Surfaces and Essences

by Douglas Hofstadter

  What Makes One Say “But” Rather than “And”?

  Occasionally one hears sentences like “I don’t know what country the florist comes from, but she seems very nice.” Why the “but” here? What kind of a zigzag in discourse space is this? Well, first consider how it would sound with “and” instead: “I don’t know what country the florist comes from, and she seems very nice.” It simply sounds like a non sequitur. One wonders what these two thoughts are doing in the same sentence. On the other hand, with “but”, there is a definite logical flow, although it’s a bit subtle to pin down. The feeling being expressed is something like this: “Despite my near-total lack of knowledge about her, I would say that she seems affable.” “Despite” is a concessive that is a close cousin to “but”. The point is that the first part of the sentence is about a hole in one’s knowledge, and the second part is about a small but significant counterexample to that tendency. Thus the first part of the sentence suggests a pattern and the second part states an exception to the pattern. Whenever we are about to tell someone a “piece of news” and just before doing so we realize that in some way or other it goes against expectations likely to be set up by what we had just told them moments earlier, we have detected the telltale signs of a but situation. The two-clause sentence about the florist has exactly that property, and that’s why putting “but” between its clauses makes sense and sounds right to our ears, whereas putting “and” there would make it sound very strange.

  Likewise, if someone says “He has big ears, but he’s really a nice guy”, it doesn’t mean (despite the way it sounds on the surface) that the speaker has a stereotype of large-eared people as being unpleasant. Rather, it means something more like, “Although this person is on the negative side of the norm in a certain physical way, he is on the positive side of the norm in terms of his behavior.” Once again, we see that the conjunction “but” signals a swerve in discourse space — the person in question is on one side of one norm and yet (despite that fact) is on the other side of another norm.

  The category of swerves that the word “but” denotes is just as real as the category of swerves made by vehicles on roads, though it is more intangible, and the use of the word “but” comes about because as people speak, they are always paying some amount of attention to their trajectory in discourse space and are categorizing its more familiar aspects in real time, just as they are always paying some amount of attention to the scene before their eyes (and the sequence of sounds coming into their ears, etc.) and are categorizing its more familiar aspects in real time.

  Sometimes a speaker becomes aware of the real-time linguistic self-monitoring going on as a background process in their brain, and this can affect the flow of speech. It can result in one verbal label being canceled and swiftly replaced by another label. One example is when someone says, “Oh, look at that horse — uhh, I mean donkey”. The following story involves such a relabeling, but the self-correction involves an event in the speaker’s linguistic output stream rather than an object in the environment.

  Frank and Anthony, lifelong friends, hadn’t seen each other in a long time and were pleasantly catching up on the news of each other’s families. Frank wanted to tell Anthony about his daughter, who had been hit by a mysterious illness and, to everyone’s relief, recovered from it after a couple of years. One of his sentences went like this: “She got to be an excellent skier during her stay in Montana, and one day on the slopes she just couldn’t keep her balance — or rather, but one day on the slopes she just couldn’t keep her balance…”

  As he launched into his sentence, Frank thought he was in an and situation, and then suddenly — or rather, “but suddenly” — when he started to flesh out the second clause, he clearly heard the abrupt swerve in what he was telling Anthony (it would strongly violate anyone’s expectations that a highly accomplished skier will, without any warning, start to fall a lot), and so he quickly spun in his tracks and, changing conjunctions in mid-stream, jumped from “and” to “but”, as he realized that from a listener’s point of view, the story he was relating involved a kind of zigzag — thus a member of the category of but situations rather than of the category of and situations.

  Further Refinements in Discourse Space

  Making the distinction between and situations and but situations is not a high art, but it is a most useful skill to pick up, and that dichotomy is perfectly adequate in many situations. However, there are numerous subcategories inside the broad categories that we’ve labeled “and situations” and “but situations”, and people, first as children and later as adults, gradually pick up the finer nuances that will help them to recognize these subcategories and thereby to choose, in real time, the sophisticated connecting word or phrase that best describes the situation in discourse space.

  Sticking to just the categories and and but while making no finer distinctions is rather like making the useful but coarse distinction between the categories of car and truck, but not venturing into finer details. The car/truck distinction is good enough for many purposes. People who are fascinated by motor vehicles, though, are eager for much more detail, and they’ll often use a much narrower category than is designated by the generic word “car”. In the same way, fluent speakers depend on making finer distinctions than just the coarse “and”/“but” dichotomy. However, just as recognizing whether a vehicle is a Honda or a Hyundai, a coupe or a sedan, automatic or manual, fuel-efficient or gas-guzzling, sporty or family-style, and so forth, takes considerable experience, so deciding whether one finds oneself in a nonetheless situation in discourse space, a however situation, an and yet situation, a still situation, an on the other hand situation (and so forth) is a subtle skill, since it requires having constructed these subcategories and having a decent mastery of them.

  We have no need to delve into the subtleties that underlie such choices. Just as it is not our aim to explain how people distinguish among studies, studios, offices, dens, ateliers, cubicles, and workplaces, or among their friends who are agitated, antsy, anxious, apprehensive, concerned, disquieted, distressed, disturbed, fidgety, frantic, frazzled, frenetic, frenzied, jittery, nervous, perturbed, preoccupied, troubled, uneasy, upset, or worried, or between situations calling for “Thanks a million”, “Thank you ever so much”, “Many thanks”, and other expressions of gratitude, so it is not our aim to explain the nature of the nuances that lead a person to choose to say “however” rather than “but” or “nonetheless” or “actually” or “and yet” or “that having been said” or “despite all that”. We are concerned not with pinpointing the forces that push for choosing one or the other of these linguistic labels, but simply with the fact that each of these different phrases is the name of a subtly different mental category — a highly characteristic, oft-recurring type of pattern in discourse space to which one can draw analogies.

  We might point out here that where English has two most basic conjunctions (“and” and “but”), Russian has three — (“and”), “HO” (“but”), and “a” (whose meaning floats somewhere between “and” and “but”). This means that Russian speakers and English speakers have slightly different category systems concerning very basic, extremely frequent phenomena that take place in discourse space. Picking up the subtleties of when to use “a” instead of or “HO” takes a long time. It’s much the same story as for any set of categories that overlap. We don’t want to give a linguistics lesson, so we’ll stop here, but the bottom line is that words that to most people seem infinitely far from the most venerable and clichéd examples of categories (such as chair, bird, and fruit) are nonetheless the names of categories, and they are so for the very same reasons, and the categories they name act very much the same.

  Ever More Intangible

  It might seem logical for a chapter on words to move from the most frequent ones to rarer ones, but we will go against expectations here. We want to finish up by talking about some of the most frequent words of all, which, like “and” and “but”, ar
e almost never thought of as being the names of categories. Consider words like “very”, “one”, and “too”, for instance. What category does “very” name? Of course we can’t literally point to members of the very category the way we can point to members of the category dog, say. Still, let’s try for a moment. Usain Bolt is a very fast runner. Cairo is a very big city. Neutrinos, they are very small. That’s very you. There; that’s enough to give the feeling. Much like much, very is a category having to do with norms built up over a lifetime of prior experience. Where Rome is a big city, Cairo is a very big city.

  We learn to use the word “very” just as we learn to use the word “much” — by hearing examples of its usage and feeling our way around in the world of sentence construction. Does the fact that the crux of the notion very has to do with the formation of sentences disqualify it from being a concept? No, not at all. The concept very is just as genuine a concept as is dog. The concept very is all about relative magnitudes, expectations, importances, intensities. All of that is deeply conceptual.

  And while we’re at it, let’s not forget that Albert Einstein was one very smart dude. Yes, no doubt about it, Einstein was one smart dude, as opposed to being several smart dudes; but why was he not just a smart dude? The word “one” can convey more information, it seems, than just the number of items that somebody is talking about. In this case, saying “one smart dude” emphasizes the extreme rarity of a genius of Einstein’s caliber; it is a subtle way of squeezing extra information into the sentence via a very unexpected channel. However, the choice of the word “one”, as opposed to the word “a”, also conveys information about the persona of the speaker (earnest, candid) as well as about the tenor of the conversation (informal, casual). Moreover, using the word “dude” strongly resonates with using the word “one”, and vice versa — indeed, when used together, these two words paint a vivid portrait not only of Albert Einstein but of a certain brand of English speakers who are prone to use this kind of phrase.

  To put it more explicitly, probably most native speakers of American English have developed a category in their minds that could be labeled “the kind of person who goes around saying ‘one smart dude’ ”. However, the category is not as narrow as this label suggests. To be sure, it would be instantly evoked if one were to hear the above remark about Albert Einstein’s intelligence, but its evocation doesn’t depend on having heard the specific words “smart” and “dude”; it would also be evoked by remarks like “Doris Day was one cute cookie” or “That’s one bright lamp!” We thus see that even bland little words like the numeral “one” intoned in a certain fashion, which might seem very close to content-free, can evoke rich and subtle categories in our minds.

  Having just considered “one”, let’s move along to “too”. Of course that word has two quite separate meanings — namely, “also” and “overly much” — so let’s focus on just the latter. What are some quintessential members of the too category? Well, perhaps the idea that eating a whole fudge cake would be too much. Or the idea that teaching general relativity to elementary-school kids would be too early. We’ll let readers invent their own too situations. The point is that doing this little exercise will make it vivid for you that there are analogies linking each too situation to other too situations, and thus to the abstract concept of too-ness.

  When we considered the concept much, we pointed out that part of its richness is how it is used in sentences. Indeed, the realm of discourse is one of the richest domains we humans come into contact with. Just as there are concepts aplenty in the worlds of linear algebra, molecular biology, tennis-playing, and poetry, so there are concepts galore in the worlds of discourse, language, grammar, and so forth, but we seldom think about them. Thus a high-school student might pen a poem in flawless amphibrachic hexameter without ever suspecting that there is a standard name for such a meter. Likewise, we native speakers of English are all past masters in the use of words such as “the” and “a” without ever analyzing how they work. But the Polish linguist Henryk Kałuża wrote a whole book — The Articles in English — to teach non-native speakers “the ins and outs” (one of his examples of “the”) of our language’s definite and indefinite articles. As it turns out, Kałuża’s book is all about the meanings of these rich words, but nonetheless, some people resist the idea that “the” and “a” have meanings, arguing that they are not “content words” but just grammatical devices. It seems that since these words do not designate tangible objects, some people think they are devoid of meaning (not unlike people who insisted for centuries that zero isn’t a number). It seems strange, however, to suggest that the difference between “the president” and “a president” has nothing to do with meaning. There is a great deal of content conveyed by the distinction between “the sun’s third planet” and “a sun’s third planet”, between “I married the man in the photo” and “I married a man in a photo”, between “the survivor died” and “a survivor died”.

  Trying to pin down how words like “the” and “a” are used in English is not our purpose here — no more than trying to specify the type of circumstances likely to evoke the word “office” as opposed to the word “study”. What we are emphasizing is that this subtle knowledge is picked up over many years thanks to one analogical extension after another, usually carried out without the slightest awareness of the act.

  And thus we have moved our discussion from fairly low-frequency words, like “hub”, “attic”, and “moon”, to the very top of pile — the most frequent word in all of English — the definite article “the”. In so doing, we have also moved from very visual, concrete phenomena to phenomena that are largely intangible and mental. But what’s crucial is that in making this move, we have never left the world of categories. Just as “hub” denotes a category (or perhaps a couple of different categories — the centers of bike wheels as opposed to certain major airports), so “the” denotes a category (or perhaps a few distinct ones, as the world-class “the”-expert Henryk Kałuża would be quick to point out).

  Carving Up the World Using a Language’s Free Gifts

  Any language has an immense repository of labels of categories that people over millennia have found useful, and as we grow up and then pass through adulthood, each of us absorbs, mostly by osmosis, a decent fraction of that repository, though far from all of it. The many thousands of categories that we are handed for free and that we welcome, seemingly effortlessly, into our minds tend to strike us, once we have internalized them, as self-evident givens about the world we live in. The way we carve the world up with words and phrases seems to us the right way to view the universe — and yet it is a cliché that each language slices up the world in its own idiosyncratic manner, so that the set of categories handed to speakers of English does not coincide with the set handed to speakers of French, or to those of any other language. In short, “the right way” to see the world depends on where and how one grew up.

  A striking example is provided by English and Indonesian. The English words “brother” and “sister” seem to us anglophones to cover the notion of siblinghood excellently, as well as to break that concept apart at its obvious natural seams. However, the Indonesian words “kakak” and “adik” also cover the notion of siblinghood excellently, but they break it into two subconcepts along an entirely different axis from that of sex: that of age. Thus “kakak” means “elder sibling” while “adik” means “younger sibling”. To speakers of Indonesian, this seems the natural way to slice up the world; they don’t feel a need to be able to say “sister” using just one word any more than anglophones feel a need to be able to say “older sibling” using just one word. It doesn’t cross their minds that something is missing from their language. Of course Indonesian speakers can say “female kakak or adik”, and that effectively means “sister”, just as we English speakers can say “older brother or sister”, and that effectively means “kakak”. Each language can express through a phrase what the other language expresses throu
gh a word. And the French language does an admirably diplomatic job with these concepts, managing to slice the world up in both ways. The male/female dichotomy tends to be the more frequently used one in French (“frère” vs. “sœur”), but the older/younger one exists just as well (“aîné” vs. “cadet”), and thus all possibilities are available. As this shows, slicing the world up at its “natural” joints is not quite so natural as one might think.

  Different ways of cutting up the world are far from being exceptional picture-postcard rarities. In order to unearth good examples of the phenomenon, one certainly doesn’t need to resort to pairs of languages that are spoken halfway around the world from each other. We can find plenty of them right under our nose, simply by poking about a bit in the languages that are closest to our native tongue, even limiting our search to words and concepts that are unquestionably central.

  Thus, nothing seems more obvious to us anglophones than what time is. We know what time it is right now, we know how much time it will take to drive to the airport, and how many times we’ve done so before. These three ideas strike us as being very clearly all about just one central, monolithic, and hugely important concept: the concept known as “time” (in fact, the most frequent noun in the English language). And yet, most strangely, there are languages that don’t see those three ideas as being about the same concept at all! If you’re a francophone, you know what heure it is right now, you know how much temps it will take to drive to the airport, and how many fois you’ve done so before. They aren’t the same word or even related words, and the three concepts labeled by the words “heure”, “temps”, and “fois” seem quite distant from each other for French speakers. As if this weren’t bad enough, the French word “temps” doesn’t denote only a certain subvariety of English’s concept of time — in addition, a good fraction of the time, it means “weather”. Thus speakers of French, in their whimsical fashion, somehow manage to confuse the weather and the time! On the other hand, we speakers of English manage to mix up the hour of the day with the number of occasions on which something has happened! Which mistake is sillier?