In my understanding, there is no problem with having a relative clause modify a noun that is the object of a preposition.
I addressed a roomful of professors who were eating lunch.
I gave the student a sampling of the books that were on the shelf.
I sorted through a box of tools that were used for grinding.
The first two work for me, also, but the third does not. When I get to 'that are' my brain stops to redefine the subject of 'are'.
Great examples! I've noticed something of interest. Tell me what you think: take a closer look at the "nouns",
roomful (derived: room that is filled )
sampling (derived: participle; verbal in nature)
If the head (i.e. sampling, roomful) is verbal in nature (i.e. derived from a verb), then of course there wouldn't be any argument at all since the relative pronoun would not be able to modify a verbal head,
I addressed a roomful of professors that was huge. :(
I gave the student a sampling of the books that was long. :(
That relative pronouns modify nominals, specifically nominals in nature, might be just what has speakers deeming the two examples above as semantically awkward.
If, however, the head is nominal in nature (i.e. box, book), then the relative pronoun would modify the closest nominal,
a box of tools that was used by my Dad. :D
[i]Note, the closest nominal, given the non-linear structure of language, will be the head of the phrase (i.e. box, book); That's possibly the reason behind the awkwardness of,
a box of tools that were used by my Dad. :(
a book of words that were used for teaching. :(
Given, the other meanings,
a box of toolsthat were used by my Dad. :(
a book of wordsthat were used for teaching. :(
I'd speculate with some weight that the speaker views the RC modification as an outcropping of the embedded noun phrase 'words'. That is, the structure is novel.
Speakers know intuitively that RCs modify nominals; that RCs modify true nominals, and not verbals (i.e. roomful, sampling); that a relative pronoun modifies the closest nominal, and that given the non-linear structure of language, the closest nominal within proximity to the relative pronoun is the nominal that heads the phrase (i.e. box, book). That's all intuitive. Structure such as a box of tools that were and a book of words that were deviate from the pattern, and hence have the reader/listener taking a double-take, sort to speak, in order to redefine the subject of the clausal verb. The structure is grammatical in terms of syntax, but in terms of semantics, one has to work out (i.e. what you had refer to in a previous post as logic) the meaning it expresses. That's a fairly big red flag. To me (and possibly others, given the heated debate you referred to) it seems a little odd or rather out of place that the speaker has to work out the meaning. I say that because Language change has always moved in the direction of creating a more efficient system. This change, in my opinion, isn't efficient, which brings to light the question, Will it survive in the system? We can describe it (i.e. it's an embedded node) and as language users if we hear it used a great deal, we too will pick it up (i.e. comprehend it), but, and here's the point, what about the time and energy it takes to differentiate which nominal within the phrase is the antecedent? Language is non-linear; there's a reason for that; adding this new change into the language adds a linear aspect. Hmm. Kewl. How will that alter the non-linear structure of the language (i.e. the way we think)? Or, if linear is there already, what does that say about how we acquire language?
You always provide the most appetizing food for thought! :D Very interesting examples, Mike! Thank you. :D
All the best,
And your analysis is, as usual, scholarly and logical. I will continue to think about the issue. :wink: