The reason is the combination of the word break and the gaNa break,
or the lack of a semblance of correlation between the two. This can
happen with grAnthika bhAsha as much as with vyAvahArika bhAsha.
Mostly this problem happens with lines that have many, if not most,
if not all laghuvus. The concurrency of the word and gaNa breaks is
more important in this case than with others.
There are two major ways of "reading" a kanda pandyam, to check the
gait of it (and thereby the adherence to the rules). One is a 4+4+4
break of the (mAtras) of the short lines, and the other is a 6+6 bbreak.
For the longer, second and fourth lines, the two breaks are
6+3+3+4+4 and (suddenly it escapes me! may there ain't one)
When the words don't neatly break along this division, the gait suffers;
or vice-versa.
Oh, the break need not always happen. For example, the long line
may have a 6+3+3+8 which happens in,
neerATa vanATamulaku
pOrATam beTlu galige? purushOttamucE
....
SreerAmuni daya cEtanu
nArUDhiga sakala janulu nourA yanagA
SreemAn mee fAdaru nA-
kO mAdiri nAnna gAre, okaTE koradA
....
I will think about and post some poems where the lack of gait is
quite apparent. When reading those poems, one repeats them with
several imposed word breaks, to check the validity. So, the upshhot
is that the rules are necessary for the gait, but not sufficient.
BTW, the same can be seen in other meters. More so in some
tETageeti and ATaveladi poems.
Ramakrishna