In the process of creating the monosyllabic roots the first time (and, by extension, the second time), I had an idea of a few disyllabic roots that I wanted. As «me», «se», and «ke» are the singular personal pronouns, I wanted «mema», «sesa», and «keka» as the plural personal pronouns. That meant, for 4-phoneme disyllabic roots, there was an MM root (‘MaMa’), an SS root (‘SaSa’), and a KK root (‘KaKa’), so I figured that all phonemes would pair with themselves. Furthermore, «me», «se», and «ke» were chosen because MKS is one of the few trios of phonemes that doesn’t conflict with one another (in terms of my Hamming distance parameters). Extending that, I figured I could pair any phoneme with another non-conflicting phoneme and itself. There were a few options available, but the best one ended up being MS, NK, PL, TY with «h» left out. For example, MS worked as follows:
It’s basically a Punnett square. So, really, MS = MS, MM, SS, SM in terms of the roots that was created from the pair. There ended up being a few extra that worked, but most of them stopped working later one when I started creating the 5-phoneme disyllabic roots.
Actually making those 5-phoneme disyllabic roots was much trickier. A while back, I figured out that I had two special trios of initial consonants: MKS and NPY. Within each trio, the consonants do not conflict with one another. And no consonants are shared between the two trios. (However, «t» and «l» are left out, and I knew I would have to integrate them later somehow.) I couldn’t cross them with themselves because each phoneme was already paired with itself, and there was already MS, which would conflict in MKS. I figured that I could instead interpolate these special trios with each other to get a bunch of non-conflicting pairs to create 5-phoneme roots, and I was right!
- MN, MP, MY, KN, KP, KY, SN, SP, SY
- NM, NK, NS, PM, PK, PS, YM, YK, YS
So, that first set creates «m-a-n-a-», «m-a-p-a-», «m-a-y-a-», «k-a-n-a-», etc. Each root would need an extra phoneme in place of one of those dashes, obviously, to make it a 5-phoneme disyllabic root.
Right off the bat, I knew I couldn’t use KN and NK since I already had those for the 4-phoneme disyllabic roots. Furthermore, that list would create way too few options. Of course, that list also had plenty of holes in it that could be filled, and I spent such a long time trying to figure out what they were. I don’t even know how to explain that process, but I can happily share the results:
- MN, MP, MY, NH, PH, TK, TL, KP, KS, KY, SN, SP, SY, YL, YH, LM, LS, HM, HT
- ML, NM, NS, MH, PM, PK, PS, TH, KT, SK, SL, YM, YK, YS, LT, LY, HN, HP, HY
Then, I had to figure out the rest of the pattern to make as many 5-phoneme disyllabic roots as possible. It came to figuring out into which position each of the inserted fifth phoneme would go.
- MN becomes «m-a-n-a», and one of those dashes has to be filled.
- I focused on one syllable at a time.
- For «m-a-», I chose «y» for the first dash, and both «m» and «l» for the second dash (because «m» and «l» don’t conflict).
- That gives «mya», «mam», and «mal».
- For «n-a-», I chose the opposite: «l» for the first dash and «n» for the second dash.
- That gives «nla» and «nan».
- Crossing those gives «myana», «mamna», «malna», «manla», and «manan».
Then, I did the opposite for NM since it is the inverse of MN. So, the syllables are just switched. «myana» becomes «namya», «mamna» becomes «namam», «malna» becomes «namal», etc. This creates a set of roots that do not conflict with one another. However, some issues did pop up later.
At the time of planning, it looked like I could have 166 disyllabic roots total compared to my expected minimum of 165. Unfortunately, that meager 166 shrunk to 162 when I had to rework the numerals—or, rather, rework the roots so the numerals could fit in the scheme I had created. However, I will share the roots a little bit further down the line.
In the next part, I will talk about the process of creating the root database!