The Language of Food: A Linguist Reads the Menu

by Dan Jurafsky

  From the earliest menus we have, these sweet dishes tended to congregate roughly toward the end of the meal, a property likely derived from their origin in Baghdad. The ordering probably comes from medieval models of health and digestion; sweets were believed to help digest heavy food. Baghdad cookbooks from the very first one, the al-Warrq’s Kitab al-Tabikh, written in Baghdad c. 950–1000, place all the sweet puddings and fritters, lauznaj, and crepes at the end of the meal. This is most mouthwatering to see in the fabulous meals of medieval Arabic literature like One Thousand and One Nights, which end in dessert after dessert, like the meal from the “Tale of Judar and His Brothers” of

  roasted chicken, roast meat, rice with honey, pilaf, sausages, stuffed lamb breast, nutty kunfa swimming in bee’s honey, zulbiyya “donuts,” qat’if pancakes folded around a sweet nut filling, and baklava.

  In “The Tale of the Sixth Brother,” after serving meat porridge, goose stew in vinegar, and marinated chicken fattened on pistachio nuts, the host presses his guest to take dessert. “Take this dish away and bring the sweets,” he says, offering almond conserve and fritters flavored with musk and “dripping with syrup” and almond jelly.

  In medieval Muslim al-Andalus the man who was credited for bringing these things west from Baghdad was Ziryab, a musician who arrived in 822 at the court of Abd-al-rahman II of Cordoba. Ziryab was the inventor of the Andalusian musical form. Legend says that he memorized tens of thousands of songs and that he stayed up all night discussing composition with the Jinns. Ziryab was said to have first proposed that meals be served in courses, starting with a lamb soup he invented called tafaya made with almonds and cilantro. The eleventh-century Córdoba historian Ibn Hayyan tells us that people even credited Ziryab with “inventing” many of the fabled desserts of al-Andalus like lauznaj and qatayif that came from Baghdad. Ziryab seems to have personified in these legends both the glories of Al-Andalus and the rich foods of the eastern court at Baghdad.

  A few hundred years later, a thirteenth-century Andalusian cookbook specifies that meals be served in seven courses, beginning with this tafaya (because it was particularly “healthy”) and ending with three courses of desserts and egg dishes. And the first Spanish cookbook published after the Reconquista, Roberto de Nola’s 1525 Libro de Cozina, says that meals at court still begin with soup and end with sweets and fruit.

  These fabulous foods spread to Europe, mainly through Muslim Andalusia and Sicily. The Manuscrito Anonimo, a medieval cookbook from Muslim Andalusia, gives recipes for dishes like zirbaja, an originally sweet-and-sour chicken stew, jullabiyya, chicken made with rose-syrup (sharâb al-jullâb, from the Persian word for rose), or lamb stewed with quince, vinegar, saffron, and coriander. These dishes were copied across Europe, first in Sicily, Naples, and England (all run by the Normans) and what we think of as “medieval” food developed: meat dishes seasoned with dried fruits, ginger, rosewater, and other Middle Eastern spices. The very first cookbook in English, the Forme of Cury, has recipes for rabbits in sugar, ginger, and raisins, or with honey and saffron; ground pork or chicken with dates in wine and sugar; and dishes like mawmanee and blankmaunger (savory puddings of sweetened boiled rice and almond milk with capon or fish that come from the medieval Arab dish ma’muniyya).

  Throughout Europe, however, the tendency of sweet dishes to be served toward the end of the meal was not yet a strict requirement in the Middle Ages. Many sweet dishes were served in the middle of the meal, and savories like capon pie or venison at the end. In fact the drop in the price of sugar around the turn of the sixteenth century led to an increase in sweet recipes throughout the meal (as well as linguistic consequences—"sweet" was one of Shakespeare’s favorite adjectives). Savory and sweet were intermingled, and a leg of mutton might be simmered with lemons, currants, and sugar, or chicken might be served with sorrel, cinnamon, and sugar, as in the following recipe for “Chekyns upon soppes” (basically chicken on cinnamon toast) from the 1545 early Tudor cookbook A Propre Newe Booke of Cokerye:

  Chekyns upon soppes.

  Take sorel sauce a good quantitie

  and put in Sinamon and suger

  and lette it boyle

  and poure it upon the soppes

  then laie on the chekyns.

  From their medieval position scattered throughout the latter half of the meal, however, sweet foods began to move slowly toward their modern place at the very end of the meal. Culinary historian Jean-Louis Flandrin carefully annotated the presence of sugar in French recipes over time, and found a sharp drop in the use of sugar in meat and fish dishes as French cuisine developed from the fourteenth to the eighteenth century, corresponding to a rise in sweet desserts. The use of sugar and fruits with meat, still prevalent in Moroccan, Persian, Central Asian, and even parts of Eastern European cuisine, slowly began to die out in France.

  The year 1600 marks about the halfway point in this transition toward a modern meal; at this time, French meat dishes were still often sweetened and dessert mainly still used to mean a light after-dinner snack or nuts, especially fruit or nuts. We know this from the first mention of dessert in 1612 in English, where it is described as a foreign “French” word in an early health and dietetics manual, William Vaughan’s Natvrall and Artificial Directions for Health: “such eating, which the French call desert, is vnnaturall, being contrary to Physicke or Dyet.”

  You might applaud Vaughan’s early warning about fat and sugar in rich “foreign” desserts, but in fact dessert still didn’t yet mean all those rich foods. Vaughan was referring to fresh fruit, and expressing concern about how hard it was to digest fruit at the end of the meal unless it was thoroughly cooked. This is an opinion that my Grandma Anna would have thoroughly agreed with; dinner at her apartment in the Bronx meant boiled chicken or boiled fish with boiled potatoes, and boiled fruit for dessert. When she visited us in California in the 1970s she would pick luscious ripe apricots and peaches from the trees and promptly boil them for compote.

  By a hundred years later, in the eighteenth century, the word dessert was borrowed into both British and American English. In British English the word retained its meaning of a light after-course. Given the American attitude toward food (something on the order of “Why eat an apple when you can eat an entire plate of cake and ice cream with whipped cream and chocolate syrup instead?”), you will not be shocked to know that by the time of our revolution, the word had shifted here to include more substantial sweet fare like cakes, pies, and ice cream. We know this because George and Martha Washington threw a party after Washington’s New York City inauguration of 1789, at their Manhattan mansion on Cherry Street, and Pennsylvania Senator William Maclay put down the menu in his diary: “The dessert was, first apple pies, puddings, etc.; then iced creams, jellies, etc.; then water-melons, musk-melons, apples, peaches, nuts.”

  By the nineteenth and twentieth century, the idea that sweet things belong only in the dessert course became relatively strict in the classical French cuisine represented by Escoffier. There were exceptions, but they were specific, like Escoffier’s canard a l’orange, and the vogue, especially after the rise of Nouvelle Cuisine, for duck magret with cherries or pan-seared foie gras with grapes or fruit conserves.

  In American food the boundary is a bit less rigid, as Catalan super-chef Ferran Adrià once observed (“A hamburger with ketchup and Coca-Cola? That’s the most intense symbiosis of sweet and savory imaginable”). And there are further remnants of medieval tastes in meats, what historian Ken Albala calls “throwbacks”—sweet-and-sour in barbecue, brown sugar and cloves for hams, fruit sauces with duck, or meats with cranberry sauce, apple sauce, and candied yams—preserved mainly in old-fashioned meals like Christmas and Thanksgiving. The great anthropologist Sidney W. Mintz, who pioneered the modern anthropology of food, notes that these fragments “demonstrate what anthropologists have long contended—that holidays often preserve what the everyday loses.”

  Besides these exceptional cases, however, we generally eat the
savory things earlier in the meal and the sweet things for dessert.

  Tracing the history of dessert demonstrated that a procession of the meal through savory courses with a sweet course at the end is a recent development in European cuisine. In other words, this particular sequence, and the idea of dessert, is something that some cuisines have (modern American, ancient Persian) and some don’t (classical Greek and, as we will see, Chinese).

  To explain how and in what way cuisines are different or similar, and how they change over time, I propose a theory called the “grammar of cuisine,” which suggests that a cuisine is like a language. The metaphor comes from linguistic grammar. The grammar of English, for example, consists of implicit rules that specify that adjectives tend to come before nouns (we say “hot fudge,” not “fudge hot”), or objects come after verbs (“eat chocolate” not “chocolate eat”). A grammar defines how linguistic parts are structured into a linguistic whole.

  Just as a language has an implicit grammar that native speakers know even if they can’t explain, a cuisine has an implicit structure, a set of rules about which foods go together, what constitutes a “grammatical” dish or meal in that cuisine. This implicit structure of cuisine consists of rules about how dishes are structured out of ingredients, meals are structured out of dishes, and entire cuisines out of particular flavor combinations and required cooking techniques. Each of these kinds of structuring helps explain the nature of cuisines and their similarities and differences.

  We’ve already seen one aspect of the grammar of cuisine: the ordering of meals. One constraint of American and European cuisine is “dessert comes at the end” and another, related to entrée, the default ordering of the American dinner, which we might represent as a kind of “rule” with sequence of dishes (using parentheses to indicate optional dishes):

  American dinner = (salad or appetizer) main/entrée (dessert)

  This rule states that an American dinner consists of a main course, preceded by an optional salad or appetizer (or both), and possibly followed by dessert.

  By contrast, French cuisine makes use of a cheese course, and a light green salad is often eaten after the main rather than before (and of course dessert at the end or, as the French say, salé puis sucré);

  French dinner = (entrée) plat (salade) (fromage) (dessert)

  Even moving one country over in Europe things change again; Italian cuisine has a distinct course (primo) that often consists of pasta or risotto:

  Italian dinner = (antipasto) primo secondo (insalata) (formaggi) (dolce)

  Some shifts in the ordering of the American meal are even more recent. Americans used to eat salad later in the meal, much as the French still do. The late MFK Fisher, one of America’s greatest prose stylists and my favorite food writer, suggests that the modern custom of eating salad before the main course arose in California in the early twentieth century. Fisher grew up in Whittier, just east of Los Angeles, around the First World War, eating fresh lettuce salad before the meal, and writes that her “Western” custom of starting a meal with salad shocked her friends from the East Coast who all ate salad after. (Meals on the East Coast in the first half of the twentieth century might instead begin with grapefruit, a custom that my other New York grandmother, Grandma Bessie, kept to all her life.)

  Despite the differences outlined above, the American and Western European meal sequences are pretty similar. By contrast, in Chinese cuisine a dessert course is not part of a meal at all. There was traditionally no exact Chinese word for dessert. The most frequently used modern translation, tihm ban in Cantonese, or tian dian in Mandarin, is most likely an extension, via borrowing from the West, of a word originally just referring to sweet snacks, not to dessert. The end of a traditional Cantonese meal, for example, is instead often marked by a serving of savory soup, or only occasionally (after the table is cleared) by fresh fruit.

  This explains why the tradition of fortune cookies developed in America as dessert. Jennifer 8. Lee’s The Fortune Cookie Chronicles tells us that little snacks stuffed with fortunes have been eaten in Japan since the nineteenth century. But only in the twentieth century in California did they begin to be served in Japanese and then Chinese restaurants as a dessert. The grammar of cuisine explains why: Chinese cuisine traditionally had no dessert course, and fortune cookies filled a kind of evolutionary niche for the final sweet cravings of American diners.

  The lack of dessert also explains why baking and hence ovens have a much smaller role in Chinese cuisine; there were no ovens in my kitchens in China, and even now Janet’s mom uses the oven in her kitchen in the San Gabriel Valley as a convenient place to store pots.

  Of course Chinese cuisine does have sweet foods, like the lovely sweet soups called tong sui (literally “sugar waters”), which can now be served as desserts, but more often act like snacks or small late-night meals. Janet and I often enjoy a one a.m. snack at my favorite sweet soup restaurant, Kowloon Tong out on Geary: the peanut soup called fa seng wu () with rice dumplings (tongyun ), tofu curds with honey, red bean soup, tortoise jelly, although I’m not quite as fond of the Chinese red dates with frog fallopian tubes (don’t ask).

  While Chinese meals don’t have the concept of a final sweet course, they do have structure of a different sort: constraints on the ingredients and their combination. A Cantonese meal, for example, consists of starch (rice, noodles, porridge) and nonstarch portions (the vegetables, meat, tofu, and so on). These can be mixed together in one dish (to form chow mein, chow fen, fried rice, and so on) or a meal can use plain white rice with the nonstarch served as separate dishes that each eater serves over their own portion of rice. Describing this in English requires the awkward word “nonstarch”; Cantonese has a word for this, sung . The word for “grocery shopping” in Cantonese is mai sung: “buying sung” (since the starch is a staple that would already be in the house). Thus a typical Cantonese meal consists of a starch plus a sung , or, to write a different kind of grammar rule, we might say:

  meal = starch + sung

  The grammar of cuisine does more than define the structure of a meal. Each cuisine also has implicit rules about the flavors that make up individual dishes. I like to think of dishes as words, and particular ingredients or flavor elements as the sounds (the “phones”) that make up a word or dish.

  Sounds differ from language to language, but they are also surprisingly similar. For example, every language seems to have a sound that’s something like English t or English p. Why should this be? The late linguist Ken Stevens (in his Quantal Theory of Speech) explained that humans all have the same tongue and mouth physiology, and sounds like t result from certain configurations of the tongue (and lips, and vocal cords) that are easy for speakers to make reliably and also result in sounds that are easy for listeners to distinguish reliably.

  Nonetheless, each language pronounces the universal t or p in a slightly different way. English t is different from Italian t or Cantonese t, French p is different from Spanish p. That’s the main cause of our accents in foreign languages: we become expert at saying a t in the English way and it’s hard to unlearn that and make a Japanese or a French t.

  Similarly, based on a different aspect of the very same human tongues, the ability to perceive flavor elements like sweet, sour, bitter, salty, and umami are universal. But each cuisine may express these universal flavor elements using ingredients that add their own culturally specific flavors.

  Each cuisine, for example, seems to have its own flavor element for sweet. My favorite is Malaysian gula melaka, a coconut palm sugar with a lightly smoky, caramelized taste. It’s easy to find palm sugar here, but not good gula melaka, so I have to bring it back the rare times I’m there, or rely on generous visitors from Asia.

  By contrast, the sweet taste of American food comes from refined white cane sugar or corn syrup, or, in special cases, maple syrup. In the gold rush days, San Franciscans instead used molasses for everything, pouring it on their food like ketchup. British and Commonwealth
desserts often use golden syrup, Mexican cuisine raw piloncillo sugar, and Thai cuisine palmyra palm sugar.

  The flavor elements for sour tend to be rice vinegars in China, tamarind in Southeast Asia, lemon juice or grain vinegar in the United States, sour orange or key lime in Central America, and wine vinegars in France (hence the name vin-aigre [sour wine]). Yiddish food is soured by crystals of citric acid called sour salt. This is what gives the sweet-and-sour flavor to my mom’s cabbage stuffed with rice, beef, and tomatoes (a dish I love but that my father refers to as “beef in shrouds”). Other universal flavor elements include salt or umami (from sea salt, salted olives, capers, soy sauce, fish sauce, fermented shrimp paste, anchovies, and so on).

  Not all flavor elements are universal. Combining different specific combinations of flavors is definitive of a cuisine, an idea that the late food scholar Elisabeth Rozin called the “flavor principle.” She pointed out that a dish made with soy sauce, rice wine, and ginger tastes Chinese; the same ingredients flavored with sour orange, garlic, and achiote tastes Yucatecan. Add instead onion, chicken fat, and white pepper (or for baking, butter, cream cheese, and sour cream) and you’ve got my mother and grandmother’s Yiddish recipes.

  Recent work has used computational techniques on online databases of recipes to test cross-cuisine generalizations of Rozin’s flavor principle at the molecular level. Y. Y. Ahn at Indiana University and his colleagues examined 60,000 online recipes to test the “food pairing hypothesis,” a recent theory proposing that tasty recipes are more likely to combine ingredients that share flavor compounds. For example tomatoes and mozzarella share the compound 4-methylpentanoic acid. Ahn and his colleagues found an interesting difference between North American and Western European recipes, which do indeed tend to pair ingredients that share flavor compounds, and East Asian recipes, which combine ingredients (like beef, ginger, cayenne, pork, and onions) that don’t have overlapping compounds. The difference suggests that preference for similar or different compounds may be part of what makes up a cuisine.