a bunch of the boys ?was/?were...

[Arnold M. Zwicky]

I was struck by the following sentence in a review (NYT Book Review,
2/27/05, p. 17) of JT LeRoy's "Harold's End" by Albert Mobilio, who
edits the fiction section of Bookforum:
-----
"Harold's End" is set in the parks and alleys of San Francisco, where a
group of teenage hustlers takes drugs and turns tricks.
-----

I would have gone for "take" and "turn" myself, but it's an arguable
point.

Ncollective + [ of + NPpl ] is taken up by MWDEU under the heading
"agreement, subject-verb: a bunch of the boys" (as in the immortal line
"A bunch of the boys were whooping it up in the Malamute saloon").  Its
advice is that when the sense is plural, as it usually is, the verb
should be too, though some grammatical sticklers insist that the PP is
just a partitive (as in "a vase of flowers") so the agreement should be
singular.

Now, there are some collective Ns that have been completely
grammaticalized as determiners and so are transparent to the number and
count/mass classification of the head N: a lot of shrubbery has thorns,
a lot of shrubs have/*has thorns.  For me, there are a few collective
Ns that are invariably heads: The committee/team was/*were working on
reports.  For me, the subject-verb agreement remains singular even when
a plural anaphoric pronoun is called for; The committee/team was/*were
working on their/??its individual reports.  (Others, especially British
speakers, have other judgments here.  CGEL, section 18.2 of chapter 5,
has an extensive discussion of the collective facts, taking "committee"
as the paradigm example of a noun allowing either agreement.)

In any case, for me, most other collective Ns permit either agreement,
though they strongly trigger plural agreement when the sense is plural,
as it is with taking drugs and turning tricks.

MWDEU displays several examples of the Mobilio sort, which it suggests
"may be the result of nervous copy editors or indecision on the part of
the writers".  Ah, the perils of instruction.

arnold (zwicky at csli.stanford.edu)