Teaching them who Buddy Holly was would be more valuable than trying to make them shun covertly inflected mandative clauses.
For grammar bullies “the subjunctive” is sacred ground. Reforms proposed for the British national curriculum in 2012 required teaching use of the subjunctive not later than sixth grade. People seem to think the subjunctive is a fragile flower on which civilization depends; without our intervention it will fade and die, and something beautiful, fragile, and important will be lost.
As usual, virtually none of the things people believe about the subjunctive or its status in English are true. Most purists who witter on about it couldn’t actually pass a test on distinguishing subjunctive from nonsubjunctive clauses to save their sorry asterisks.
But then they don’t have to: Merely mentioning the subjunctive approvingly and urging that it be taught is enough to establish one’s credentials as a better class of person — one who knows about subjunctives.
This post is simply an attempt at surveying the facts (imperfectly; but see Rodney Huddleston’s beautiful treatment in The Cambridge Grammar of the English Language, henceforth CGEL, pages 993–1,000).
It’s not about verbs. English has an odd fondness for homophony or homography in verb forms: Grammatically distinct forms of verbs often share spellings or pronunciations, so you get fewer distinct shapes than you might have expected in the inflection table; but it’s crystal clear there is no point in having a “subjunctive” box anywhere in that table. Not a single verb in the language has a special subjunctive shape. CGEL distinguishes three tensed forms and three untensed for typical verbs. Here’s the array for shake and bake:
| plain present | take | bake | |
| TENSED | 3rd-singular present | takes | bakes |
| preterite (simple past) | took | baked | |
| past participle | taken | baked | |
| UNTENSED | gerund-participle | taking | baking |
| plain form | take | bake |
Shake has distinct shapes for the preterite and the past participle; bake doesn’t. Both share shapes for the plain present and the plain form (the latter is used in imperatives, in infinitivals, after modals, and elsewhere). Yet one verb in English—the weirdest yet commonest one, namely be—has a plain form different from all its other forms: be does not share its shape with is. Notice the contrast between present tense and imperative with shake and be (I underline the crucial forms):
| [1] | He shakes it vigorously. |
| Shake it vigorously. | |
| She is careful. | |
| Be careful. |
With both verbs you can contrast present-tense clause with imperative; but with be you actually see a different verb form shape.
Now I’m ready to define English subjunctive clauses. They are finite yet tenseless clauses with their verb in the plain form. Virtually all are subordinate clauses, usually introduced by the standard finite-clause subordinator that. (A few optative main clauses with subjunctive form survive, relics of a bygone age: Heaven help us, God damn you, God be thanked, Long live democracy, So be it, etc.)
The subjunctives that are most robustly present in contemporary English are what CGEL calls mandative subjunctives. They go with verbs and adjectives of necessity, cruciality, or demand. A typical example (with the subjunctive clause underlined):
| [2] | It’s vital that he be more punctual. |
Finite clauses with pronoun subjects have to have nominative pronouns, hence we find he, not him. The finite declarative subordinator that is the same as in Jill says that he is punctual, but in [2] the subordinate clause verb (be) is in the plain form.
Today many speakers of Standard English substitute the plain present for the plain form in mandative subjunctives, and say It’s vital that he is more punctual and so on. Yet even they have a subjunctive-clause construction, covertly, because they too see sentences like [3] as ambiguous:
| [3] | Jill insists that I wear shoes. |
No contrast in meanings has been lost: It says either that Jill demands shoe-wearing or that she affirms I’m already a shoe-wearer.
By replacing first person in the subordinate clause by third person, we can make the difference pop into view for those speakers who use overt subjunctives, because two shapes emerge:
| [4] | Jill insists that Jack wear shoes. | [unambiguously subjunctive] |
| [5] | Jill insists that Jack wears shoes. | [nonsubjunctive for all, covert subjunctive for some] |
These patterns are simple, robustly entrenched—and not very important. If overt subjunctives disappeared completely, it would matter very little: Everyone would then be a covert-subjunctive speaker. Ambiguous sentences would be more frequent to some tiny degree (Lane Greene discusses one here). But English-speaking civilization does not hang on this, and there are many things more worth a sixth-grade teacher’s time than whether the final s should be left off wears in [5]. Teach them what planets are; why dogs die; what the courts do; who Buddy Holly was; something more important than the footling matter of whether overt or covert mandative clauses are better.
[One footnote: Many discussions of English grammar refer to the were of If only there were something we could do as a subjunctive, often as the "past subjunctive"; but it's a totally different construction, the irrealis clause. The irrealis is overtly distinct only for the verb be (with every other verb the preterite is used), and only in the first and third persons, and only for some speakers. It occurs in certain clauses describing situations not claimed to hold in this world. It doesn't talk about the past at all: I wish it were a different color certainly isn't the past tense of *I wish it be a different color!]