A summary of various phenomena involving the comparative P (subordinator or preposition) than, focusing on non-standard variants.
Background. Crudely, there are three sorts of things we’re dealing with when we talk about comparatives:
SEM: the semantics of comparison, involving some property with a scale and rankings along that scale; two things are compared along that scale — the focus (Foc) and the basis (Bas).
equative semantics (eq): Foc and Bas are ranked at (about) the same point on the scale;
differential semantics (diff): Foc and Bas are (significantly) different on the scale, with Foc higher or lower than Bas;
polar semantics (pol): Foc is at an extreme of the scale, either highest or lowest.
MORPH: the morphosyntactic devices available for use in syntactic constructions, including both the inflectional marks — base grade (BSE big, talkative) for eq constructions, comparative grade (CMP bigger) for diff constructions, superlative (SUP biggest) for pol constructions — and periphrastic expressions of the comparative (CMP more/less talkative) and superlative (SUP most/least talkative) grades.
SYN: the syntactic constructions devoted to expressing the semantics of comparison; these include both the central constructions (all with two parts, focus + basis) —
eq marked by as BSE … as …
diff marked by CMP … than …,
pol marked by (the) SUP, with various expressions of the basis (via of +NP, for example)
— and also an assortment of other constructions, for example the same … as …, (would) rather … than …, and prefer … to …
[Comparison is related in complex ways to other rankings on a scale (in particular, temporal or spatial precedence), to expressions of similarity and dissimilarity, and to expressions of exclusion (except, but + NP, etc., including other than).]
As I noted above, the central syntactic constructions come in two parts (one part for Foc, one for Bas), with separate marks for each part: a primary mark on Foc (FocM) — as, CMP, SUP — and a secondary mark on Bas (BasM) — as, than, of.
[From the point of view of logic, only FocM, conveying eq, diff, or pol semantics, is essential; BasM is in fact redundant, though of course communicatively useful (as redundant marks are in general). That is, English could have developed to treat FocM + Foc as a transitive verb, with Bas serving as its direct object, and no BasM at all. In this alternative English, we’d have:
Kim is as big Sandy / as talkative Sandy.
Kim is bigger Sandy / more talkative Sandy.
Kim is (the) biggest the class / most talkative the class.
This is not, of course, the way real English works, but my point is that BasM is in principle dispensable, since it’s determined by — selected by, governed by, licensed by — FocM.]
The main event. At this point, I’ll narrow my focus to constructions with BasM than. There are four sets of phenomena I’ve seen discussed (so far):
(1) government by the nearest;
(2) temporal sequencing with implicitly negative adverbs (scarcely, hardly, barely);
(3) implicit comparison;
(4) truncated comparison.
These waters are muddied by the tendency of commentators (including, on occasion, MWDEU) to appeal to blending (sometimes under the label “confusion”) or to the language of non-native speakers (either through transfer from their native languages or through imperfect learning of standard English, with extensions of the English syntax they know in non-standard ways). While both things can happen with than, when you exclude non-native speakers and occurrences that are likely to be inadvertent errors that would be disavowed by those who produced them, there are still plenty of types of phenomena left.
(Syntactic blends do occur as inadvertent errors in language production, but most apparent combinations of syntactic constructions are not production errors; instead, they are intended as produced — whatever the chain of historical events that might have given rise to these intentions.)
Now, the phenomena. Each of these deserves a posting or more on its own; here I’m just flagging the topics.
Set 1: Government by the nearest. The prototype here is
as big or bigger than
with than as BasM. MWDEU treats this as a syntactic blend of as big as … and bigger than …, yielding a failure of syntactic parallelism (the “correct” version is as big as or bigger than), but examples like this occur in such massive numbers that an inadvertent blend analysis for most of them seems preposterous. And in fact, as BSE or CMP than falls in with a big collection of other coordinations of the form
FocM1 + Foc Coord FocM2 + Foc BasM2 + Bas
where in non-coordinate examples FocM1 selects BasM1 and FocM2 selects BasM2, so that there’s a conflict between BasM1 and BasM2 for the last element in the coordination. If it’s not avoided by rewording, this conflict is uniformly resolved in favor of BasM2, which is the nearest element to Bas.
This is government by the nearest, examined in some detail here (with some earlier discussion here).
Set 2: Sequence marking with temporal negatives. Here the models (indisputably standard) have the explicitly negative no sooner:
(SVP) I had no sooner entered the room than the dog lunged at me.
(SAI) No sooner had I entered the room than the dog lunged at me.
(On the choice between the clause constructions SVP and SVI, see this posting.)
Than is licensed here by the CMP sooner, though many speakers allow when instead of than; there’s a competition between the explicitly temporal when and the explicitly comparative than, which could be resolved either way.
This competition persists in temporal sequencing marked with the implicit negatives hardly, barely, and scarcely:
(SVP) I had barely entered the room when/than the dog lunged at me.
(SAI) Barely had I entered the room when/than the dog lunged at me.
These implicitly negative temporal adverbials also convey a ranking (of events in time) and so are semantically similar to comparatives, which convey a ranking on a property scale.
The facts are (1) that there’s a lot of variation with respect to when vs. than and SVP vs. SAI; and (2) that a fair number of handbooks view than in these constructions as a vulgar error (possibly arising from “blending” with no sooner than examples). A topic for another posting.
Set 3: Implicit contrast. Here the classic is different than (vs. different from or different to), which deserves a lengthy posting of its own. The prepositions from and to have clear semantic motivation (exclusion for from, (dis)similarity for to); Jespersen suggested, long ago, that what motivates than is implicit contrast (or separation), as with other than.
The sociolinguistics of variation is again complicated, but in this case different than is now clearly a standard alternative to different from/to (no matter how much vitriol peevers pour on it).
In the same family are differ than, pretty obviously back-formed from different than; twice [etc.] as likely … than (if something is twice as likely as another, then it’s more likely than it); inferior/superior than (like different); and the most … than. Each deserves a separate posting.
Set 4: Truncations. Here we have a set of examples with (apparently) truncated licensors of than, with various motives for the truncation — starting with sooner than later, with truncated rather (which might also be seen as rather absorbed by the preceding CMP, or simply as CMP licensing than on its own). The historical origin here is pretty clearly the truncation of idiomatic (and frequent) sooner rather than later, leaving out the element that’s most dispensable in discourse (truncations of idioms are very frequent).
Then there’s safe than sorry (< better safe than sorry), considered by Neal Whitman (here) along with sooner than later; again, the historical licensor is omitted, the marker than remaining to carry the construction.
Then no N … than (< no other N … than), again with the licensor omitted. And, similarly, the rather of preference omitted in an assortment of contexts (discussion here), most especially would rather … than.
All these types — there are probably more — show a historical trend towards omitting the primary mark of a construction while maintaining a secondary mark, which then serves as the new primary mark. This is a familiar development, most famously tracked (in many languages) in the way in which original primary marks of negation are supplanted by secondary (originally reinforcing) marks (ne … pas > pas in French, for example), which then become the new primary marks.
Leave a Reply