Webmaster's note: In the AUE FAQ, there is a discussion of the
various uses in English of the present and past subjunctive. That
discussion is presumably based upon the assumption that there is a
well-defined subjunctive mood in English. Neil Coffey questions whether or not the (so-called) English subjunctive
really constitutes a subjunctive mood, and suggests that what people typically
refer to by the term 'subjunctive' is not a single phenomenon but several
different phenomena.
Some languages have what we can say are subjunctive forms:
that is, verb forms which, in certain syntactical constructions,
replace the 'ordinary' forms (e.g. French 'vienne', from the
verb 'venir', replaces forms 'vient'/'venait'/'viendra'/
'viendrait' in certain subordinate clauses).
English has a handful of constructions which, because they
derive from previously-existing subjunctive forms, can on the
surface appear like the subjunctive forms we observe in other
languages. However, this superficial similarity does not undermine
the fact that the defining characteristics of subjunctive forms
are syntactic rather than semantic; the fact that a particular
instance of a verb in English appears to denote e.g.
contrary-to-factness (a notion which, somewhat exaggeratedly,
is typically associated with "the subjunctive") is not
by itself an indicator of whether that verb is a subjunctive
form or not, and we should note in particular that there
is no such thing as a single 'subjunctive meaning': in
languages that have them, subjunctive forms are used in
constructions that denote a variety of meanings, and the
use to which they're put differs from language to language.
In order to show that the so-called subjunctive in English
is nowadays probably simply a conspiracy, let us consider the
following sentences which might traditionally be thought of
as constituting examples of 'subjunctive':
[1a] They requested that he be banned from driving.
[1b] They requested that he not be banned from driving.
[1c] They requested that he should be banned from driving.
[2a] Be that as it may.
[2b] So be it.
[2c] God praise the queen.
[3] If I were you, I'd be careful.
Now, on closer analysis, it turns out these constructions
differ in their nature to the subjunctive forms we
observe in other languages. In particular, true subjunctive
forms behave as conjugated forms with regard to phenomena
such as negation and contraction. In English, on the other
hand, not only do so-called subjunctive forms in e.g. construction
(1a) resemble the infinitive 100% of the time, but they actually
behave like infinitives with respect to the formation of their
negative (cf. 1a->1b) and the impossibility of contraction
(*he've; *she've). Sentences such as (1a) are consistent
in their behaviour with actually constituting a covert
modal (which we'll denote as [M]) followed by an infinitive:
[1a] They requested that he [M] be banned from driving.
[1b] They requested that he [M] not be banned from driving.
[1c] They requested that he should be banned from driving.
[1d] They requested that he should not be banned from driving.
Clearly there are then no more grounds for calling (1c)
an example of subjunctive than (1a) or (1b): a modal is involved
in each case, and it so happens that a different modal
([M] or 'should', and also probably 'might') can be chosen.
This then leaves fixed expressions (2a-2c), which can
be analysed in a similar way or else excluded from analysis
on the grounds that they are fixed expressions, and (3),
which must be analysed as a fixed formula.
Note that in saying that the 'were' in (3) was a past
subjunctive form (as is popularly suggested), we would be
proposing that speakers acquiring the language detected the
presence of an entire paradigm in which only a single verb
had a form different from the indicative (and even that
resembles another form of the verb 'be'); this is somewhat
implausible.