spec language of calling ThrowCompletion seems strange

[FrankYFTang]

I think the language used in the spec text about calling ThrowCompletion() may need to be change via editorial PR

We now have

1. Let error be ThrowCompletion(TypeError).
   in https://tc39.es/proposal-intl-messageformat/#sec-selectpattern

a. Let error be ThrowCompletion(ReferenceError).
in https://tc39.es/proposal-intl-messageformat/#sec-resolveexpression

a. Let error be ThrowCompletion(ReferenceError).
i. Let error be ThrowCompletion(TypeError).
in https://tc39.es/proposal-intl-messageformat/#sec-resolvefunction

i. Let error be ThrowCompletion(ReferenceError).

1. Let error be ThrowCompletion(ReferenceError).
   in https://tc39.es/proposal-intl-messageformat/#sec-resolvevalue

but ThrowCompletion is expecting a an ECMAScript language value
see https://tc39.es/ecma262/#sec-throwcompletion
but TypeError and ReferenceError are "Global Name" as defined in
and https://tc39.es/ecma262/#sec-well-known-intrinsic-objects

Is "Global Name" an ECMAScript language value ?

Looks other part of ecma262. it seems the language in spec text is

1. Return ThrowCompletion(a newly created TypeError object).

i. Let error be ThrowCompletion(a newly created TypeError object).

a. Let error be a newly created TypeError object.

1. Let error be a newly created AggregateError object.
   etc

[ljharb]

Indeed, ThrowCompletion does not take a constructor, it takes an error object.