Re: To non-native English writers: expunging the solecism "This allows to do something."

[Stephen Berman]

On Mon, 25 Jan 2016 10:35:55 -0500 Richard Stallman <address@hidden> wrote:

> [[[ To any NSA and FBI agents reading my email: please consider    ]]]
> [[[ whether defending the US Constitution against all enemies,     ]]]
> [[[ foreign or domestic, requires you to follow Snowden's example. ]]]
>
>   > This does, however, also return several false positives, most of them
>   > with "require" (one is written as `require' and one or two of the others
>   > should probably be quoted like that).
>
> I don't know for certain what you found, but "require to do" has the
> same English usage rules as "permit to do" or "allow to do": the
> entity that will do it must be stated.

Several of the false positives are use-mention ambiguities arising
because "require" isn't quoted as `require', e.g.:

./lisp/ChangeLog.2:3416:        Use require to load x-mouse.
./lisp/ChangeLog.17:23939:      Tweak requires to silence compiler.
./lisp/ChangeLog.11:1159:       * calc/calc.el: Move require to end of file.

(The latter is of course not even syntactically an instance of the
solecism ("to end" is not an infinitive), but was caught by the regexp's
wide net.)

The object of "require" can also be relativized, leading to another
false positive, in fact, one found several times in the Emacs sources,
since it occurs in the GFDL appended to every info manual:

./info/todo-mode.info:1946:     material this License requires to appear in the 
title page.  For

(Though this usage is different from the typical solecisms, since the
object here is not agentive.)

Sometimes the object of "require" can be elided, if it's clear from the
context; this is quite common in "ChangeLog-Speak" and leads to another
false positive:

./lisp/gnus/ChangeLog.3:7094:   (time-date): Require to make some autoload 
issues on XEmacs go away.

But this is not an instance of the solecism, since the infinitive is not
the complement of "require" but an adverbial purpose clause (like the
first two false positives above).

Steve Berman