Uh... no. Well, the second part, yes.
I really ought to clean up the terminology I'm using. By "language file", I mean "file containing a sample of the language", not "DF language raw file." That is, the file's contents should look like this:
kulet alak bidok nicol anam gatal mabdug zustash sedil ustos emar agak izeg bemong gost ontak tosid feb berim ibruk ermis thoth thatthil gostang libash lakish asdos roder nel biban ugog ish robek olmul nokzam emuth fer uvel dolush agek ucat ngarak
...
and not like this:
language_DWARF
[OBJECT:LANGUAGE]
[TRANSLATION:DWARF]
[T_WORD:ABBEY:kulet]
[T_WORD:ACE:alak]
[T_WORD:ACT:bidok]
[T_WORD:AFTER:nicol]
[T_WORD:AGE:anam]
[T_WORD:AGELESS:gatal]
[T_WORD:ALE:mabdug]
[T_WORD:ANCIENT:zustash]
[T_WORD:ANGEL:sedil]
[T_WORD:ANGER:ustos]
[T_WORD:ANIMAL:emar]
[T_WORD:ANUS:agak]
[T_WORD:APE:izeg]
[T_WORD:APPLE:bemong]
[T_WORD:ARCH:gost]
[T_WORD:ARM:ontak]
[T_WORD:ARMOR:tosid]
[T_WORD:ARROW:feb]
[T_WORD:ARTIFICE:berim]
...
You make one of
those files, select it in DFLang, and tell it to generate a language. You can find some examples of these files in DFLang/languages.
To make a roots file: start as if you were generating a language, but check the box next to "Use symbol file" and click "Generate Roots!". A file will appear in DFLang/languages called "<something>_roots.txt"; inside you'll see a list of all of the symbols in DF matched up with short strings of letters. These shorts strings are
roots for the language you're about to generate.
Back in DFLang, check the box next to "Use roots file" and make sure you have the correct file selected. If you generate a language now, whenever DFLang generates a word that contains one of those roots, it'll try to assign it a meaning that belongs to that root's symbol. If ARTIFICE's root is "kal", and DFLang generates the word "mikalit", it'll try to make that word mean DAGGER or BOAT or CLASP or some other artifice-related concept.
Also: While I was answering this question, I found a little bug. Expect it to be fixed in a day or two.