It's just slightly sloppy terminology to avoid introducing nearly synonymous words.
A model of a (first-order) language would be just a set together with appropriate interpretations of all constant, function and relation symbols.
A model of a theory would then be a model of the language in which all the sentences/formulae of the theory are true.
You can argue that a language specifies its formulae, and therefore a model of a language should morally be a model of all formulae of that language, as per the second point, but then you have to find a way to bootstrap yourself to the second definition.
In practice this shouldn't cause too much difficulty, as whenever you come across model in Just-Weese (or elsewhere) it should be clear from context which meaning is intended.
(Sometimes structure or interpretation is used for models of the first type, reserving model for models of the second type.)
At times it's almost enough to make you wonder if the sloppiest expositions in mathematics come from the AMS's 03-XX subject classification....
Added due to edit.
I'm almost certain that Just-Weese do not use language and theory as synonyms. Consider the second paragraph of quoted text in your previous question.
And, no, ZFC does not consist of only nine axioms. In particular the Axiom Schema of Separation (sometimes called (Restricted) Comprehension, or Specification) consists of an infinite list of axioms. But I believe this comes up later in Chapter 7 of Just-Weese.
You might even see later (in Chapter 12) that there is no finite axiomatization of ZFC (assuming the consistency of ZFC).