Punctuation following an interruption with a dash
What you have written is incorrect, several times over. First, you never follow a dash with punctuation; it simply isn’t done. Second, you didn’t use a dash there, and you should have done so. These are your three main choices here, with some variation in number 2:
-
Unspaced em dash:
- You didn’t use a dash there—and you should have done so. (no space)
-
Spaced em dash:
- You didn’t use a dash there — and you should have done so. (normal space)
- You didn’t use a dash there — and you should have done so. (shorter space)
-
Spaced en dash:
- You didn’t use a dash there – and you should have done so.
Generally speaking, and not without exception, American publishers for the most part use em dashes spaced or flush, while British publishers tend to use spaced en dashes. Use whichever style your publisher or journal tells you to use, and use it for all instances calling for a regular dash.
Note that ranges take unspaced en dashes, as in 10–20 years go. So too do equal compounds, like the Boyer–Moore algorithm. Don’t use hyphens for either of those.
With a few exceptions, the general rule is that dashes separate things whereas hyphens join things. Below are lines of five each of em dashes, en dashes, minuses, and hyphens, all separated by a normal space.
- — — — — — (em dashes)
- – – – – – (en dashes)
- − − − − − (minus signs)
- ‐ ‐ ‐ ‐ ‐ (hyphens)
Here are the code points for the Unicode dashes that are of the Common script, along with their general categories. Do not use the first one to mean a dash; it never does.
U+002D - GC=Pd HYPHEN-MINUS
U+058A ֊ GC=Pd ARMENIAN HYPHEN
U+2010 ‐ GC=Pd HYPHEN
U+2011 ‑ GC=Pd NON-BREAKING HYPHEN
U+2012 ‒ GC=Pd FIGURE DASH
U+2013 – GC=Pd EN DASH
U+2014 — GC=Pd EM DASH
U+2015 ― GC=Pd HORIZONTAL BAR
U+2053 ⁓ GC=Po SWUNG DASH
U+207B ⁻ GC=Sm SUPERSCRIPT MINUS
U+208B ₋ GC=Sm SUBSCRIPT MINUS
U+2212 − GC=Sm MINUS SIGN
U+2E17 ⸗ GC=Pd DOUBLE OBLIQUE HYPHEN
U+2E1A ⸚ GC=Pd HYPHEN WITH DIAERESIS
U+2E3A ⸺ GC=Pd TWO-EM DASH
U+2E3B ⸻ GC=Pd THREE-EM DASH
U+301C 〜 GC=Pd WAVE DASH
U+3030 〰 GC=Pd WAVY DASH
U+30A0 ゠ GC=Pd KATAKANA-HIRAGANA DOUBLE HYPHEN
U+FE31 ︱ GC=Pd PRESENTATION FORM FOR VERTICAL EM DASH
U+FE32 ︲ GC=Pd PRESENTATION FORM FOR VERTICAL EN DASH
U+FE58 ﹘ GC=Pd SMALL EM DASH
U+FE63 ﹣ GC=Pd SMALL HYPHEN-MINUS
U+FF0D - GC=Pd FULLWIDTH HYPHEN-MINUS
Not also that U+00AD, SOFT HYPHEN
, is neither a dash nor a hyphen, and is in fact a control character.
Beyond the most basic considerations, punctuation is governed more by (often conflicting) style manuals and ‘house’ rules in expository prose, and by personal preference in fiction, than by universally accepted rules.
Few, if any, publishers would put spaces around an em dash†; In print and other carefully formatted text, dashes are strictly distinguished from hyphens; but usage is different on the internet, where few users have the knowledge or patience to employ non-ASCII characters.
As for your two sentences, which appear to be contrived for fictional contexts:
- My own inclination in both of these would be to use parentheses rather than dashes.
-
In both I would recast so the parenthesis did not break the connection between the main clause and its punctuator. The second is easy:
What on earth could he be expecting from her? (She had long ago ceased to believe he cared for her.)
The second is tricky. The problem is that you want to hang the parenthesis off of surprising, but that creates an unacceptable deferment of what surprised him. You run the danger that your reader will lose her sense of the syntax before she gets to the car.
One strategy is to hang the parenthesis off the car—after all, how surprising is it to see a blue car on the street?:
He was walking down the street when he saw something quite surprising: a blue car. (Being easily surprised was one of his characteristics.)
Another would be to recast so the parenthesis and its hook come earlier:
Walking down the street he was surprised (he was easily surprised) to see—a blue car! What the devil was that car doing here?
But context is everything.
EDIT:
† I am happy to see tchrist's correction and expansion of this. I don't like unspaced disjunctive dashes, either em or en, so I probably think them to be more universal than they in fact are. My preference is for the dash opening a parenthesis to be preceded by a space, and that closing a parenthesis to be followed by a space; but that is very idiosyncratic, followed as far as I know by nobody else.