4.7.1 Direct Element Constructors

Norm

"We start with the rules for normalizing a direct element constructors'
content."
    You'd get a more logical flow (top-down) if you started with rules
    3 and 4 (and most of the paragraph that precedes them).

"a direct element constructors' content."
    s/s'/'s/

"Literal XML character data (CDATA)"
    It would be better to say "CDataSections", if that's what you mean.
    (And if that's not what you mean, what *do* you mean?)

"is assumed to be processed directly at parsing level so it does not
require any formal treatment."
    It's not entirely clear what "processed" means.

"An element-content unit is ..."
    Make the sentence a definition?
    At least emphasize the phrase "element-content unit" somehow.

"a contiguous sequence of literal characters"
    Note that "contiguous" doesn't mean "maximal", which I think you want.

"literal characters (character references,"
    Insert "including" after open paren.

Example: "<name>Dizzy Gillespe</name>"
    "Gillespie", if you mean the trumpeter.

"After boundary-space is stripped"
    s/boundary-space/boundary whitespace/
    ("Boundary-space" is the name of the declaration and the policy, but
    not the stuff that gets stripped.)

"It contains one XML comment, followed by one enclosed expression ..."
    Occurrences of "one" in this sentence sound odd. Change to "a"/"an".

Norm / rule (1|2)
[[]]_ElementContent-unit
    This construct only appears on the LHS of mapping rules, so how would
    it ever be invoked? Perhaps, in Norm / rule 3 / RHS, the
    'ElementContent' subscript should be 'ElementContent-unit'.

    Use of the normalization-subscripts 'ElementContent-unit' and
    'ElementContent' seems to be backward.  That is, I would expect the
    'unit' subscript to be concerned with normalizing content units, and
    the other with normalizing the whole content. Swap 'em!

    Also, 'Unit' would be more consistent than '-unit'.

Norm / rule 2
"n > 1"
    This uses undeclared syntax for tacking a premise onto a normalization
    rule.

Norm / rule 2
"fs:item-sequence-to-node-sequence( ... )"
    The declaration for this function says that it takes a single
    argument (of type item*), so there should be an extra pair of
    parentheses around its current args. (As in the example following the
    rule.)

"We need to distinguish between multiple element-content units ..."
    If this is supposed to hark back to the second sentence of the
    previous para, change it to something like:
        "We need to distinguish the results of consecutive
        element-content units ..."

"the rule for converting sequences of atomic values into strings apply"
    s/apply/applies/

Norm / rule (3|4)
    Change "AttributeList" to "DirAttributeList".

    Also, change "[[ QName ]]_Expr" to just "QName".
    (The only LHS that the former matches is 4.2.4 / Norm / rule 3,
    and you don't want it to.)

"2. that character references have been resolved to individual characters
and predefined entity references have been resolved to sequences of
characters"
    Why would a PER resolve to a *sequence* of characters?  Change to:
        "2. that character references and predefined entity references
        have been resolved to individual characters"

    But actually, rule 5 looks like it *doesn't* assume this, since it
    includes mention of CharRef and PredefinedEntityRef.

"3. that the rule is applied to the longest contiguous sequence of
characters."
    This makes it sound like, "of the many sequences of characters, the
    rule is only applied to the longest one".

    Anyhow, this assumption appears to forget that the element's content
    has already been partitioned into element-content units, which
    guarantees that the assumption holds. So you can delete it.

Norm / rule 6
    Italicize 'DirPIConstructor'.

Norm / rule 7
    Italicize 'DirCommentConstructor'.

(completeness)
    You're missing a rule for [[ DirElemConstructor ]]_ElementContent,
    which probably just maps to [[ DirElemConstructor ]]_Expr.