Noncompositional

Understanding how a language changes over time is a crucial part (most people would agree) of developing a full understanding what language is (what are possible languages, how could a person’s mental model of a language be organized, and so forth). One part of language change is semantic change, whereby some word (or possibly construction) changes in its denotation or connotation. People are familiar with modern variations on word meanings, like colloquial uses of various evaluative adjectives, such as cool, tight (which has a range of different colloquial uses now than it did a couple generations ago), and sweet. Everyone knows about gay, of course. Less familiar are the history of of the meanings of computer and nice, though I assume that many students of linguistics will have heard about the latter in their historical linguistics class. Constructional examples might include the various uses of subject-aux inversion (Is linguistics interesting?, boy, is linguistics interesting!, physics is far more interesting than is linguistics, may your children study linguistics, and so forth), or in Japanese the use of a “hanging” conditional as a suggestion (where something that means If you run can mean Who not run? or You should run. Compare to English How about if you ran?).

Take the example of spill. A caricature of its change over time might be, “originally it meant kill, but now it means accidentally cause a liquid to come out of a container.” Of course this is clearly a simplification, since it glosses over large swaths of time when it had multiple meanings; and dictionaries are generally no help in getting a sense of how meanings evolved, since the job of a dictionary (like the OED) is to give clear examples of unambiguous uses, not the crucially ambiguous or “creative” (e.g., novel-metaphoric) uses that often give rise to semantic shift.

One sort of semantic change that is not often discussed in detail, but which was certainly recognized by historical linguists like Gustaf Stern, was where changing culture and technology changes, but the language doesn’t. He gives the example of a ship, the modern varieties of which look rather different from what might have been called a ship over a thousand years ago. Nonetheless some continuity was observed enough by language users that the word stuck. Similar, more modern cases include the telephone, which include not just cell phones but also wireless headsets and computer software that utilizes VoIP, and perhaps Japanese 土木学, which means civil engineering though character-by-character it is earth-wood-study. Another interesting one is the verb tape, which is used to mean copy audio/visual information onto some permanent medium: it can still be used even when there is no physical “tape” onto which the recording is done (just search for “tape * on my tivo”, and peruse the results).

It seems to me that such a “change,” involving some continuity of referent, can also be seen for other sorts of signs, like icons. One was recently pointed out in a BBC article on the demise of the floppy disk:

Interestingly, software giant Microsoft seems to be keeping the flame alight for the floppy. Its newly-released operating system Vista still pays homage to it by continuing to use a floppy disk as the icon for saving a document in Microsoft Word 2007.

Though you may not be able to find a manufacturer that produces computers with floppy drives built in by default, everyone will still understand that the floppy disk symbol means save. But in fact this is not new just with the end of the medium itself. That icon has long been used to mean save, even if it was to a hard disk: I hardly expect clicking on a button with the floppy disk icon to default to the floppy drive. Especially since such a drive does not exist on the machine I’m using now. (hmm, that makes me wonder if the verb drive is not also a similar case)