One can have an extremely good grasp of grammar without the slightest knowledge of the elaborate (and wholly artificial) theory that grammarians have retrospectively applied to it. In other words, grammarians are not the only ones who know how to write correctly. To put it another way, I don’t need to know, as I write my latest masterpiece, whether my nouns are abstract, my adjectives predicative, or my verbs transitive.

Also, please check up the word “homophone” in a good dictionary, such as the O.E.D., where you’ll find that Question 5 includes two correct answers. The essential point about homophones is that they sound the same. According to the Shorter O.E.D., they may differ in spelling, meaning or origin.

And who on earth puts (a specific number of) fungi in the soup? What’s wrong with good old mushrooms? Excessively artificial and unlikely examples are self-undermining and counter-productive. I wouldn’t “put three fungi in the soup” for the same reason I wouldn’t ring the emergency services to say “I believe my spouse is experiencing a myocardial infarction”.

Finally, multiple-choice tests are a poor way of testing grammar, particulaly if the underlying intention is to teach, and not merely to test. This is because each of the wrong answers must be read – and, to some extent, absorbed – when trying to identify the correct one. This is likely to reinforce – or, indeed, even to create – incorrect associations, which must then be unlearned. The sad fact is that multiple-choice tests are widespread in language teaching because only they are so easy to construct and to mark, and not because they are of any real benefit to students.