Language Log

Whether either

It's been a time of puzzles about whether and either: a case where either is used in place of standard whether, and an assortment of coordinations, of varying levels of oddness, involving these two items.


1. Concessive either.  First, either for whether -- in an ADS-L posting by Phil Cleary on 9 December, intended to document a possible eggcorn (advance someone's course for advance someone's cause) but incidentally providing an instance of concessive either X or not 'whether X or not':


It looks like advance ... course is a moderately common error -- an eggcorn, perhaps, or a flounder, a word confusion based on similar pronunciation (especially, in this case, in non-rhotic varieties) and overlapping meaning.  Here are a few more examples (found on 12/9):


That day I also found plenty of hits for {"either you like it or not"} in the appropriate sense, e.g.:


(You can find more examples with {"either you like it or you don't"} and other variants.)  I hadn't noticed this before, and it seems not to have caught the attention of the usage advisers, but it's definitely out there.

Now, this is not just "confusing" either and whether.  It has a basis, I think, in one of the ways you can use "either you like it or not" (equivalent to "you either like it or not"), namely to convey indifference: 'it doesn't make any difference whether you like it or not'.  (Similarly for examples with clauses other than you like it.)  That's just what whether ... or not conveys.  So either bleeds into whether's territory.

[Long digression: This would be a good place to point out that people who revile whether ... or not in complement clauses --


because the or not is "unnecessary" (Omit Needless Words, or ONW) don't always recognize that the or not is OBLIGATORY in standard English in concessive clauses:


Scrupulous usage advice recognizes the distinction.  (And sensible usage advice doesn't insist on the omission of or not.  MWDEU: "this use of or not is more than 300 years old and is common among educated speakers and writers.  It is, in short, perfectly good, idiomatic English.")

A further wrinkle: there are plenty of occurrences of the non-standard truncated concessive (without the or not), e.g.:


That would be like the truncated as far as of


which I've mentioned a number of times on Language Log (for instance, here).  Here we have two situations where vernacular speakers omit words because's they're needless in the context, but guardians of the standard insist that you must Include All Necessary Words (IANW).  The guardians' judgment is in fact based on social criteria -- who uses the variant, an antipathy to what's perceived as innovation -- but the criticism is couched in terms of "efficiency" (ONW) or "logic" (Two Negatives Make a Positive) or some other abstract principle.  The rationale is SECONDARY, as I termed it in my discussion of at about a while back.  Omitting needless words is ok only if you do it in tune with the prescribed standard.]

2. Correlative either ... either.  A few weeks before, inspired by the data I'll talk about in section 4 below, I looked for examples of correlative either ... either.  Searching on {"either you * or either"} on 15 November got many hits, among them:


I'd noticed correlative either in student writing for years.  It's non-standard, but not (so far as I can tell) proscribed in the manuals, although it would be a natural object for ONW scorn.

It's easy to see how it would arise.  The standard disjunctive correlative either ... or has an accented word, either, marking the first disjunct, but normally unaccented or in the second.  In speech you can put an accent on the or to convey that the alternatives are of equal status, but in writing that's hard to do.  Using either on both disjuncts does the trick.  Yes, it's non-standard, but it's communicatively effective, and you can see why people might like it.  (It's also likely that correlative whether, as in the next section, promotes correlative either.)

[More digressions.

In addition to disjunction marked with either, English has (of course) unmarked disjunction, as in


Here the two disjuncts have equal status, just as conjuncts do:


Somewhat surprisingly, there is little antipathy towards marked disjunction in the advice manuals, though you'd expect ONW to be invoked against either.  To my mind, this is a good thing: relentless ONW legislates implicit (rather than explicit) marking of content, and this is just an irrational prejudice.  Saving a word means a lot when you're writing telegrams or writing to some limit on word counts, but in plenty of other contexts that extra word does some work.  Explicit marking is often just what you want.]

Notice that the problem with correlative either ... either is not a failure of parallelism.  From the point of view of standard English, it's TOO MUCH parallelism.

3. Correlative whether ... whether.  Formally similar to correlative either ... either is correlative whether ... whether, as in these examples Neal Whitman sent me last month:

 
Neal found these examples (all concessive) a bit edgy, but I find them fully grammatical, and in fact I generally prefer them to coordination of full clauses WITHOUT the repeated whether, like:


This sentence is entirely grammatical, but I prefer the explicit marking on both disjuncts, because it gives equal status to the disjuncts.

[Yet another digression.  Either and whether have a lot in common, but the details of their distribution are significantly different.  Among these differnces is that in formal standard English either marking a first disjunct is generally optional, but whether in the same context is obligatory:


But in informal English, such concessives are just fine.  This is another sort of implicit marking, a kind of parataxis -- one sentence glommed onto another.  Such implicit marking, which depends on the hearer or reader supplying the connection between the two bits of content, is a good thing when the people involved are generally on the same page about what's going on.

The really big point here is that it's loopy to legislate in general against explicit marking (ONW) OR implicit marking (IANW) when both are available.  Each has its own virtues, depending on the context.]

Neal's examples are all concessives, but correlative whether is equally at home in complements:


4. Correlative subjects.  Now we get to some possibly WTF coordinations.  It all started when I heard the following:


(This was the impetus for my investigations in section 3, which in turn took me to the stuff in section 2.)

On 14 November I googled up a few more examples, among them:


These have correlative SUBJECTS marked with whether, in a Right Node Raising (RNR) configuration that goes beyond the ordinary, since it groups together a subordinator whether with the subject of its associated complement clause, as against the VP of that clause; the disjuncts are not constituents.

There are similar examples with other introductory WH elements (some involving disjunction, some conjunction):


And with if:

And with that:


(I haven't found any examples with either or with subordinating conjunctions like although and because.)

Non-constituent disjuncts/conjuncts are just what we expect in RNR: in


the object NPs money and support are grouped together with a preposition associated with another argument of their verbs, even though the two don't make a syntactic constituent.  RNR is like that.

My first reaction to the whether examples was to shrink back, and probably many people will find them unacceptable.  But I've grown to find these RNR examples that "cut into" clauses not so bad at all.  As far as I can tell, the advice literature hasn't noticed them.  (How would the manuals label them?  They're certainly not failures of parallelism -- once again, they are in a sense more parallel than they'd have to be -- though they could of course be faulted as violations of ONW, if you care passionately about omitting every single omissible word.)

5. Bonus WTF coordination.  And now a real WTF coordination case (involving disjunction, so there's a tenuous tie to the preceding sections), from Bob Ray on 21 November:


Ray noted that there was plenty of room on the form for a more complete sentence, in particular:


That is, the have is functioning both as a main verb (have Medicare) and as an auxiliary, in the perfect construction (have applied for Medicare).

This looks like a one-off error, the result of pasting together two formulations of related ideas, with have as the hinge, much as in the Elmore Leonard quotation Mark Liberman posted about earlier today:


The other cases I looked at above are systematic -- though maybe non-standard or in the gray area between standard and non-standard -- and not inadvertent errors.

Posted by Arnold Zwicky at December 20, 2007 10:30 PM