September 18, 2006

How safer is America today?

Yes, that's the headline on a UPI story (of 8/31/06) by Claude Salhani, UPI's International Editor, on the current terrorist threat.  I think that question deserves an asterisk of ungrammaticality: "How safe is America today?", ok; "How much safer is America today?", ok; "How safer is America today?", no.  Comparative adjectives like "safer" just don't take the same degree modifiers as plain adjectives like "safe" -- in standard English.  But it turns out that there's a fairly widespread non-standard English out there in which comparatives can work just like plain adjectives.

Cornelius Puschmann, who alerted me to the headline (after checking with a native speaker of English and eliciting an "ick" response much like mine), speculated that "how safer" was either a typo or a piece of non-native English.  Either thing might be possible in this particular case -- though we probably can't blame Salhani, because he almost surely didn't write the headline, and the odd comparative doesn't occur in the article -- but you can find plenty of examples of modified comparatives like "how safer", apparently from native speakers, on the web.  I don't know how long this has been going on, or what the social distribution of the non-standard comparatives is, or even whether the non-standard comparatives coexist with standard ones or (more likely) supplant them.  But they're out there.

Now some background about the system of standard English...

1.  English has two schemes for comparing adjectives (and adverbs, though I'm going to focus on adjectives here): a morphological scheme, for "inflected comparatives" in -er, like safer and handsomer; and a syntactic scheme, for "periphrastic comparatives" with more (or less), like more comfortable and more handsome.  There's a lot to be said about which adjectives use which scheme, but that's (fortunately) beside the point here.  The first big generalization is

Same Syntax: Inflected comparatives and periphrastic comparatives have almost entirely the same syntax.

This is no great surprise, since comparison by inflection and comparison by periphrasis have the same semantics.  But notice the hedge "almost entirely"; alternative expressions virtually never have EXACTLY the same syntax.  The differences between the distribution of inflected and periphrastic comparatives are (again, fortunately) beside the point here; it's a similarity we're interested in, a similarity in the kinds of modifiers they can take.

2.  To get at this, I'm going to introduce a piece of terminology: the MODIFIER SET for a word is the set of all expressions that can combine as an adjunct with this word as head; and the modifier set for a word class is the union of the modifier sets for all the words of that class.  To illustrate: the modifier set for the class of (plain) adjectives is the set of degree expressions, among them very, pretty, how, so, and more.  And the modifier set for the class of nouns comprises both adjectival expressions (safe, handsome, comfortable) and determiner expressions, both determinatives (words especially devoted to serving as determiners, like the, a, much, many, more, and this) and expressions of other types, like the noun lot of a lot of and possessive NPs like the doctor's

An important complexity: the determiners in the modifier set for nouns are different for different types of nouns.  In particular, (singular) mass nouns, singular count nouns, and plural count nouns take somewhat different determiners; much goes only with mass nouns (much shrubbery), a only with singular count nouns (a bush), many only with plural count nouns (many bushes), while more and lot go with "extended" nouns (mass or plural count: more shrubbery, a lot of shrubbery; more bushes, a lot of bushes).  Hang on; we're going to need these facts in a little while.

In any case, we can now state the relevant special case of Same Syntax:

Inflected comparatives and periphrastic comparatives have the same modifier sets.

Examples (from standard English): much/a lot handsomer, much/a lot more handsome, *very/so handsomer, *very/so more handsome.

Put another way:

Same Modifier Set: The modifier set for inflected comparatives is the same as the modifier set for the degree word more.

3.  And the modifier set for the degree word more (a quantity modifier) is something of a surprise.  To start with, it's nothing like the modifier sets for other degree words.  Many degree words don't allow any modifiers, but some do, and these are themselves familiar degree words: so very big, with so modifying very; very surprisingly big, with very modifying surprisingly.  But, as we've just seen, very and so can't modify degree more.

In case you were trying to think of the degree word more as just an adjective, you'll now see that that can't be right, since the modifier set for adjectives is just the familiar degree words, and none of these can modify degree more.

Instead, the modifiers of degree more (and less) are quantity modifiers: much, no, any, a lot, lots, a bit, a great/good deal.  Where else do we see these?  Astonishingly, they also function as DETERMINERS, in the modifier set for nouns -- specifically, for mass nouns: much shrubbery, any confusion, no information, a lot of trouble, lots of rice, a bit of grass, a great/good deal of wine, parallel to much more handsome, no more satisfactory, any more intelligent, a lot more complicated, lots more ridiculous, a bit less convincing, a great/good deal less tasty.  Not all mass determiners of quantity can also modify degree more -- some and all can't -- but the relationship is very close.

At this point, it looks like degree more is acting like a noun.  But that can't be right, either, because the nominal determiners lot, lots, bit, deal don't get the of that's required when they combine with a noun.

4.  However, there's one more place where we see essentially the modifier set much, a lot, etc.: as the modifier set for DETERMINER more with mass nouns: much more shrubbery, no more information, any more confusion, a lot more trouble, lots more rice, a bit more grass, a great/good deal more wine.  We even pick up at least one more modifier that degree more shares with determiner more, namely somewhat (somewhat more handsome, somewhat more lettuce, but *somewhat lettuce).  In other words:

DetDeg: The modifier set for degree more is almost identical to the modifier set for mass determiner more.

(We have to specify the MASS determiner more here, because determiner more can also be used with plural count nouns, in which case its modifier set reflects the properties of the plural count noun: many more bushes vs. much more shrubbery.)

Another way of thinking about these facts is:

One More: There is only one lexical item more here, but it gets used in in two different constructions, in two different syntactic functions, degree modifier and determiner; the facts about the modifier sets are (virtually) identical because they are facts about a single lexical item.

(There's now a question about what the CATEGORY of this lexical item is -- a fascinating question, but one that would take us even further afield.)

5.  These connections between different constructions have long been known -- the early monument of the literature on English comparatives is Joan Bresnan's 1973 monograph "Syntax of the comparative clause construction in English", published as a 70-page article in Linguistic Inquiry -- but I've always thought of them as one of the coolest facts about English, not at all something that you'd be likely to predict from first principles.

In any case, the standard English system hangs together in a remarkable way: Same Modifier Set and One More, together with a stipulation of the modifier set for this item more, do the job.

6.  So what's happened with the non-standard degree modification, as in "How safer is America?" and other examples like those below (involving both inflectional and periphrastic comparatives)?

Computer Newbies- it should show them why to switch and how safer and easier it is than other browsers. (link) just how healthier are we going to be when volatile climates have wiped out half the world's wheat crop one of these years? (link)

... and above all this the products are very, very safer for both myself and family and the environment. (link)

And nobody knew what will happen if we reach communism so it was pretty safer for the party to say we're just on the way. (link)

But how more efficient is this system when compared to current ones? (link)

And, if you're embarrassed about it, think of how more comfortable sucking your thumb is when you have one. (link)

I would feel very more comfortable if there was a Carepaq style support mechanism. (link)

I can tell a story that would get you to a place like this.  You extend the modifier sets for plain adjectives to comparative adjectives, thus eliminating the oddity of modification for comparatives.  And somewhere along the way, you stop treating the periphrastic comparatives as involving degree modification in the syntax; instead, you treat more plus adjective as a kind of compound adjective, the whole thing having the comparative property, just like an inflected comparative word.

By the way, the result is briefer comparatives; you don't need that much any more.

This development would make One More dispensable, though you'd still have stipulated oddities for determiner more.

On the other hand, you could stick to One More, in which case determiner more would have its syntax altered as well.  And there are a few hits suggesting that this has happened for some people:

General good security practices will suffice for web based access since if done right, there will be very more information available than is available on ... (link)

Here I break out in asterisks.

[Addendum: I left out the Shakespeare -- "How sharper than a serpent's tooth it is / To have a thankless child!" (King Lear), one of my favorite quotations -- but now Russell Borogove (and the Mome Raths?) has written to remind me.  The quotation suggests the possibility that the current modified comparatives, or at least some of them, are survivals rather than (re)inventions.  I don't know a damn thing about the history; as I keep telling people, I don't do REAL historical linguistics (or REAL phonetics or REAL formal semantics), but just appeal, whimpering, to the specialists.  In any case, I hope to post soon about yet another case where survival and (re)invention are both live possibilities.]

zwicky at-sign csli period stanford period edu

Posted by Arnold Zwicky at September 18, 2006 02:31 PM