I was typing out an explanation/justification to the effect that when speaking aloud, the punctuation at the end of the sentence is 'heard' in the last word, even if the word is quoted. We insert a pause before the start of the quote to indicate it was direct, but there isn't a way of indicating whether, say, a question mark belongs inside or outside the quotation.
But then I sounded out some examples in my head and realised it would be totally obvious. The inflection of the question is apparent from the beginning of the fragment (¿), not just from the inflection on the last word. I knew this already, of course, but I didn't really grok it, I guess because of the limitations of how we write out punctuation in English.
The intonation that people think of as characterizing questions is actually specific to yes/no questions. It can be the only feature that marks the sentence as interrogative: "You know him?" is a valid sentence, and it is distinct from the indicative "You know him."
There are also what I think of as "question word" questions, and you might think of as "fill in the blank" questions: "Why are you here?"
These do not necessarily have the intonation that applies to yes/no questions. Like yes/no questions, they are characterized by subject-auxiliary inversion (the word order constraint you mention), and also like yes/no questions, that inversion is not guaranteed to be present in a question of this type: "Who sent you?"
(Note that inversion is possible for "You know him?" ("Do you know him?") and impossible for "Who sent you?"; these are different phenomena.)
I would argue that the yes/no question is primarily marked by intonation and the question-word question is primarily marked by the presence of a question word. Word order is affected in both cases, but not the primary indicator of what's going on. (Compare "He said what to the king?" - again, inversion is possible here ("What did he say to the king?"), but not required. In this case, the inverted version of the question is unmarked (normal), and the uninverted version suggests that the speaker wishes to place a special emphasis on something.)
Re. Intonation: I actually distinctly remember my first grade teacher having us practice a rising tone at the end of a “question sentence”. (US, 1980)
Also I remember being taught quite rigidly the “A,Eric an” style of quotation-punctuation relationship, and I naturally rejected it from the beginning as illogical.
Sentences of the form <some_statement>? with rising intonation are common and mean roughly <some_statement>, right? or <some_statement>, yes? There's no marker other than the intonation when spoken or the question mark when written to indicate that it's a question.
But then I sounded out some examples in my head and realised it would be totally obvious. The inflection of the question is apparent from the beginning of the fragment (¿), not just from the inflection on the last word. I knew this already, of course, but I didn't really grok it, I guess because of the limitations of how we write out punctuation in English.