December 30, 2007


Mark Liberman and I have gone one round on blame Y on X, thanks to the 1915 Funk & Wagnalls Faulty Diction booklet, which categorically rejects the usage.  Through the mediation of MWDEU we were led to the apparent source of the peeve, Alfred Ayres's 1881 book The Verbalist, and now we're into meta-matters -- not so much about the syntax of blame, but about the advice literature on it.

1. Why this antipathy to innovation?  I'm only going to scratch the surface here (see note below on the placement of only), and very briefly, but at least eight attitudes and beliefs might contribute to this antipathy.

[Note: please don't write me about how only is "misplaced" here, because it should be right next to the thing it modifies, scratch the surface.  I mean, think about it: putting only right next to scratch the surface gives you going to only scratch the surface, which is a split infinitive even I am uncomfortable with, and locating the only before to scratch the surface interrupts the very tight idiom of prospective be going to -- going only to scratch the surface, ugh -- so only going to scratch the surface is by far the best version.  In any case, I'm not at all set against "high" placement of only and even (preceding a VP that contains the modified constituent), as in I only saw one dog 'I saw only one dog'.]

(1) Belief: innovation of new variants is pointless elaboration.  Why invent new expressions when old ones do fine?  In the case at hand, why concoct blame Y on X when we already have blame X for Y?  (There are answers to these questions, which I'll sketch later.  Here I'm trying to unpack the reasoning of usage critics like Ayres.)

(2) Belief, lying behind (1): for the most part, alternative expressions are paraphrases, differing at most in style.  In particular, blame Y on X and blame X for Y are (it is assumed) paraphrases.

(3) Belief: when alternative expressions differ in style, a "higher" variant (standard rather than non-standard, of course, but also formal rather than informal, written rather than conversational, and in general use rather than restricted socially or geographically) is intrinsically better than a "lower" variant and therefore should be preferred.  So insofar as blame Y on X is perceived as being "colloquial" (informal and conversational) or restricted to certain groups of speakers, it is to be avoided.

(4) Attitude: variation should not be tolerated.  In any context, there should be One Right Way.  In particular, either blame Y on X or blame X for Y, but not both.

(5) Attitude: the past should be preserved.  In particular, blame X for Y should continue in use.  Putting this together with (4): blame Y on X should not be tolerated.

(6) Belief: innovations threaten older alternatives.  Even if you don't subscribe to (4), this belief in combination with (5) means that innovations are to be rejected.  In particular, blame Y on X should not be tolerated, because it threatens the older variant blame X for Y.

(7) Belief: most innovations arise "from below", from the working class, the uneducated, the uncultured, the frivolous young, and so on, and so are to be resisted on this basis alone.

(8) Belief: most innovations arise from ignorance or laziness or both, and so are to be resisted on this basis alone.

[Note: "Alfred Ayres" is a pseudonym used by Thomas Embly Osmun, described in his 1902 NYT obituary as an "elocutionist and critic of dramatic expression".   With Richard Grant White, one of the great American grammar/usage ranters of the 19th century.]

2. Why innovate?  And why tolerate alternatives?  There's something a bit off-center in all of the attitudes and beliefs above, but here I'll look at just a few, beginning with the question of why people don't just leave the language alone, why people innovate.  Why don't they just preserve older forms, as in (5) above?

One part of the answer is that people are always trying to find ways to express what they want, and they're willing to stretch things a bit for their purposes.  Everyday conversation is full of language play of all kinds, novel figures of speech (metaphors and metonymies), other extensions of meaning, extensions of syntactic patterns, exploitations of implicature, intrusions from other varieties, and more more.  So is more elevated speech and writing.  Everybody innovates, all the time.  And that's a good thing.  (Of course, people also use a lot of expressions as wholes, formulas, and prepackaged routines.)

Another part of the answer is that people can't possibly know what the language as a whole is like, so that in an important sense much of the time they won't know whether they're innovating or just using existing patterns.  After all, people say lots of things they've never heard before, or don't recall having heard before.  If there's some backing for a usage, then go for it!

But on to specifics: diathesis alternations in English involving direct objects (DOs) and another complement.  Diathesis alternations are alternative distributions of syntactic arguments of heads (I'll stick to verbs here), where arguably the "same" head -- that is, a head with the same semantics, phonology, and morphology -- occurs with different sets of syntactic arguments:

Kim ate something.  [SU x, DO y]  (where SU = subject)
Kim ate.  [SU x]

Standard English has many patterns of diathesis alternations for verbs (Beth Levin's English Verb Classes and Alternations is a significant beginning of an inventory of them), including a number in which SU remains constant but DO can be switched with PO (prepositional object), among them:

the verb present (recipient x, thing presented y):
  I presented Kim with an award.  [DO x, PO y, P = with]
  I presented an award to Kim.  [DO y, PO x, P = to]

smear-class verbs (material x, location y):
  They smeared paint on the wall.  [DO x, PO y, P = on]
  They smeared the wall with paint.  [DO y, PO x, P = with]

"dative alternations (2O is "second object", not marked with P)":
  transfers (recipient x, thing transferred y):
    I gave Kim an award.  [DO x, 2O y]
    I gave an award to Kim.  [DO y, PO x, P = to]
  benefactives (beneficiary x, thing affected y):
    I baked Kim a cake.  [DO x, 2O y]
    I baked a cake for Kim.  [DO y, PO x, P = for]

(Please don't write me to tell me about more types -- these are merely illustrative -- or about further details of these types.  The literature on diathesis alternations is staggeringly huge, even just for English, even just for some of the types, like the dative alternations.)

I have to stress here that these alternations are entirely standard in modern English and have been so for some time.  The point is that speakers of English have every right to assume that there can be different ways of "saying the same thing" via different deployments of DO and PO.

But of course -- contra (2) -- these are not quite ways of saying the same thing.  The truth conditions for the alternatives might (or might not -- there are disputes about particular cases) be the same, but the alternatives are not otherwise equivalent; in particular, they do different things in discourses.  This is not just pointless elaboration (cf. (1)).  In contrast to the assumption in (2) that variation is (except perhaps for style) free, I've repeatedly argued that variation is typically unfree: there are contexts in which the alternatives do different things (even if they're mostly interchangeable) -- a moderated version of Dwight Bolinger's position that there is no formal difference without a semantic difference.  (Some discussion here, here, here, here.)

In all of the DO alternations, including the blame case, there's a difference in the status of the DO, versus PO or 2O: it's closely associated syntactically with the V, leading to the expectation that its referent is in some way "focussed on" or foregrounded; and it precedes the other (PO or 2O) argument, leading to the expectation that its referent is a given rather than a new item in the discourse (old before new).  In addition, the linear order favors short NPs (especially pronouns) as DO, as againt longer NPs, which are preferable in later constituents (shorter before longer): Blame it on Canada works better than Blame Canada for it, for instance.

All of these effects, and others I haven't mentioned here, are familiar from the literature on diathesis alternations involving DOs and other non-subject arguments of a V.  Put them together with the existence of blame X for Y, and you have a formula for the innovation of blame Y P X, for some preposition P.  It would be a way of promoting the resultant/caused situation NP to the position of prominence/givenness immediately after the V, and postponing the source/cause NP (usually, but not necessarily, referring to a human being) until later in the VP:

the verb blame (source/cause x, resultant/caused situation y):
  I blame Kim for the problem.  [DO x, PO y, P = for]
  I blame the problem P Kim.  [DO y, PO x, P = ?]
The only question is what P to choose.  Here, Faulty Diction alludes to a possible source for the P:

... [either] "I do not blame the President for the defeat," or "I do not lay the blame . . . upon," etc.  Here [in "I do not blame the defeat on the President"] two points of view essentially different are confused.

That is, the idiom family

lay/put/place/fix (the) blame (up)on ...

supplies a model with a candidate P, namely on (earlier upon).

And so it was.  The innovative construction blame Y on X supplies a result/caused situation DO to go along with a source/cause PO, and another construction with similar meaning provides the P on.  It's hard to believe that speakers at the time batted an eyelash over the innovation -- it would have been easily comprehensible, and in fact most people probably wouldn't have recognized it as something they hadn't heard before -- which means that it was able to spread quickly, simply because it was a Good Thing, because it does useful work communicatively. 

(Why it didn't arise earlier is a knotty question, the answer to which is likely to be: sunspots.  Lots of changes are likely to happen, for language-internal reasons, and they do happen, with modest frequency here and there, but only a few catch on -- because the planets are properly aligned, or whatever.)

More generally: most innovations have a communicative rationale -- not necessarily the same one as in the blame example -- and are not just pointless elaborations. 

Yes, I know, some lexical innovations have primarily a social rationale, as when new items mark off the social groups that use them or establish personas for particular speakers in context.  And I know that the spread of variants runs significantly along social lines (I am, after all, some kind of sociolinguist), but I'd also like to point out that a great many variants are out there just because they're good to go.

3. A few brief remarks, on things I'm flagging for possible posting on a later occasion.

3.1. Innovations as threat.  Although neither Ayres nor Faulty Diction makes this explicit for blame on, innovations are routinely seen as threatening to the older variants ((6) above).  If you see alternants as free variants ((2) above), then they're in competition, and the newcomer is a threat to the oldtimer.  (Indeed, there are a fair number of well-known cases where innovative variants have in fact supplanted older ones.)

But when the alternants are differentiated in use, stable variation can result.  This seems to be the situation for blame on, which has co-existed with blame for for at least 125 years.

3.2. The class issue.  Ayres on blame on: "a gross vulgarism, which we sometimes hear from persons of considerable culture".  Faulty Diction: "indefensible slang".  There are several things to unpack here, starting with the assumption that blame on originated "from below" ((7) above), in the vulgar mob, the users of slang, and therefore is to be rejected on that basis alone ((3) above).  No source I've seen provides evidence for this claim -- I assume that it's the result not of observation, but of reasoning from first principles, in particular from the assumption that innovations GENERALLY originate from below -- and I suspect that it might be false.  Innovations that provide communicatively useful variants, as blame on does, could in principle originate with speakers of any sort.  The fact that Ayres in 1881 was already hearing the innovative variant in "persons of considerable culture" suggests that it might have originated among such speakers (and, quite likely, independently among speakers of other sorts as well).

3.3. The education/culture issue.  Why should usage advisers assume that innovations generally originate from below?  Probably because they take innovation to be the result of ignorance ((8) above); the innovators don't know better (they should look it up!).  Educated and cultured people, on the other hand, are assumed to be able to differentiate novelties from established forms, and to know the norms that are recommended in the advice literature.  That is, the assumption is that education and cultural refinement provide a kind of synoptic knowledge of linguistic variants and their social values.  This is, on the face of it, a preposterous idea -- not even scholars of variation have such comprehensive knowledge -- but there it is.

3.4. Fashions in peeves.  The peeves that make their way into the advice literature are not an even sampling of (what are judged to be) "lower" variants (non-standard, informal, primarily spoken, geographically or socially restricted) -- I am constantly coming across such variants that get no, or very little, press from usage advisers -- nor are the variants that get the most press and excite the most passionate criticism necessarily of much, or even any, significance on rational grounds.  There are fashions in peeves, some of which persist more than a century.  (Some are extraordinary.  I have a posting in preparation about one of them, which I'm nominating for a Peevy Award for Lifetime Achievement.  But blame on is certainly a contender for the award.)  As Mark noted in his posting,

Ayres' perspective in this case is an inspired one, precisely by virtue of lacking any pretense of logical or empirical support. There are no grounds for refutation -- grammatical logic is irrelevant, and if many excellent writers often use the construction, that simply shows that their culture, although perhaps "considerable", is in fact inadequate.

and concludes with a wonderfully pointed critique, worth repeating here:

... is there any constellation of facts that could prevent Ayres' little off-hand expression of prejudice from echoing down the centuries in the unconsidered repetitions of his cultural copyists?

From a functional perspective, I suppose that this is an ideal sort of prescriptive norm. Since it's an entirely artificial policy, with no basis in the past century of speaking and writing, there's no way to learn it simply from attending to even the "best" speakers and writers. The only possible source is works on usage. This adds essential value to such works, by giving those who read them carefully and credulously a reason to feel superior to every one else.

Posted by Arnold Zwicky at December 30, 2007 02:16 PM