As your file is read, each word is checked against the following lists. The first match found determines what will be done with the word.
01. External preprocess list (RSTYLE.PRE)
02. Internal command keyword list if first word in command or next word after DEBUG
03. Internal set keyword list if next word after SET
04. Internal keyword list
05. Internal casing list
06. Internal variable list
If not found yet it is considered a non-keyword.
07. External expansion list (RSTYLE.EXP)
08. External casing list (RSTYLE.CAS)
09. External variable list (RSTYLE.VAR)
If not found, each word is cased as a non-keyword but added to RSTYLE.NEW without alteration. If you check RSTYLE.NEW after each run you will be able to put those words in RSTYLE.CAS or RSTYLE.VAR and case them the way you want if not that way already. If you leave words in RSTYLE.NEW things will slow down as the file grows so check on it occasionally.
Also, words in RSTYLE.NEW will not be used in casing. If you see the case changing to all lower on new words, just move them to the appropriate files and they will be properly cased from then on.