markup or whatever impractical compared to having it handy on the keyboard in my opinion – but it doesn't have to be fast as it's only a few glyphs at a time.

The great advantage to my method is that it lets you know more or less how to type a glyph if you know your phonology, or if you don't then you'll learn the phonology of that glyph as a bonus which makes it easier to remember for the next time you need it or even for hunting and pecking: "Let's see, that sound is near the middle of my mouth isn't it ... so I'll use the '5' key then. Oops, that's not it – but the neighbor works, yay!"

It'll also open up for writing African latin scripts which is a bit complex but would be a good thing. My proposed system should have room for capitalized versions of the needed glyphs so you could write Berber and Togo latin script and several others at the least. Not as proper locale layouts but for referring to these languages.

As for the crowding issue, it wouldn't take up any important real estate but a priori overload dead key mappings that don't do anything beforehand. Non-interested typists could safely ignore it and only notice it when they make a typo after a dead key. It'd only be a matter of documentation, which could be kept separate from the main one.

I'm considering using the AltGr+Shift mappings for these dead keys instead, as that'd be simpler.

Heh, a fascinating idea. The camelcase approach is probably more intuitive than Sampa? Although I'd expect Th for the þ sound, both from phonology and usage (and Nj for ñ). So it's not all that easy. I expect it's because þ is a fricative – but it doesn't quite sit with me nevertheless.

Your Flownetic is pretty, but goes too far for me. Like Colemak, it's better to build on something recognizable like IPA does – to keep the learning and implementation curve manageable. Parts of IPA work very well, with a recognizable base letter and consistent elements denoting articulation (palatal and retroflex hooks); other parts are more chaotic unfortunately. Otherwise, Tolkien's Tengwar script is almost phonetical and featural like Flownetic – very pretty but not very useful for us!

You're obviously a resourceful and creative person! ^_^

I need to write IPA, and I want to do it properly. IPA isn't perfect but it's the standard so that's what I'll use. And I'll want to do it without installing anything special except the keyboard layout which I need anyway.

Nobody claimed that Hangeul were phonetic, just that it's beautiful. :)

Mellon! :)