A theory for semantic drift

Originally published on Language and Philosophy, April 1, 2011

Okay, here’s a theory to account for some of the variety in semantic drift, the gradual change of a word’s meaning.

Going back to old Saussure, the sign has two sides, the physical shape (for speech, this is the sound of the word) and what the word means. You’d think that semantic shift would only happen on the meaning side, but semantics plays on both sides of the sign because both sides relate to other signs in the language.

Let’s expand Saussure’s duality a bit with Frege’s distinction between sense (something like idea) and reference (the real world objects determined by the idea). The meaning of the sign can shift if the idea drifts, expanding (losing information) or contracting (becoming informationally richer) or just replacing some information with new information. The many pressures or inclinations on idea drift have been well observed and studied in the literature.

The physical sound-shape side of the sign can be the source of meaning shift as well, odd as that may seem. Why would an arbitrary sound have an effect on meaning? Well, for example, sound shapes in English that bear strong resemblance to Greek or Latin words tend to be treated as more serious and formal than monosyllabic Anglo-Saxon words. That seriousness affects their semantic value. Ask a class of students which endeavor is more fun and which more serious, athletics or sports, and you will get a 90% agreement that sports are a less formal activity, although if you then ask whether they denote the same set of activities, you’ll get 100% agreement that they do (and a puzzled class of students).

Speakers have a blind spot in their linguistic capacity. They have great trouble distinguishing between the word and its meaning. There’s nothing surprising about it: we do a lot of our thinking in words, so the two — thought and word — incline to meld into one another.

So it is also not surprising that the formality of a word can influence its semantic shifts. Our social attitude to a word’s sound-place in the language is part of its meaning.

So far we’ve got the following aspects of meaning:

– idea

– social attitude to the idea and relation of the idea to other ideas

– attitude to the sound shape and the relation of the sound shape to other sound shapes

– reference.

Less observed is the possibility of drift caused by the reference.

The set of real-world objects referred to by a word is determined by the idea. The idea of the word “cat” determines the set of felines. When the idea is employed colloquially as in “hep cat,” the idea determines that “cat” denotes the set of counterculturally acceptable males. Because the referent depends on the idea, drift in the idea has received most scientific attention. But the referent, even though fixed by the idea, can be the source of semantic shift as well, because social attitudes to things is not fixed.
“Democracy” denotes a specific set of government types, but that set has not always been valued in the past as it is today. In the U.S., “democracy” is viewed as almost synonymous with “just” and “right.” It wasn’t always and it need not always be.

Consider the denotation of “woman.” The boundaries of the set haven’t changed, and the properties that make a human a woman haven’t changed, but social attitudes have and the social place of women has. Surely these changes, which relate to the referent, not to the idea or the sound shape, have shifted the meaning.

“God” is another word that has surely shifted through changing attitudes to its referent, though discussing it presents the difficulty of dealing with an elusive referent.

Consider “computer.” The rapid technological advances have altered the set almost beyond recognition, from a room-sized machine exclusive to universities and military labs, to the palm pilot. That’s a reference change that directly shifts the idea.

One reason linguists don’t spend much effort observing reference-based shifts, is that those shifts depend on non linguistic phenomena, and so don’t instruct much about the nature of language itself. It would be useful, however, to learn the extent to which reference shift can be tolerated by a word and the broader linguistic effects of reference shift.

Then there’s metaphor. In a separate post I mentioned that metaphors trade on a few specifics of an analogy. They can dilute information as in

– the foot of the mountain

now means just the bottom — the toes, heel and arch are lost.

This is all just a start, but the point here is that semantic shift can occur on either side of the sign, the reference or the idea, the signified or the signifier.

I’m going to stop here for now, but there’s more in the first post on this blog, which gives a bunch of surprising facts and more surprising dynamics about semantic drift in gender words. The big surprise there is that euphemism frequently causes its opposite, pejoration.