Take the example of
There is very little that a conforming POSIX.1 application can do by catching, ignoring or masking SIGSYS
(From the SIGSYS article)
This can be rewritten as
There is very little that a conforming POSIX.1 application can do by the catching, ignoring or masking of SIGSYS
Both the and of must be added or the sentence becomes ungrammatical. Is there a grammatical explanation for this?
Morning! The below is just an hypothesis, but it sounds convincing enough to me.
A gerund is special kind of word: it is both noun and verb at once (just as a participle is both verb and adjective). In its function as a verb, it can have an object:
Augustus condemned his daughter's adultery.
By condemning Julia, he set an example for the Empire.
But it can also have an article and an of attribute like most nouns:
The public condemnation of his own daughter was part of his new policy of chastity.
His laws punishing adultery would have been hypocritical without the condemning of Julia.
Whenever the is used, the gerund is marked as a noun; that is probably why it cannot have an object then, since nouns normally can't have objects. This forces the secondary argument of the word to turn into an of attribute.
Conversely, whenever it has an object, it is marked as a verb, so that it cannot have an article. If there is no secondary argument, the article is free.
Whenever of is used, it is marked as a noun, just as with the. Even though nouns can normally exist without articles, somehow of usually forces the gerund to take the article. Apparently the pattern the + gerund + of became dominant enough to render no article + gerund + of unidiomatic. I can't really say what could have caused this, except that the lack of article may somehow mark it as a verb in this case. Perhaps some more pondering will bring inspiration.
The two sentences you have written are not exactly the same in meaning.
"There is very little that a conforming POSIX.1 application can do by catching, ignoring or masking SIGSYS"
Here catching, ignoring and masking are gerunds and they function like verbs. In other words they have a subject (a conforming POSIX.1 application) and an object (SIGSYS). It is clear that the specific application performs these acts.
"There is very little that a conforming POSIX.1 application can do by the catching, ignoring or masking of SIGSYS"
The gerunds function as nouns here. In order to attribute genitive to nouns we use 's or the preposition of, the latter being the case in this sentence. It is not clear who or what does the catching, ignoring or masking to SIGSYS since nouns can't have a subject like verb forms (unless of course there is a technical reason I am not aware of which allows everyone to understand it is the specific application which performs the above mentioned acts and nothing else). In the meantime, the has been introduced to mark this change of function in the sentence. Without the, all three gerunds behave like verb forms and of is ungrammatical.
There is no such thing as a gerund in present-day English (CGEL p. 1220-1222). They are only present participle verb forms (or gerund participles if you wish) and (gerundial) nouns. Verbs take objects and nouns don't. Nouns take determiners and verbs don't.
So you have catching chickens was hard work, where the subject is a non-finite clause headed by a present participle with the chickens as its object.
This the same catching in he is catching chickens
or that in he hurt himself catching chickens.
The alternative is the catching of the chickens was hard work in which the subject is the noun phrase the catching of the chickens. In this case, of the chickens is the complement of the noun catching. Like other singular countable nouns, This requires a determiner; thus we have the catching...
I want to try to analyze this without the onerous term "gerund". That's just a distraction, I think. I think what this is about is ambiguity resolution. Let's consider the 4 possibilities of article + preposition:
1 The Taming of the Shrew
2 Taming the Shrew
3 *The Taming the Shrew
4 ?Taming of the Shrew
1 and 2 sound fine but perhaps have subtly different meanings. 3 is right out. 4 is iffy but all right. The question here is, why does this situation obtain?
Look at 3, which I think we can all agree is ungrammatical. Here the word 'taming' must not be acting as a verb form. If it were a verb, then the transitivity of tame would allow 'the shrew' to be a direct object. But here, we know that 'taming' is not a verb because it has a determiner, the article 'the'. Determiners modify nominals. So a determiner with an -ing word means we are interpreting the -ing word as a noun.
[Det] [-ing word] => [-ing word] = [N]
But this means 3 must be of the form
[Det] [N] [Det] [N]
Which we don't know how to interpret as grammatical.
Now consider 1. This is straightforward:
[Det] [-ing word] [Prep] [Det] [N] => [Det] [N] [Prep] [Det] [N]
That is, a noun phrase modified by a prepositional phrase. Easy.
When there is no [Det], we by default interpret an -ing word as a verb.
[-ing word] => [V]
which means 2 is like:
[-ing word] [Det] [N] => [V] [Det] [N]
which we interpret as a verb and its direct object, based on default SVO word order assumptions.
4 is interesting since we could have 2 paths for resolving the ambiguity.
[-ing Word] [Prep] [Det] [N] => [V] [Prep] [Det] [N]
which could be just a V with a prepositional object. Except transitive tame takes a plain DO (direct object), not a prepositional object, so it's not quite right. But:
[-ing Word] [Prep] [Det] [N] => [N] [Prep] [Det] [N]
could just be a noun phrase modified by a prepositional phrase again. The first noun just doesn't happen to have a determiner, which is ok. So 4 is easy to interpret.
So I think, again avoiding loaded terms like gerund, that in English we want to be able to assign N or V categories to words, and then semantic roles like DO based on those categories. When we can't do so unambiguously, we call the sentences ungrammatical.
