July 15, 2006

I was (probably) wrong

I may very well have been wrong when I suggested that "drop ceiling" is derived from "dropped ceiling", in the way that "ice cream" derived from "iced cream". Michael Robinson wrote:

I never thought that "drop ceiling" had any relationship at all to "dropped ceiling". I've always assumed that "drop" was used in the same sense as it is used in theatrical scenery, where a "drop" is something that is lowered down from the space above the stage. You talk about a "drop curtain", never a "dropped curtain". A "drop curtain" is a curtain that's a drop.

"Drop ceilings" are not very common on stage, but not unknown, as when representing a very confined space such as a ship's cabin.

At any rate, it seems to me that the usage of "drop ceiling" is influenced by "drop curtain".

In contrast, I've always assumed -- without thinking about it -- that a "dropped ceiling" is a false ceiling that has been "dropped" or "lowered" relative to the true ceiling, and a possible connection to theatrical drops never occurred to me. Whatever the derivation, it's certainly the recency illusion that made me think that "drop ceiling" was an innovation. People have been using both forms for at least a century, and the earliest citation I could find is to "drop", not "dropped".

In the December 1901 issue of Harper's Bazaar, the author of the "Household Decoration" column advises "An Admirer":

Use a pale but decided cream shade in window draperies, with white shades between the sash and long curtains. Pure white would be harsh with the mellow tones already planned. Arrange the writing-table between the bay-window and the bookshelves: the big Davenport in the northeast corner, the mahogany sofa in the corner southeast, and that superb table in the centre of the room. Choose cream Brussels net curtains for the reception-room, edged either with pointe Arabe or Renaissance lace. Why not vary the walls here by using an Empire paper in green tones with rococo or pure Empire festoons, in which the least bit of rose color occurs? Have a drop ceiling of deep cream, divided from the paper itself by a narrow gilt moulding. Your Empire furniture would show to great advantage in such a setting, and the half-defined rose shade suggested would answer some tone in the mahogany as well as in the damask covering, which is usually an accompaniment of Empire furniture.

On the other hand, the same column in the September 1909 issue tells "M.T.":

It is very gratifying to hear that the suggestions I gave you last year have been carried out and have pleased you so much. I am sending a sample of tan paper for your living-room, and with it would use dull bronze green and brown furnishings. Cover your couch with either a bronze green or brown -- not with bottle green. It will be necessary to be very careful in your selection of just the right tone of green. [...]

It would be very attractive to use the same paper in both of your rooms, with bronze green for a contrasting color in one room, and a greenish brown for a contrasting color in the other room, but if you wish to have them different, the green paper of which I sent you a sample before will be beautiful in the sitting-room. Here, too, I would confine the colors to brown and green so that the two rooms will combine harmoniously. Inner curtains of bronze-green silk would be lovely in the living-room, and of light brown in the sitting-room. I would use dropped ceilings again if you think they make the rooms look in better proportion.

The earliest use I could find in the NYT (March 2, 1913, "Spring Styles in Wall Papers") has "drop":

In some rooms drop ceilings are still used, for it gives the sense of largeness. As the decorators say, "It spreads the room."

But Diana Rice, 10/16/1921 (p. 83) "Here is Finest Rural School", has "dropped":

The modified Spanish style of architecture, the many windows, the broad, low entrance steps and wide doors present a pleasant and encouraging friendliness to the chance visitor. Pushing into the rough-plastered vestibule and thence into the roomy halls, with their deep cream dropped ceilings and soft green walls, inviting vistas through low, graceful Moorish arches of quiet classrooms, study halls and library, lure the visitor on.

By the 1930s, real-estate ads are using "dropped ceiling" and "drop ceiling" promiscuously. The OED, alas, offers no information.

Of course, the pronunciation of -ed in this context is just [t], and the phonetic difference between [pts] and [ps] is not a very salient one, especially in fast speech. Partly for this reason, syllable-final /t/ and /d/ are often deleted in English. Much of the long history of research on this topic is summarized in Chapter 5 of Andries Coetzee's 2004 dissertation:

In English, a coronal stop that appears as last member of a word-final consonant cluster is subject to variable deletion – i.e. a word such as west can be pronounced as either [wɛst] or [wɛs]. Over the past thirty five years, this phenomenon has been studied in more detail than probably any other variable phonological phenomenon. [...]

The factors that influence the likelihood of application of [t, d]-deletion can be classified into three broad categories: the following context (is the [t, d] followed by a consonant, vowel or pause), the preceding context (the phonological features of the consonant preceding the [t, d]), the grammatical status of the [t, d] (is it part of the root or is it a suffix). The contribution of each of these three factors can be summarized as follows: (i) The following context. [t, d] that is followed by a consonant is more likely to delete than [t, d] that is followed by either a vowel or a pause. Dialects differ from each other with regard to the influence of following vowels and pauses. In some dialects, a following vowel is associated with higher deletion rates than a following pause. In other dialects this situation is reversed – i.e. more deletion before a pause than a vowel. (ii) Preceding context. In general, the more similar the preceding segment is to [t, d], the more likely [t, d] is to delete. Similarity has been measured in terms of sonority (higher deletion rates after obstruents than sonorants), but also in terms of counting the number of features shared between [t, d] and the preceding consonant. (iii) Grammatical category. Generally speaking, [t, d] that is part of the root (in a monomorpheme like west) is subject to higher deletion rates than [t, d] that functions as a suffix (the past tense suffix in locked).

In the case of the many paired two-word phrases with and without the -ed suffix on the first element, this phonological variation intersects with morphological and semantic variation. In my earlier post, I gave the list ice(d) cream, skim(med) milk, ice(d) tea, wax(ed) paper, roast(ed) beef, shave(d) ice, cream(ed) corn, whip(ped) cream, and Barbara Zimmer wrote to suggest adding screen(ed) porch and steam(ed) crabs. Is it a porch that has been screened in -- a screened porch, like a covered wagon -- or a porch whose walls are made of screens -- a screen porch, like a stone house (or a screen door)? Either formation is consistent with the broader syntactic and semantic patterns of English, and with the forms and meanings of the particular words involved. And either phrase could easily be misperceived for the other one, given both the low phonetic salience of the pronuncation difference at best, and the general tendency for that difference to be omitted in speaking. So it would make sense to find word histories in which X Y turned into X-ed Y, as well as histories that go the other way.

[By the way, if you have access to the Proquest American Periodicals Series Online, or to a library with old issues of Harper's Bazaar, I can recommend the experience of reading a half dozen or so of the Household Decoration columns from the first decade of the 20th century. As you assimilate those idealized descriptions of upper-middle-class living spaces in the time of Theodore Roosevelt and Edward VII, I predict that you'll soon feel an intense need for free verse, cubism, and Freudian analysis, even if you're normally as immune as I am to the attractions of Modernism.]

[Update 7/16/2006: Dan Bruno points out that "old fashion(ed)" should be added to the list of common pre-nominals that alternate between V+ed and N forms, as in these examples.]

Posted by Mark Liberman at July 15, 2006 08:08 AM