Arnold Zwicky's Blog

A while ago (on 5 May), I posted to ADS-L (somewhat revised here):

NSAID is an abbreviation for “non-steroidal anti-inflammatory drug”.

How is it pronounced?  Apparently, as /ɛn sɛd/ (with compound stress, greater stress on the first element than on the second).  At least that’s what I hear in television ads. That is, the N is pronounced as a letter name (as in initialisms) and SAID is pronounced the same way as the verb form said (as in acronyms).  So this is one of those mixed abbreviations that we don’t really have a name for.

[There are all sorts of abbreviations. My interest here is in those composed of initial letters of words, of which there are two main types: initialisms, which are pronounced as a sequence of letter names (MIT or M.I.T., standing for Massachusetts Institute of Technology); and acronyms, which are pronounced as if they are ordinary words (PET, standing for positon emission tomography and pronounced just like pet). (The terminology here is mine, and the one followed in my Language Log postings, but other writers have used different labels.) NSAID is an instance of yet another type of initial-letter abbreviations.]

I’ll now go to to comment a bit more on NSAID, to suggest a name, and then to make some further observations about these hybrid abbreviations.

First, the name of the class of drugs is sometimes spelled Nsaid (by people who don’t like long chains of capital letters, I suppose). This is in line with a trend towards spelling acronyms in lower case with an initial capital letter, rather than in all-caps — a trend Geoff Pullum remarked on in British newspapers here.

An interesting feature of NSAID is /sɛd/ for SAID.  The /ɛ/ here is exceptional; we’d normally expect /e/, as in paid, raid, laid, braid, maid, afraid. But the spelling SAID represents an existing English word, so its pronunciation takes precedence over the general letter-sound correspondences.

On to a name for the hybrid initial-letter abbreviations: I suggest initacro, pronounced trochaically (with stress on the first and third syllables, higher stress on the third). This name reflects the fact that these abbreviations (well, the examples I’ve collected) consist of an initialism (sometimes just one letter, as in NSAID; sometimes two, as in SFMOMA, the San Francisco Museum of Modern Art — SFMoMA, SF MOMA, SF MoMA, and S.F. MoMA are other variants) followed by an acronym; that is, they have an init portion and then an acro portion.

Now a few remarks on stress in initacros. As far as I can tell, this is quite regular:

(a) an initacro with a one-letter init portion has compound stress (“forestress”, with greater stress on the first element);

(b) an initacro with a two-letter init portion has compound stress within init portion, but has phrasal stress (“afterstress”, with greater stress on the second element) overall.

So, for (a), we get a stress pattern like that in dog house, White House, etc.: NSAID, NWAV(E) (New Ways of Analyzing Variation (in English), a linguistics conference), MCAT (Medical College Admission Test), JPEG (Joint Photographic Experts Group, the name of the committee that created the standard), SCHIP (State Children’s Health Insurance Program).

For (b), the init portion is forestressed; SF in SFMOMA gets greater stress on the S, this despite the fact that free-standing two-letter initialisms (like L.A./LA, B.A./BA, and O.U./OU, referring to Ohio University) are afterstressed. I’ll offer an explanation for this difference in a moment. (On periods in initialisms — there’s a lot of variation out there — see my discussion here.) And then, the whole abbreviation is afterstressed, with greater stress on the acro portion than on the init portion, as in phrases like a white house.

So SFMOMA ends up with a stress pattern in which the first three syllables all have some stress, but the third syllable has the highest level, the first syllable the next highest, and the second syllable even less. (I’ve resisted using numbers for stress levels, because what’s important here is the relative levels of stress, not some absolute levels.) Similarly for MS-DOS/MSDOS and CD-ROM/CDROM/Cdrom. Note that SFMOMA, MS-DOS, and CD-ROM are all parseable as modifier-head expressions, so it makes sense for them to have phrasal stress.

[Digression on terminology: I’ve been using the terms compound stress and phrasal stress because these have a considerable history in the analysis of English, but they’re just labels, not definitions. In particular, things that are clearly compounds do not always have “compound stress”. For example, the common noun hummingbird lane, as in “This is our hummingbird lane”, referring, perhaps, to a lane where we often catch sight of hummingbirds, has forestress; the proper name Hummingbird Lane, however, has afterstress; and the otherwise very similar proper name Hummingbird Street has forestress. There’s a system to all of this; some discussion can be found in an old paper of mine here, along with some other systematic — and some idiosyncratic — exceptions to the generalization that compounds have “compound stress”.]

On to the question of why SF should be forestressed in SFMOMA, even though such an initialism would normally be afterstressed. This is an easy one: English has (that is to say, English speakers have) a strong drive towards “alternating stress”, an alternation between stressed and unstressed syllables. (This is scarcely a specifically English feature; it’s found many places, and there’s a big literature on it.) One manifestation of this tendency is an avoidance of “stress clash” (two heavily stressed syllances in sequence) when that’s possible, via “stress retraction”, a shift of stress on the earlier element to a still earlier syllable. So we get, in the textbook example,

Q: How many people did you see? A: ThirTEEN.

but

Q: How many people did you see? A: I saw THIRteen people.

In the second example, thirTEEN would not be wrong, but THIRteen is better, because it avoids stress clash.

The facts are complex. In some circumstances, stress retraction is possible, in others not. In still others, it seems to me to be virtually obligatory; I think that’s the case in SFMOMA and probably in other initacros with two-letter inits. 

The thing is that the acro portion of initacros seems to be always initially stressed (most of the time, it’s a monosyllable, in fact). So it’s not surprising to see stress retraction in the init portion — to the point of virtual necessity. (I find it hard to wrap my head around SFMOMA with higher stress on the F than on the S, but maybe that’s just me; someone should look at what people actually say. Note: reports about what people think they say are unreliable, even if heartfelt.)

[A final note: there are lots of other types of spoken abbreviations out there, many of them involving clipping. I’m not denying their existence, only focusing on one class of cases that can be carved out from the whole assortment.

One of my favorites: initialism + clipping, as in UConn or U Conn, for the University of Connecticut, and UWash or U Wash for the University of Washington. Even better: UDub or U Dub for the University of Washington (and possibly some other universities as well), where Dub represents the dub syllable of the letter name double-u, via the initialism UW for that university.]

This entry was posted on May 18, 2009 at 6:13 pm and is filed under Words. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.