Given that English is compulsory, the only students who wouldn't be linked to the English candidature would be those who are accelerating and/or accumulating their HSC. There are only a small number of these students.
Courses don't need to be directly linked to the English candidature in order to be scaled. So long as the matrix of courses is sufficiently interconnected, all courses can be scaled. Problems would arise with a disjoint course (or set of disjoint courses) that is (are) isolated from all other courses. For example, in the case of a single disjoint course, none of the students taking that course would be taking any other courses. It is not possible to scale disjoint courses.
Small candidatures do not always present a bar to scaling. In my experience, it is often possible to scale courses with as few as four students. However, it is not always possible. The greatest problems posed by small candidatures are, I believe, in relation to the non-linear transformation that is used to actually scale the marks once the scaling parameters have been determined. The transformation (which requires further parameters to be determined through an iterative process) will not always converge; or, if it does, it may converge to a solution which severely distorts and/or skews the distribution, even though it produces the desired scaled mean and standard deviation. A linear transformation is used where the non-linear transformation is inappropriate.
If the scaling of a particular course becomes a mathematical impossibility - i.e. the equations simply have no solution - then it is my understanding that no scaling occurs. The standardised marks (mean=25, stdev=12) are allowed to stand.
