"He never calls subroutines "functions"; I'm not sure where that usage comes from, but certainly by BASIC and LISP there were "functions" that were at least implemented by subroutines."
I think the early BASIC's used the subroutine nomenclature for GOSUB, where there was no parameter passing or anything, just a jump that automatically remembered the place to return.
Functions in BASIC, as I remember it, were something quite different. I think they were merely named abbreviations for arithmetic expressions and simple one line artithmetic expressions only. They were more similar to very primitive and heavily restricted macros than to subroutines or functions.
FORTRAN had both functions and subroutines. A function returned a value and was invoked in an expression (eg. S=SIN(A)). A subroutine was invoked by calling it (eg. CALL FOPEN(FNAME, PERMS)).
In the flang-new compiler, which builds a parse tree for the whole source file before processing any declarations, it was necessary to parse such things as statement functions initially so that further specification statements could follow them. Later, if it turns out that the function name is an array or regular function returning a pointer, the parse tree gets patched up in place and the statement becomes the first executable statement.
I think the early BASIC's used the subroutine nomenclature for GOSUB, where there was no parameter passing or anything, just a jump that automatically remembered the place to return.
Functions in BASIC, as I remember it, were something quite different. I think they were merely named abbreviations for arithmetic expressions and simple one line artithmetic expressions only. They were more similar to very primitive and heavily restricted macros than to subroutines or functions.