September 13, 2006

The tyranny of the majority, and other reasons for choosing a variant

Suppose you've come into a substantial amount of money and want to spend it founding an institution of higher learning in Lake Wobegon, Minnesota.  What will you call it?

Well, you start by making a list of ingredients for the name.  The two essentials are a head noun X denoting an institution type (University, College, etc.) and a modifier, which is most often a proper name N, either a place name (Lake Wobegon) or a personal name (Garrison Keillor), though there are other possibilities, and several more ingredients are possible.  Let's keep it simple: just these two.

Then you have to package them into a name, via premodification (N X: Lake Wobegon University) or prepositional postmodification (X of N: University of Lake Wobegon).  How to choose?

One way is to look at the way other people have made their choices, and let their decisions guide you.  You might look at the numbers, which say that:

(1) If X is anything other than University, N X is strongly preferred: Lake Wobegon College, Garrison Keillor Institute.

(2) If X is University, then:

    (2.1) If N is a place name, X of N is strongly preferred: University of Lake Wobegon.

    (2.2) If N is a personal name, N X is very strongly preferred: Garrison Keillor University.

Now these are statistical generalizations, and there are plenty of counterexamples that no one views as ill-formed: College of Wooster, Boston University, University of René Descartes (ok, the last one is a bit dodgy, maybe too Frenchy, but the other two are not an issue).  In the case of U.K. usage when place names are involved, both forms are usually acceptable, but official naming practices favor University of N; in the U.S., there is virtually no choice.  Still, you could say: let's go with it, choices as above.

When we came into this discussion, some postings ago, Shen Hong was quoted as advocating that the choice in (2.1) was a RULE: PlaceName University is just wrong (at least in formal contexts), according to him.  What's going on here is that a statistical preference is being bumped up to, elevated to, a rule of grammar, by a kind of majority rule: instead of seeing two constructions in competition, with one much more frequent than the other, the facts are conceptualized as a general rule plus a scattering of individual exceptions, each of which is a kind of idiom (syntactically ill-formed, but nevertheless occurring, so having to be memorized one by one). 

You can see this sort of majority rule reasoning elsewhere -- in discussions of restrictive relative which, for example, where it's sometimes pointed out that that is now the preferred variant, so why not go all the way and use that all the time?  Recently, Stanford student Doug Kenter and I have been looking at (and talking in public about) sentence-initial linking however vs. but --

I expected a multitude.  However, the audience was minute.
I expected a multitude.  But the audience was minute.

and examining Bryan Garner's advice (in many places) that however should not be used here, but should be replaced by an alternative construction, preferably sentence-initial but.  Now, we found a strong preference for but over however in writing (extraordinarily strong for some writers: in my Language Log postings through July of this year, I used but 72 times and however not a single time -- I wasn't counting as I went along; Doug extracted the counts after the fact), and one response we've gotten to these findings is that writers should take Garner's advice.  That way, their writing is bound to be ok.

This reasoning assumes, first of all, that the variants really do not differ in meaning or discourse function (in many cases, this assumption is dubious at best, but in the university naming world, at least, it seems to be correct); and then treats free variation as something in need of a fix, with one variant either confined to informal style (as has sometimes suggested for PlaceName University) or barred entirely.  There must be One Right Way.  Garner himself picks the but variant on grounds of taste -- more on this in a later posting -- rather than on the numbers, but he still thinks a choice needs to be made.

Well, I don't think variation is in need of a fix when we're talking about choices within the standard language.  There's nothing wrong with minority variants.

(Some of you may be thinking that I'm being inconsistent, by rejecting the tyranny of the majority in this case, but accepting a kind of majority rule with respect to the spread of variants into general use, in particular general use in formal contexts -- what some people think of as the laxness of descriptivists.  I don't think the two situations are at all comparable; in the interests of saving space, though, I'll postpone that discussion to another posting.)

And in fact, when you look at particular contexts of language and groups of people, you see very significant differences in usage among practiced writers.  On Language Log, for instance, the but/however ratio for sentence-initial linkers ranges from Mark Liberman at 3.28 to Ben Zimmer at 15.00 to Geoff Pullum at 24.25 to me, with no however at all.  (To put this in a larger context, the ratios for ALL uses of but to ALL uses of however in the British National Corpus is 7.95, in the Brown corpus 7.94 -- that is to say, roughly 8.)  Mark turns out to be exceptionally fond of sentence-initial linking however, Ben to be somewhat averse to it, and Geoff and I to be strongly averse to it.  However (there! I've broken the run), all of us defend this use of however, as something available to anyone, in formal writing and elsewhere.

Back to naming educational institutions.  I reject the tyranny of the majority, and move on to more interesting and subtle things.  Even when variants do not differ in meaning or discourse function, they may be perceived to differ in other ways.  Look at Name University vs. University of Name.  Some differences that might be relevant to a choice between them:

(3) Name University has the advantage of brevity (though only by one or two words, depending on whether you count the the in the University of Name).

(4) But Name University can be uncomfortably left-heavy, which would favor the prepositional form.  Still, that hasn't held back Northern Arizona University, Southern Illinois University, Western Reserve University (one of the ancestors of Case Western Reserve), or even Northeastern Illinois University.  (Or, for personal names, George Mason University, George Washington University, and many others.)

(5) The two forms differ in where they assign prominence: in the premodifying form, the primary accent is on University, while in the prepositional form, the primary accent is on N, suggesting that  the denotation of N is particularly significant.  As a result, University of Kutztown would sound rather silly, since Kutztown is a really small town, and University of Garrison Keillor might suggest that the university is all about him.

(6) For more complex names, in particular those incorporating subject-matter in the name, the prepositional names might introduce a potential confusion with the names of other institutions.  So, Beijing University of Language and Culture and Language and Culture University of Beijing might suggest (incorrectly) a connection to Beijing University/University of Beijing.

Similarly, in the U.K. some of the newer universities have to be distinguished from older universities in the same places.  One strategy is to add a personal name to the place name.  But PersonalName PlaceName University (Ruskin Anglia University) or PersonalName University of PlaceName (Ruskin University of Anglia) would invite confusion with PlaceName University (Anglia University) or University of PlaceName (University of Anglia), respectively, so there are now names of the (to me) odd form PlaceName PersonalName University: Anglia Ruskin University, Liverpool John Moores University, Oxford Brookes University.  (Thanks to Andrew Gray.)

In addition to these formal considerations, it's also possible for the choice between variants (here as elsewhere) to pick up, construct, or convey social meanings.  Different people can even have divergent connotations for the same variants.  So it is with university names.

(7) Given the traditional official practice of U.K. universities, in favor of University of PlaceName, this variant is now seen by many (on both sides of the Atlantic) as formal, with PlaceName University viewed as an informal abbreviation.

(8) Again given the official practice of most U.K. universities, the University of PlaceName version is seen by some as traditional, as against the modern PlaceName University.  What values you then attach to the different forms depends on your attitudes towards tradition and modernity.

(9) Some in the U.S. associate premodifying PlaceName University with private institutions, prepositional University of PlaceName with public institutions.  (These connotations will then frequently contradict the ones in (7) and (8).)  This is an especially interesting case, illustrating the very frequent phenomenon of conflict between fact and cultural construction, with cultural construction aligned with social meanings.

The facts are that PlaceName University is not mostly private, though University of PlaceName is mostly public -- but this latter fact is a consequence of other things, not a direct association.

Background observations:

(10) Combinations of University with a personal name, IN EITHER VERSION, tend to name private institutions.  Private institutions are very frequently named to honor people.

(11) Combinations of University with a place name, IN EITHER VERSION, tend to name public institutions.

From observation (10) and observation (2.2) above, favoring premodification for universities with personal names in them, it follows that PersonalName University will be mostly private: Harvard, Yale, Brown, Stanford.  This is, strictly speaking, irrelevant to the nature of PlaceName University institutions, but when you think of Name University examples, you're going to pull up personal name examples, and the premodifying form might then become associated in your mind with private status (and prestige).  It's also true that there are a few highly salient (famous, etc.) PlaceName University cases like BU and NYU -- you're not likely to think of Kutztown U or NAU or NEIU when you're dredging for examples -- so your attention will be drawn to private PlaceName University, reinforcing the effect from the personal name cases.

From observation (11) and observation (2.1) above, favoring prepositional versions for universities with place names in them, it follows that University of PlaceName will in fact be mostly public.

This brings us to the University of Rochester, a private institution with a prepositional name.  Frank Townsend (a U of R alumnus) reports a rumor that an outgoing president a while back wanted to change the name to Rochester University, which to him sounded private, like Boston University and New York University and unlike University of Buffalo (oh dear, I see that SUNY Buffalo now bills itself as "University at Buffalo", with "at") or University of Massachusetts.  Meanwhile, on ADS-L Larry Horn has described the situation at an earlier time:

My undergraduate alma mater, the University of Rochester, first spent decades zealously correcting anyone who dared misrepresent its brand as Rochester University (on the assumption that the correct version put it in the category of the University of Chicago, while the latter might mislead prospective applicants and their parents into inferring that it was just another big city school like (perish the thought) Syracuse University (and would then cause them to wonder why our tuition was so much higher).  Then, having decided that the difference between "the U of R" and "R U" wasn't sufficiently robust, the powers that be began contemplating changing the name completely, but presumably no sufficiently classy alternative was found, since it still seems to be the University of Rochester.

To wrap all this up, Bob O'Hara vaguely remembers "an edu-linguistic meme that used to circulate on the net years ago", a list of "rules", each of them illustrated by a counterexample: "Schools called 'University of Placename' are always public (e.g. University of Southern California)"; "Schools called 'X College' have no graduate programs (e.g. Dartmouth College)".  The point was, apparently, that there wasn't much pattern to this stuff.  (Anybody have a line on this list?)  But in fact there are patterns here, generalizations like (1), (2), (10), and (11), but generalizations with a significant number of exceptions.  No inviolable rules, true, and plenty of complexity, but it's not just chaos.

zwicky at-sign csli period stanford period edu

Posted by Arnold Zwicky at September 13, 2006 07:55 PM