When to use the subjunctive: some difficult cases

On the previous pages, we saw various circumstances where the French subjunctive is triggered. In general, these circumstances fit in with our statement that the subjunctive expresses a non-assertion. There are some other cases that are more difficult to rationalise and to some extent are just a matter of convention.

espérer takes the indicative

The verb espérer, meaning "to hope", is the verb par excellence for introducing a non-assertion. But conventionally, it takes a normal indicative:

j'espère qu'elle pourra venir
I hope she'll be able to come
Feedback Suggest a change / proposez une modification

One view is that espérer nonetheless leans in favour of the event actually being realised, i.e. introduces "more of an assertion" than, say, désirer or souhaiter, which do trigger the subjunctive (Leeman-Bouix, 1994, quoted in Rowlett (2007:149) and Jones (1997:189)). But not only is this argument tentative, but it is to some extent unnecessary. There is a tendency in informal speech for speakers to in fact use the subjunctive with espérer:

j'espère qu'elle puisse venir
I hope she'll be able to come
Feedback Suggest a change / proposez une modification

So an alternative view is that speakers do indeed 'naturally' treat espérer as taking a non-assertion triggering the subjunctive, just like its other synonyms. The use of espérer with the indicative is largely an 'artificial' prescriptive convention.

si takes the indicative but que (meaning 'if') triggers the subjunctive

The conjunction si generally takes indicative, again despite the fact that it seems an excellent candidate for introducing a non-assertion:

si elle vient, on peut lui demander...
if she comes, we can ask her...
si jamais tu veux nous contacter, voici notre adresse
if ever you want to contact us, here is our address
s'il pleuvait, je serais obligé de rester chez moi
if it rained, I'd be forced to stay home
Feedback Suggest a change / proposez une modification

On top of this, que, when used to mean 'if' or 'whether', triggers the subjunctive. The latter can be used in coordination with si, or on its own:

s'il fait beau, et qu'il ne fasse pas trop chaud, on peut...
if it's fine, and not too hot, we can...
que tu viennes ou que tu ne viennes pas, moi j'irai
whether you come or not, I'll be going
Feedback Suggest a change / proposez une modification

"Assertion toggling", and the verbs penser and croire

The 'cognitive' verbs penser and croire generally behave as though they introduce direct speech, and take the indicative:

elle pense/croit que c'est vrai
she thinks/believes it's true
Feedback Suggest a change / proposez une modification

However, they often trigger the subjunctive in either the negated or, more particularly, interrogative forms1:

elle ne pense/croit pas que ce soit vrai
she doesn't think/believe it's true
croit-elle que ce soit vrai?
does she believe it's true?
Feedback Suggest a change / proposez une modification

However, when the verb is both negated and interrogative, the form generally "flips" back to the indicative (Abouda, 2001: 20):

ne croit-elle pas que c'est vrai?
does she not believe it's true?
Feedback Suggest a change / proposez une modification

Thus, negation and interrogation are sometimes referred to as "assertion toggling" operations. In isolation, either can switch the indicative to the subjunctive, but together they 'cancel each other out'. Note that this toggling effect can also apply to adjectives such as possible (cf Rowlett, 2007:150).

Aussi/tout/quelque/si ... que ..., Autant que ...

Prescriptively, autant que... ("as far as...") takes the indicative except in the construction autant que je sache (Thomas, 1971:44). But in case that was ever actual usage, there's now a tendency to extend the subjunctive to other verbs. Price (2003:374) suggests that the choice depends on the 'degree of certainty or uncertainty the clause is intended to express'. A quick Google search suggests a roughly 50:50 split between (pour) autant que je m'en souviens and (pour) autant que je m'en souvienne ("as far as I remember").

The constructions aussi ... que, tout ... que and quelque ... que, si ... que are used in expressions such as:

tout père que je suis
father as I am
tout riche qu'il est
rich as he is
tout riche qu'il soit
however rich he might be
aussi bizarre que cela puisse paraître
si bizarre que cela paraisse
however strange it may appear
quelque difficile que cela paraisse
however difficult this may seem
quelque difficile que puisse être ce travail
however difficult this work may be
Feedback Suggest a change / proposez une modification

tout triggers indicative or subjunctive depending on whether the 'factual nature of the statement is stressed' (Price 2001:378). The use with the subjunctive corresponds roughly to cases where English uses however. The other constructions appear to systematically trigger the subjunctive.

Subjunctive "leaking"

There are some cases in both English and French where a logically unexpected form occurs because one word influences or 'leaks' on to a nearby word or, structurally, one language feature 'percolates' or 'leaks' down from one word to words that it governs. Interestingly, the French subjunctive is an example of such leaking. Rowlett (2007:153) gives the following example:

je ne voulais pas qu'elle croie que je sois triste de rester
I didn't want her to believe that I was sad to be staying
Feedback Suggest a change / proposez une modification

Strictly speaking, there's no logical reason to expect the subjunctive here, at least in terms of what verb directly governs what: voulais selects a subjunctive, croie. But qu'elle croie que... is a non-negative, non-interrogative form of croire, which under such circumstances usually takes the indicative. A negative main clause generally triggers a subjunctive in its subordinate clause, but que je sois... isn't directly subordinate to the main clause. What seems to have happened is that the negation plus preceding subjunctive form have somehow 'induced' or 'leaked' a subjunctive in the bottom clause2.

1. Opinions appear to differ on how common the indicative is with other forms of interrogation (est-ce que and intonation). Abouda (2001:17) cites Huot (1986) as 'clearly demonstrating that the subjunctive is really only acceptable (and in common usage) if the verb of opinion is in a total interrogative construction characterised by subject-clitic inversion' (my translation). Contrast this with Hawkins & Towell (2001) who give the example est'ce que Jean pense que Pierre soit venu?.
2. Rowlett gives a possible structural mechanism for this 'leaking' happening, which would assume it is a linguistic process. Note that this leaking pattern is by no means systematic, however. It may be a conscious 'hypercorrection' rather than a truly linguistic process that requires a formal structural mechanism to explain it.

 French grammar index
 French-English dictionary
 English-French dictionary

This page written by Neil Coffey. Copyright © Javamex UK 2017. All rights reserved.