More ideas related to Tirelat spelling and history.

I'm looking at patterns and trying to see if I can get any clues from 
them. Doubled consonants are phonemic in Tirelat, but not every 
consonant is found doubled. E.g.

pika "shovel"
pikka "width"

khatu "grammar"
khattu "to preserve"

dapaku "uranium"
dappaku "to make someone laugh"

Consonants that can be doubled are /f/ /k/ /l/ /m/ /n/ /p/ /r/ /s/ /t/.

Consonants that I haven't got examples of doubled are /b/ /d/ /dz/ /g/ 
/gh/ /j/ /kh/ /ł/ /ŋ/ /rh/ /š/ /ts/ /v/ /w/ /z/ /ž/.

I can think of two explanations for the difference. On the one hand, 
doubled consonants like /bb/ or /dd/ may have existed in older versions 
of Tirelat, but the distinction was lost. There's a hint that there were 
two /d/-like sounds in earlier Tirelat: perhaps those were actually /d/ 
and /dd/.

nidu > nidu (Dialect A), niru (Dialect B)
viddu > vidu (both dialects)
viru > viru (both dialects)

(Or modern /vidu/, with [ð] as an allophone of /d/, could be a remnant 
of an older /ð/ phoneme which is no longer phonemic, while original /d/ 
merged with /ð/ in some dialects and /r/ in others.)

The second explanation, which seems more likely, is that the doubled 
consonants are the result of assimilation. Perhaps /pikka/ was 
originally /pitka/, /khattu/ was /khaptu/, /nallu/ was /nanlu/, and 
/dzimma/ was /dzighma/. In that case, modern Tirelat spelling might 
preserve the original clusters. As a counterexample, the word "tepta" 
does have a /pt/ cluster, which hasn't assimilated to /tetta/, and there 
are words with /kt/ clusters.