The images in Mapping for the Continuous Phonetic Based Input Method show the mappings between English tokens and their equivalent codepoints in each of the target scripts supported. The CONSONANT category means the mapping is between the English tokens and consonants of the script. The VOWEL category means that mapping from English tokens and vowels of the script. The OTHER category includes mapping of characters that do not exhibit the properties of consonants and vowels (whose form does not change depending on the surrounding character).
The keywords CONSONANT, VOWEL and OTHER also mean that these characters are part of Unicode standard. The section SPECIAL CONSONANT, SPECIAL VOWEL or SPECIAL OTHER means that though in principle these characters display the properties of consonants, vowels or others they are not officially part of the Unicode standard and are font dependent. They are assigned codepoint values in Unicode Private User Area. They are supported in Oracle Solaris UTF-8 locales and the mapping may not work in a different platform.
These mapfiles are not the same as the ones in your system, but slightly edited ones for removing unneeded keywords for the context of this discussion.
In the VOWELS and SPECIAL VOWELS section, an independent form and a dependent form is displayed for the same English token depending on the context. See How the Continuous Phonetic Input Method Works.
The Malayalam script contains a special ‘CHILLU’ section, that is actually the SPECIAL OTHER category.