Re: Bug#795158: explain spelling-error-in-description for 'allow to'
Niels Thykier wrote:
>> The spelling-error-in-description correction "Allow to" -> "Allow one
>> to" is hard to understand without further explanation.
> @English: Do you have a suggestion for how we explain this simpler than
> the below? I suspect the long (sentence theoretical) explanation is not
> going to help the average reader.
> My understanding of it is: Allow (when used with "to") is always a
> "transitive verb". That is, it must "apply" to an "object".
> Consider the following sentence:
> I allow my dog to go outside.
> Here "allow" is a "transitive" verb, which is applied to "my dog", which
> is the "object" (sentence structure wise). What the correction is
> complaining about is that it sees an instance of:
> I allow to go outside.
> This sentence is "invalid" and is missing something. It could have been
> a "passive voice" missing a verb and in wrong tense (e.g. "I am allowed
> to go outside") or it could be missing an "object" (a la the previous
> example above).
Yes - "allow to open files" is missing an object. Most English
transitive verbs are pretty easygoing about being turned into
intransitives like this, but "to allow" is one of the exceptions
(along with e.g. "to merit" - you can't say "does this merit?" - or
"to recognise - you can't say "I recognise").
It can be hard to notice the absence of an object with "allow" since
it can get one in any of several ways -
• "this option allows you to open files"
• "this option allows opening files"
• "this option allows files to be opened"
(That third one's particularly silly if you imagine the files begging
to be opened...)
It also gets some camouflage from the fact that "allow to open files"
is already missing a *subject*, which readers might take to imply that
this is the sort of context where conciseness matters more than
grammaticality. (In fact, no, it's just phrased as an imperative, and
imperatives regularly drop their subject.)
Then, some regions of the US allow "allow" as an intransitive meaning
"believe", but this is strongly dialectal.
You can often get away with it, just as you can get away with
pronouncing "world" as "vurlt", but it makes you really easy to spot
as a non-native-speaker. When I'm doing documentation reviews on
debian-l10n-english, a package description that starts with "This
package allows to..." is usually the first sign that I'll have lots of
work to do.
> Note: In the suggested correction, lintian always uses "one" as the
> "object". I am not sure if a place-holder might have been better. E.g.
> "allow to -> allow <missing-word> to"
That proposed replacement text with "one" is probably the most likely
quick fix to work in the largest number of cases, but it does also have
the disadvantage of introducing a use of "one" which is slightly
clunky (especially if the text has already been referring to generic
users as "you").
Using "<missing-word>" has the drawback that it makes it even harder
to guess just what's supposed to be missing. Are you saying it needs
to be "allow trying to"?
It occurs to me that you might try:
"allow to -> allow users to"
which is likely to work about as often. Mind you, when this comes up
on d-l-e my preferred fix is usually to try to eliminate the word
"allow" from the text completely, since it's almost always useless
padding, and often technically untrue - I was already *allowed* to
open files before this package came along! And indeed this menu
option isn't proposing to change the user's authorisation levels.
>> With openscad, it reports the "Allow to open multiple files" menu line.
> Which, to my understanding, grammatically does not make a lot of sense.
> If it is a menu line, it might make sense to word it slightly
> different. An example could be:
> "Open multiple files"
> Menu lines tend to "imperative" anyway ("orders" to the computer).
(Which as I was saying is why there's no explicit subject.)
>> I've checked with several native speakers and a dictionary, and nothing
>> gives me an indication on 'allow to' being wrong; at best, people have
>> suggested "Allow opening multiple files", but on a "either would work"
> I am not sure there is a general consensus that "either would work" here.
> * @English, what is your take on this?
I've given this particular non-nativism pride of place on my "d-l-e
style guide" page:
>> The report overview shows that the 'allow' case makes up roughly half of
>> the spelling-error-in-description cases.
>> Could you clarify on this?
> Presumably, people are not in a hurry to fix spelling mistakes or/and
> (like you) were unsure on how to fix the particular spelling mistake.
Also, an awful lot of packages get uploaded once with halfhearted
package descriptions and then left to rot. If lintian could detect
missing definite articles, the overview for that would show errors in
lots of packages that nobody would ever bother NMUing.
JBR with qualifications in linguistics, experience as a Debian
sysadmin, and probably no clue about this particular package