Yes, switching to roman is what one normally does, in captions or the like.

thank you!

Your idea is a bad one in the first place. We have generations of experience with bilingual English/French documents in Canada and no one would contemplate setting one language (always the implicitly inferior language) in bold, italic, or another typeface.

A bilingual document must communicate the equality of both languages typographically. Abandon your plan, please, and typeset both languages in roman by default.

I agree with Joe Clark: don't treat one language as "secondary" or inferior to the other. You can use position to separate the English and French texts, without any change of typeface.

But in general, the question of using italics for text, and then dealing with "italics" within that text, comes up again and again. There's no ideal solution.

First of all, if you're going to set any lengthy text in italic, make sure that you're using a type family whose italic was *designed* for extended reading. An italic derived from the chancery italic handwriting (and fonts) of the 15th and early 16th centuries is usually much better for this than one derived from the more rounded italics of the 18th and 19th centuries, which were usually meant only as companions to the roman. That lets out Times Italic, for instance: we're all used to seeing it, so it's familiar, but it's really not designed for continuous reading. (In fact, I think that all by itself, Times Italic is responsible for a good deal of the prejudice against *ever* using italic for text.)

The usual practice for dealing with titles or emphasized words within an already-italic text is to set them in roman. That's logical, and expected; but it has always bothered me, because it's almost always ugly. When the "italic" words are sparse, and the typeface has the requisite characters, I sometimes use small caps for emphasis or titles within the italic passage. (All small caps, not caps & small caps.) Of course, if capitalization within those words is important, then this won't work. But in an old-style typeface, it can look quite elegant, and easy to read.

If you insist on using the italic, you can get a typeface with an upright italic. Nara, for example.

“Nara includes two different secondary scripts for situations where you need to emphasise a selection of text that is already italicised. The designer of Nara also explored the notions of what defines a secondary script. While the Cursive version is perfectly upright with a stroke structure inspired by handwriting, Italic uses a 12 degree slant.”

John: An italic derived from the chancery italic handwriting (and fonts) of the 15th and early 16th centuries is usually much better for this than one derived from the more rounded italics of the 18th and 19th centuries, which were usually meant only as companions to the roman.

Really? The rounded italics of the 18th Century were based on contemporary secretarial hands, just as the italic hands of the 15th and 16th century were, and both were used extensively -- in both written and typographical form -- for extended text. Bodoni, for example, set the entire introduction to his Manuale Tipographico in italics.

I think there is a good argument to be made -- and a testable one -- that the 18th Century neo-classical and romantic italics may be easier to read in extended text because a) they are more regular than many of the renaissance italics, having consistent slant angle and consistent stroke contrast patterns, b) the stroke contrast typically follows the slant angle, rather than being independent of it, c) letters are typically less cramped than in the more compressed renaissance italics and closer in proportion to roman types.

''A bilingual document must communicate the equality of both languages''

i agree, i decided to do it italic just because the original text was written in english.

and the french is a translation of the original. I could make them both roman.

The novel Birdy by the pseudonymous William Wharton, published in 1978 by Knopf, had a similar problem. The story was told by two different narrators, usually speaking in alternating chapters.
The book's designer (whose name, alas, I do not know), solved the problem by using W.A. Dwiggins' Electra font, designed for Linotype in the 1930s. Dwiggins' original design included a Roman and an Italic, but his Italic was an oblique Roman, generally considered easier to read in mass than the more conventional cursive italic. It was Dwiggins' original oblique Roman type that was used for the second narrator in Birdy.

P.S. No. 1: In the 1930s, Linotype's customers found Dwiggins' oblique Roman a little too avant-garde, so Linotype sent Dwiggins back to the drawing board to create a cursive version of Electra, and that's pretty much the only Electra Italic you see these days. Check around, though, I'll bet someone has made a digital version of the original Electra Italic.

P.S. No. 2: Birdy became a bestseller and was published as a mass-market paperback, set in another Dwiggins face, the then-ubiquitous Caledonia. Caledonia has a cursive Italic, and the Italic chapters in the mass-market paperback Birdy are damn near impossible to read.

If you have a good font clan, setting one language in serif and the other in sans might work, too.

If text blocks are flagged by typeface as language-specific, this indicates a shortcoming of the layout.
Ideally, the reader should be able to determine the conventions of the document design immediately, established on the first page, have this confirmed at the first page turn, and then be able to anticipate where his/her eye should go on subsequent page turns, without having to look for cues such as typeface, or having to start reading text to determine which language it is.

**

However, as to your point about the French being a translation, there is a case to be made for identifying the secondary status of translation. For instance, in a book of poems by Lorca (Penguin), the verse is set in standard manner, in Spanish, and the English translation is in footnotes, prose style. This respects the primacy of the original language, and the limits of translation.

Somewhat related, in a recently acquired album (Ozawa/TSO, late 1960s), I was intrigued to find that the English and French liner notes were written independently, with different information and critical perspective.

John: Really? The rounded italics of the 18th Century were based on contemporary secretarial hands, just as the italic hands of the 15th and 16th century were, and both were used extensively -- in both written and typographical form -- for extended text. Bodoni, for example, set the entire introduction to his Manuale Tipographico in italics.

I'm aware of the relationship between 18th/19th-century italics and contemporary roundhand, but I think that by the 18th century the connection between type and handwriting had gotten much looser than in the 16th. And as far as I know, italic type in the 18th century was mainly used as a secondary face, to complement the roman; not intended for use in long passages of text.

It would be interesting to try to test your hypothesis that 18th-century italics are easier to read than chancery italics. Maybe it's just my eye, but to me, every factor that you name in favor of the later style is one that makes it less readable.