Well, no, this student doesn't. And come to think of it, neither did I at his age, in any direct sense; my knowledge of horseback riding came mostly from reading Westerns, together with large quantities of "horse books," vaguely descended from Black Beauty and urged on me by my older sister, such books being to us at one stage what videogames and DVDs are to kids today. But the agrarian world in which so much of our language is rooted had already vanished for the most part, dooming to slow (almost excruciatingly slow) extinction many idioms that had done good service for generations. The day had passed when people kept either pigeons or forty-pound grindstones in their sheds, and so the day was coming when otherwise bright youngsters would say hone in on, not home in on, mishearing it, oblivious to the underlying metaphorical logic. John Deere tractors had long since replaced oxen in the spring fields, so it was fated that a passionate letter writer would one day denounce "the yolk of tyranny." The rules of boxing had long since changed, and no doubt a significant minority had already begun to conceive of good behavior as a matter of towing the line. Unlike these whoppers, free reign at least makes good operational sense; the difference between leaving a king free to rule and leaving a horse free from rule will not be crucial for most purposes.

The wonder, really, is how long such usages survive the metaphorical and linguistic logic that first produced them: how long and how usefully. We still speak of reckless driving, knowing perfectly well what we mean, even spelling it correctly, without the slightest sense that reck once meant "obey," or that the underlying meaning is "disobedient driving." We say that such driving is "beyond the pale," registering only the faintest suspicion that the idiom has something to do with fence slats. If the DA chooses to prosecute it ruthlessly, we still understand, but unlike Milton we can no longer "melt with ruth" in the opposite mood. We complain of a spouse who wants us to be at his beck and call, but no longer realize that beck means a commanding c'mere-you gesture with the finger: a bit of vocabulary that has vanished quite unaccountably, since there seems to be no adequate replacement. When someone writes beckon call, a rather plausible misreading that replaces the dead word with an etymologically related live one, most of us still somehow recognize this as a mistake, resisting the folk etymology that tries to reintegrate the phrase into active paradigms. In such cases, it is solecism that has logic on its side, while correctness is the irrational passion. If English were logical, every child knows, reckless driving would be safe driving. The past tense of go would be goed, there would be daymares to go with nightmares, and half the lexicon would be spelled differently.

But, of course, the justification for all those senseless spellings that children must memorize one at a time, in sadistic languages like English and French, is historical. Through is spelled as it is because the word once was pronounced quite differently, and those letters in that order really were the best approximation to what people heard and said. The problem with proposals to reform English spelling is that they would only begin the cycle of change all over again. Throo might work for a while, but by 2050 the solecistic troo might be more accurate, while the grammarians of 2200 would have to anathematize the seductive, because perfectly accurate, dreh. On balance it is best to leave spelling alone for centuries at a time, condemning each new generation to much rote memorization of words whose spoken and written forms have drifted apart.

In just the same way, the lexicon swarms with semi-archaic usages that have outlived the linguistic logic that first produced them, but that persist as stubborn singularities, like stone outcroppings from an earlier geological era. Irregular verbs are the obvious example, preserving the remnants of a transformational logic that once applied much more broadly. The stem-changing paradigm that takes you from drink to drank once governed much of the domain of Old English. These days it still applies tolerably well to sing (sang), run (ran), sit (sat), and so on, but apply it to blink and you get blank, to glint and you get glont, to think and you get thank — or thunk, as the common joke has it, compounding the faulty past tense with a failure to distinguish the past participle. In the case of a highly irregular verb like be or go, the forms are so different that at first no logic at all seems to apply, but that is not quite right. Stroll from present-tense be to past-tense was and you seem to have lost the stem, but keep on going to the past participle, been, and there it is again, cooperating with a basically familiar rule that forms the past participle by adding –en, as in broke/broken, spoke/spoken, see/seen. Go/went/gone works the same way, with the same odd combination of lawlessness and vestigial orthodoxy. The reason, as Steven Pinker explains in Words and Rules, is that a kind of tectonic collision, in late Medieval times, fused entirely separate verbs. Many particular forms were obliterated in the cataclysm, and those that were left reorganized themselves into colorful hybrid forms, of decidedly mixed ancestry.

Clearly, like bones, trees, and fingernails, words do a great deal of useful work after they are mostly dead. We consent quite cheerfully to being disgruntled, inept, and uncouth, though almost no one can tell you what we were when we were gruntled, ept, and couth. We gird for battle, determined to give tit for tat, but only English teachers, ill-bred fellows that we are, giggle at the image of a jockstrap in the first case or hear a hint of bawdry in the nonsense syllables of the second. Our vocabularies are chock-full of expressions that, like chock-full, we understand less than perfectly, often using them in just one or two particular phrases. The memory of how they have been used before is sufficient to guide us in the next usage and the one after, without significant recourse to the derivation or root metaphor. One speaker in ten pauses to notice that the fury of a thunderstorm really is like the sudden swirling hostilities of a dog and cat fight; the rest of us say It's raining cats and dogs simply because that is what you say in such cases. No one likes to be left in the lurch, and on TV recently, I heard a Texas congressman, speaking of Iraq, insisting, "We will not leave a single American soldier in any lurch." But who really knows that lurch, in such cases, refers to a losing position in the game of cribbage?

A few years ago, someone spoke of the need to jump-start the economy, and the phrase has gone on to a slightly annoying vogue. These days, just starting is never good enough: if you are to start something at all, you must jump-start it. One suspects that the well-groomed, busy, New York-y commentators who are the chief culprits know rather little about cars, and no longer understand or clearly picture the concrete series of events that first made the metaphor so effective: one vehicle stalled, another driven carefully up alongside, long heavy jumper-cables used to connect the two, the charge from the second reviving the battery of the first, the first car now able to proceed without further assistance. But what the commentators do preserve, mostly, is the basic abstract meaning: the idea that something can be stalled and yet need only a brief and easy rescue, after which it will go on just as desired and intended. What began as expressive art persists as behavioral convention, at least for a while; we know how to use the phrase even when we don't quite get it. But with no clear image left to ground it, no nourishingly clear connection to paradigms operating elsewhere in the language, the precision of the abstract phrase is somewhat fragile and tenuous.

This state of affairs is worrisome, and we English teachers make our livings (such as they are) urging students to discover what words "really" mean, to use the dictionary, to attend to underlying metaphors and derivations. I have an exemplary lesson of that kind that involves Wilfred Owen's great poem, Dulce et Decorum Est. A soldier who has not been able to fasten his gas-mask in time is dying, horribly, at the front in World War I. The poet exclaims:

After the students and I have talked a bit, sometimes quite a bit, about the historical context and the poem's general drift, I sometimes ask, "By the way, what exactly does guttering mean here?" It quickly emerges that the students do not really know; the pure sound of the word, in which one hears the boy's desperate coughing, seems enough to justify its place in the line. Quite right, I concede; but to get the full force of the image, you have to imagine a time when people are more used to staring at candles than we are. You have to picture what happens when a candle has burned steadily for a long time, but then a gap forms in the thin rim holding in the melted wax. Suddenly wax leaks down the side of the candle — a gutter, get it? — and the flame flickers wildly or even goes out: just like the young soldier's life. What lovely precision!

Past a certain point, though, the hunt for real meanings becomes a fool's errand, ending in pedantry and etymological fallacy. Even here, we do not want to go all the way back to the root sense of "to form a gutter"; it is the more generalized metonymic extension of that meaning, "to flicker and die," that is truly apt. I had a professor in graduate school who once lamented that no one, these days, seemed to understand that the root sense of ecstasy was "standing apart," that is, out of oneself. This for him was an instance of that old bugbear, "the decay of language," a calamity that never happens or always happens, depending on your vantage point. But surely this was too pessimistic. Given that the abstract sense of ecstasy has remained stable, clear, and current for centuries, the attenuation of the image seems a normal tradeoff, a bargain price for the increase of conceptual range in a word that can connect such diverse experiences as a religious vision, a Super Bowl victory, a bite of fudge. In his brilliant recent book, The Unfolding of Language, Guy Deutscher argues that all languages are characterized by a continuous "flow from the concrete to the abstract," with names for things and actions gradually turning into names for ideas. Words do eventually die in the process, but the stream is constantly replenished, and the overall process is not a breakdown but a normal cycle, the way the work of language gets done. The tools that wear out are the ones that have been most useful.

Saussure famously asserted the separateness of language in its "synchronic" dimension from the "diachronic" or historical dimension. What a word means now, Saussure insisted, is in principle a separate question from that of the word's history or origin. Meaning is determined not by where the term has come from or what it did before, but by its atemporal relationship to the "field of signifiers"; it is langue, the snapshot of the lexicon in a frozen instant, that counts. Fiercely abstract and ultimately problematic, this formulation is nevertheless illuminating in all kinds of ways, and rough confirmation is never hard to find. If someone leans over to you at a party and whispers, "I think he's gay," knowing the pre-1960 sense of that term takes you farther from the intended meaning, not closer to it. If someone says you are nice, you know that you have not been called fussy. Frederick Pottle offers the example of the word fruition, which derives from the Latin verb for "to enjoy," and first enters English with the meaning "enjoyment." Once situated within the force field of English signifiers, however, the word becomes irresistibly associated with fruit, people begin saying things like fruition of our plans, and before long "completion" or "realization" is the accepted meaning, much to the grief of traditionalists. In the same way, the French chaise longue, on loan to English, gets hammered into "jayz long" or "shezz lounge" as we struggle to pronounce and make sense of it.

If the divorce of synchrony from diachrony were ever adopted as a prescription for usage, however, the result would be madness and anarchy. Part of the problem is that English (or any language) is no single integrated system, but a huge palimpsest of overlapping, partial, warring, half-destroyed systems, a site on which countless cities have risen and fallen and contributed to the wreckage. But even if we still spoke the language of Adam and Eve, single and pure, the intramural relations of words to each other would never be enough or nearly enough to determine meaning in particular cases. One needs, in addition, context, a working hypothesis as to intention, and abundant precedent. A hubcap is not a kind of hat, a parkway is not a place to park, and extra virgin olive oil is not Popeye's girlfriend tagging along on someone else's date, but there is no special reason they could not be. If we really experienced language as a timeless field (not that this is what Saussure is saying), the relative certainty of what phrases do mean would be swallowed up in the phantasmagorical vistas of everything they could mean.

Indeed, something like this does happen in each generation with children, who absorb language so quickly, in such huge quantities, with so few repetitions of so much of it, that for a while they really are like little ideal Saussurians, struggling furiously to squeeze meaning out of words in and of themselves. The consequent difficulties are often hilarious. I can remember believing that "ambush" meant to spring out from behind a bush, specifically, to attack someone. When I was younger still — too young to remember — I got jammed into a restaurant booth at a big family gathering, and as usual needed to go to the bathroom. I pulled my grandfather's sleeve (so the family tells the story) and said, "Grandpa, I need to go potty and I can't get out." He pointed to an available escape route and said, "Just go under the table." We ended up changing booths, and the waiter was not happy.

Three decades later, when my son was six, I instructed him one day to go fill our tub with lukewarm water so that we could bathe our large docile German shepherd. My plan was to get in the tub with her, the better to scrub and shampoo. I had put on my swimming suit, and I held her chest-high as I stepped barefoot into the tub — then leaped back out, screaming in pain. The terrified animal vanished; my son stared at me with deep dismay. "Dad, you told me to!"

When I could find my normal voice again, I said, "Jay, what does 'lukewarm' mean?"

"Really hot!"

Well, who could blame him? He was doing his best to get with the program, but the language offered no help; English does not describe tea as luke sweet or drab weather as luke sunny. There is no luke anything elsewhere in the language, but he could not have known that. Without realizing it — for such things happen unconsciously and instantaneously — he had taken his best guess, which turned out to be that luke was a prefix meaning "much more than" or "not just." One moral of the episode is that langue, that gloriously Platonic construct, is never really available in the existential sense: we all must work from our own experience in the language, which means a constant struggle to intuit the larger vision from insufficient data. Small wonder that we must cheat, inventing cultural helps like printing, dictionaries, mass communications, universities, cults of elocution, and English teachers, all working in one way or another to stabilize the meanings of words. Combining all these with our uncanny natural aptitude, we succeed so well that we can understand Shakespeare, separated from us by four centuries, though his contemporaries could not have understood Chaucer, separated from them by a mere two centuries.

Even so, there is a constant tension between what words do mean or have meant, and what I want them to do in this next utterance and the one after. One of the papers in the batch I am just now correcting uses the word gutturally to mean "courageously," by way of a bizarre derivation from "gut." Another author speaks of "the captain's maneuverability" in The Secret Sharer, meaning his skill in guiding his ship. These oddities will never catch on, but every so often, in the larger arena, an insurgent malapropism unseats an established correct meaning. "Beg the question" used to mean — as in formal contexts it still does — "to assume the point in question," that is, to argue tautologically. But these days, to my pain, the phrase is widely used to mean "raise the question" — nearly the opposite of what is still the dictionary meaning. The problem seems to be that we Americans no longer much use "the question" in the British sense of "the issue we are debating," so we hear not "begs [for rather than honestly wins] the question," but something like "begs [one to ask] the question." We have other idioms — dodge the question, evade the question — that are operationally so much like the original meaning that its loss is not felt, while raises the question is so drably unemphatic that the new, solecistic sense of the old debate-club formula fills a real need. Like fruition, the phrase lost the anchor that once held it in place, then drifted till an opportunistic scavenger tied it up in a new location. The ambiguity of words, the very thing that makes them so adaptable and inexhaustibly useful, also creates constant openings for misappropriation.

Over the centuries, poets have often claimed to be the creators of language, a madly presumptuous assertion if taken literally. Yet nearly everything in the phrasebook of English seems to have been somebody's bright idea at some point, and there is great pleasure to be had, often enough, in rediscovering the core of poetry in what we think of as literal speech. Deutscher's book shows very powerfully how all aspects of language, even grammatical systems, can be built up over the millennia by the steady accretion of worn-out metaphors, as if language were a great reef. Everything, it seems, has its brief poetic phase, when Speaker A captures some piece of the unsaid and delivers it to Speaker B, in a blaze of light. But this is followed by a much longer phase — a zombie phase, a fossil phase — during which we repeat the trick more or less unthinkingly, and creation dims into convention. During most of the new invention's service life, it is not expressiveness that stabilizes meaning, but habit and mimicry, our excellent unconscious memories for usage and our slavish disposition to follow suit. In this phase, it is not the poet we have to thank, but his or her dark double: the eternal curmudgeon, the gloomy Gileadite, the bilious knight of the red pen, guarding the threshold against the false poetry of solecism, keeping back the chaos of indeterminacy that always threatens.