In case anyone wonders, there is an actual basis for the vowel groups: Dr. J. C. Wells's lexical sets. The purpose of the sets is to describe phonological differences between dialects. I decided to use them in an attempt to simplify the importation process, since US and British vowels differ considerably. Inlis doesn't follow either of these dialects exclusively. Rather, it's meant to be a pan-English dialect that's closer to creoles than to the rival dialects. Speakers should always sound somewhat foreign and exotic: this is one time neither Received Pronunciation nor General American is welcome.
It should be obvious that a and o are overloaded, especially a. In my dialect, the seven mappings of o are just three distinct vowels--or technically just two. Such phenomena are common.
Still, a is overloaded, and I've tried to reduce the load, so far unsuccessfully. The current mappings seem to be the best.
Probably the most controversial items will be the mappings to o of the LOT, CLOTH, PALM, and THOUGHT sets. From a (Western) US perspective, it would make more sense to map these to a, further overloading that overworked vowel. But I chose o because
1. the sounds are commonly associated with that letter both in the main English dialects and in creoles
2. it's easier to map all of these to a single vowel, and a is already overloaded, while apart from these, o would be somewhat underloaded.
This does lead to some oddities. For example, if I imported wander and wonder directly, wander would become wonda and wonder wanda.
The big weakness is that the mappings make the English tendency toward homonyms even worse, but since I'm trying to rein in synonyms anyway, I can avoid a lot of the problem.
Next time I'll explain a bit about the grammar and importation mechanism.