Dear Piya,

> I'm not sure if chamhi can be divided into cha-mhi, because there is
> no such syllable or sound in Pali. But there is nha as in nhataka.
> The oddity is best explained as metri causa.

The Saddaniiti explains the following group of combined letters: ~nh,
.nh, nh, mh, yh, lh, vh, .lh. Apparently, the pronunciation of the h
in these combinations is not the same as the h alone as in hoti. The
difference is that the former is an urasija (born of the chest) while
the latter is a ka.n.thaja (a guttural).

I'm not at all clear how the urasija h is to be pronounced. I have
thought of an aspirated sound but I find it impossible to pronounce
the h in yh as an aspirated sound in combination with y.

Your example of nhataka with initial nh shows that these combinations
have to be pronounced similtaneously. Another convincing example is
.lh. I don't think anyone would think of dividing ruu.lhi into
ruu.l-hi.

I reproduce below Aggava.msa's explanation:

ya-ra-la-va-.lha-pa~ncamehi yutto ha-kaaro urasijo, kevalo ka.n.thajo
va, saasanikappayogato pana "na-kaaravajjito:

~na-.na-na-mehi sa.myutto tathaa ya-la-va-.lehi ho
saasane oraso ~neyyo, ka.n.thajo yeva kevalo;

ta~nhi ta.nhaa nhusaasumha muyhate vulhate tathaa
avhito ruu,lhi icc ete payogaa honti saasane;
(Sadd III 609)

A study of the metre of the above two verses might clarify how the
syllables are to be divided. Note the 3 short syllables at the
beginning of each verse if we assume ta-~nhi ta-.nhaa (sssl) of the
second verse.

Best wishes,
Jim