subject-verb agreement

[Arnold M. Zwicky]

the latest addition to my (large and still growing) collection of
english usage manuals, handbooks, and dictionaries -- Fowler [H.
Ramsey, not H.W.] & Aaron, The Little, Brown Handbook (8th ed., 2001)
-- arrived in the mail not long ago, and when i unwrapped it, it fell
open to p. 337, where the following rule for subject-verb agreement
(the fourth of eleven such rules, some with several subparts) is
enunciated:

   (0)
   "When parts of a subject are joined by _or_ or _nor_, the verb
   agrees with the nearer part."

there are three clauses:

   (1)
   "When all parts of a subject joined by _or_ or _nor_ are singular,
   the verb is singular; when all parts are plural, the verb is plural"

   (2)
   "When one part of the subject is singular and the other plural,
   avoid awkwardness by placing the plural part closer to the verb
   so that the verb is plural"

   (3)
   "When the subject consists of nouns and pronouns of different
   person requiring different verb forms, the verb agrees with the
   nearer part of the subject.  Reword if this construction is awkward"

now, this is a topic i've thought about, off and on,for some time, so i
read this advice with some interest.  six comments.

(a) the handbook doesn't actually say what the problem is, and the way
the solutions are presented conceals it.  the fact is that clause (1)
(slightly amended, to get the person issue out of the way) is utterly
uncontroversial; i'm not aware of variation on this point, and i doubt
that anyone needs to be told what to do when confronted by disjunctive
subjects of the same number (and person).  clauses (2) and (3),
however, treat circumstances where there is actual variation, and where
writers are often genuinely puzzled; here's where the problem lies, in
disjunctive subjects that differ in number or person (with verbs that
show these differences morphologically).

(in general, i think it's a mistake for manuals to treat
uncontroversial facts and disputed usage as covered by "rules" of the
same sort.)

now, extrapolating from the clear cases in (1), we can frame the
general principle:
   Disjunctive Agreement (DA): with a disjunctive subject, the verb
agrees
     with *each* of the disjuncts.
DA covers the cases in (1) and also (correctly) predicts that the cases
in (2) and (3) are problematic -- because they give rise to conflicts
as to which verb form to choose:
   Either my friends or my next-door neighbor ?are/?is going to have to
help.
   Either you or I ?are/?am going to have to help.
DA also makes sense semantically; each disjunct is interpreted
separately as the subject of the verb (in contrast to conjunction,
where the conjuncts taken together are interpreted as the subject of
the verb).

why doesn't the handbook explain at least some of this?  no way to tell
for sure, but possibly because of the unclarities in its notion of
"subject".  for the purposes of rule (0) (and for the corresponding
rule for conjunction), the subject is the *whole* subject, which has
"parts"; by talking merely about "parts", instead of saying that these
subjects are compound and have multiple subjects as their parts, the
handbook avoids the whole area of constructs of type X that have
multiple parts of type X, a notion that many people find mind-bending.
in addition, it's quite clear from the discussion of other examples
that elsewhere what counts as the subject is a single word -- the *head
word* of the subject, in modern syntactic terminology --  and the
authors of the handbook might have wanted not to draw attention to
their shifting use of the word "subject", so they didn't delve into
details.

in any case, it would probably have been useful to explain to the
reader that the problem in situations (2) and (3) is a conflict between
conditions imposed by the grammar of english, and that nothing else in
that grammar resolves the conflict.

(b) the handbook tries to cover all the situations might arise, and
with a single principle.

reasoning on first principles, you might think that if there is a
conflict between conditions, then the construction is simply blocked.
on the other hand, blocking frustrates the expression of content, so
that reasoning on other first principles, you might think that
conflicts are usually resolved.  indeed, Optimality Theory takes off
from the claim that one condition usually "wins", outranks the others
(so that expressibility is served as much as possible).  OT-style
analyses are natural when the conditions in question can be ranked with
respect to one another -- but the problem with DA is that we have no
independently motivated way of picking out the winning disjunct, so
that if there's to be resolution, it must be on the basis of some
*extra* principle.  the handbook provides such a principle -- agreement
with the nearest disjunct (AND) -- that is supposed to cover all the
cases, including the uncontroversial cases in (1).

now, there is some precedent for claiming that DA conflicts are
irresolvable.  charles fillmore suggested, many years ago, that some
conflicts (these in particular) produced gaps in the predictions of the
grammar, analogous to paradigm gaps in morphology.  there are some
things you just can't say.  and, in fact, a great many speakers are
deeply uncomfortable with *any* resolution of the conflicts in (2) and
(3).  given that, the best advice to a careful writer would probably be
to avoid the situations that give rise to DA conflicts.  (there's no
point in being "technically" correct, if what you write is going to
annoy or baffle a significant number of your readers.)

(c) the handbook assumes that there is exactly one right way to do
anything.  not only doesn't it allow for zero right ways -- blocking --
but it doesn't allow for alternatives, either.  yet when you collect
judgments from speakers, some of them are equally happy with more than
one resolution of DA conflicts (and different speakers sometimes prefer
different alternatives).  alternatives are all over the place in
languages, and it seems gratuitous to insist that the *grammar* of a
language should always pick out One Right Way.

(d) there is evidence, in the handbook's own discussion, for a
resolution principle different from AND.

the crucial point here is that clause (2) has the writer *reword*
sentences so that sg+pl disjuncts will have plural agreement:
   "Awkward   Neither the owners nor the contractor agrees.
    Revised     Neither the contractor nor the owners agree."
that is, the handbook prefers agreement with the higher number (AHN),
but only in combination with the resolution principle it prescribes,
AND.

the fact is, when you collect judgments (and examples from corpora)
some speakers resolve via AND, regardless of the order of the
disjuncts:
   Neither the owners nor the contractor agree.

indeed, some speakers resolve by agreement with the higher person
(AHP), so that instead of rewording to avoid "awkwardness" in the way
that the handbook prescribes:
   "Awkward   Either Juarez or I am responsible.
    Revised     Either Juarez is responsible, or I am."
the speakers can resolve without rewording:
    Either Juarez or I is responsible.

(e) there is evidence, in speaker judgments and examples from corpora,
for still another solution to DA conflicts in the singular: defaulting
to 3rd person sg.  that is, examples like the following can be found,
and some speakers find them acceptable:
   Neither you nor I is responsible.

though the evidence from actual usage is quite complex, the handbook
recognizes only AND.  now, agreement with the nearest (disjunct or
conjunct), case-marking by the nearest, and determination of governed
verb form by the nearest are all attested in various languages as the
*standard* resolutions of conflicts, so that AND for english wouldn't
be extraordinary from a crosslinguistic point of view.  but then
neither would principled resolutions of other sorts, or defaulting, or
blocking, for that matter.  all are attested, and often there's dialect
variation with respect to the details.  the point is that AND is in no
way privileged on the basis of first principles or crosslinguistic
considerations.  nor is it strongly supported by the evidence from
actual usage.  as i said above, the best advice to the careful writer
is probably: don't go there!

(f) the handbook falls back on appeals to "awkwardness" (cited above).
  similar appeals -- to "euphony" or "smoothness" or "naturalness" or
what "sounds right (or wrong)", that is, to some sort of sprachgefuehl
-- are astonishingly common in works of advice for writers.  (writers
are fairly often told to avoid stranded prepositions -- unless they'd
sound right!)  they are problematic: after all, if the readers already
had a feel for what is awkward vs. effective, smooth, etc., they
wouldn't need the handbook.

now, the handbook does provide (a few, well, two) exemplars of what
counts as "awkward" in the world of DA sentences, but provides no hint
as to how to extrapolate from them.

i understand that writing is an art and craft, and that teaching by
exemplars ("do as i do" or "do as she does") is both honorable and
effective.  in that context, it's best to avoid rigid arbitrary "rules"
in most situations.  you may choose to accept such conventions, as when
you propose to write a sonnet or a haiku, but otherwise it's a matter
of developing a feel for what is likely to work and what isn't.  in
this context, some analysis of the nature of the choices involved can
be very useful to the learner.

*but*, as i've said here several times before, it's a bad move to
elevate advice about effective choices to claims about the grammar of
english -- in this case, to insist that AND (or any other of the
schemes for dealing with DA conflicts) is just a rule of english
grammar.

arnold (zwicky at csli.stanford.edu)