The following was performed on Mavericks 10.9.5, with Pages '09 v4.3, and Pages v5.2.2.
I just retrieved a Medical dictionary from here. It is intended for use with MS Word, or OpenOffice, and contains 66,239 terms. By double-clicking on the .zip file, a all-unicode.dic file becomes available. It is a binary encoded file. Plopping this all-unicode.dic file into home/Library/Spelling, logging out/rebooting, and attempting to catch spelling errors in either Pages did not work.
So, I thought, ok — perhaps it is supposed to be text as is the LocalDictionary file in the above location. I wrote a very short Python script to convert it from UTF-16 to a UTF-8 text file (so accented characters would survive), and used this as Med.dic. Again, logging out, or rebooting, did not change either Pages' ability to correct mispellings of known terms in this Medical dictionary file.
#!/usr/bin/python
import codecs
# read in a binary leximed medical terms dictionary intended for MS-Word
# write out a utf-8 text file
f = codecs.open('all-unicode.dic', 'r', 'utf-16')
fo = codecs.open('Medical.dic', 'w', 'utf-8')
for l in f.readlines():
fo.write(l.replace('\0', ''))
Next, I saved a copy of the home/Library/Spellings/LocalDictionary file. Using TextEdit, I added the Medical dictionary content to the end of the existing LocalDictionary content, and then resorted alphabetically. Removed Med.dic from home/Library/Spellings, and added back the updated LocalDictionary file. Rebooted. Neither Pages would fix deliberate medical term mispellings, and would only offer to guess.
Tom, any further thoughts?