Halliday & Matthiessen (2014: 227-8):
The clauses are the same experientially; they both represent a configuration of Actor + Process + Goal. But they differ in how these rôles are mapped onto the interpersonal functions in the modal structure of the clause. In the ‘operative’ variant, the Actor is mapped on to the Subject, so it is given modal responsibility and in the ‘unmarked’ case (in a ‘declarative’ clause) it is also the Theme; and the Goal is mapped on to the Complement, so in the ‘unmarked’ case it falls within the Rheme. However, in the ‘receptive’ variant, it is the Goal that is mapped onto the Subject, so it is assigned modal responsibiity and it is also the Theme in the ‘unmarked’ case; and the Actor has the status of an Adjunct within the Rheme of the clause and, as an Adjunct, it may be left out …