Perl 6

WITCH lcurly: Revision 2

^ Mnemonics

"code", "block", "enclose"

** primary use is to enclose literal code
** also used for hashes

^ Used as a brace pair or quote pair

{} "creates a Closure"<>
{} except with specific contents inside or as part of a variable term denotes "use for hashes"<http://perlcabal.org/syn/S04.html#Statement_parsing>
{} empty as part of a term allows hashes to be "interpolated"<http://perlcabal.org/syn/S02.html#Literals>
{} during hash declaration sets up "user defined indexing"<http://perlcabal.org/syn/S09.html#User-defined_array_indexing>
{} injects hash information in POD "delimited blocks"<http://perlcabal.org/syn/S26.html>
{} when used as a user supplied delimiter to a quote, match, or substitution, is merely a general quoting character

^ When used inside regexes

{} within a rule institutes an "action"<>
{} also in a "substitution"<http://perlcabal.org/syn/S05.html#Modifiers> injects code execution in a match
{N} or {N..M} postfix to a match term is a "repeat count"<http://perlcabal.org/syn/S05.html#Bracket_rationalization>
{{<{}>}} in "extensible metasyntax"<http://perlcabal.org/syn/S05.html#Extensible_metasyntax_(%3C...%3E)> allows injection of rules generated by code.
{{<?{}> or <!{}>}} also "extensible metasyntax"<http://perlcabal.org/syn/S05.html#Extensible_metasyntax_(%3C...%3E)> performs an assertion.

^ As an adverb

{} can be used as parenthesisers in adverbs

^ Old, deprecated, or other language uses

{N,M} is used as a repeat count in many regular expression implementations. In Perl6 that is {N..M}