The San Francisco midsummer fog was late in coming this year, which means Janet and I got a fantastic view of the July 4th fireworks (legal and not-strictly-legal) from the top of Bernal Hill. Hot days are rare in San Francisco, so random strangers have been smiling at each other on Mission Street and the lines are extra-long on the sidewalks in front of the ice creameries.
You may not be aware of the close relationships among these summer phenomena. Ice cream was invented by modifying a technology originally discovered for fireworks. And the way ice creams flavors are named turns out to have a surprising relationship with the evolutionary origin of the human smile.
Ice cream has always been popular in San Francisco; Swensons, Double Rainbow, and It's It were all founded here, and Rocky Road ice cream was invented across the bay in Oakland during the Great Depression. Prices are not Depression-era at the latest upscale creameries, though, where you'd be lucky to walk away with a pint of ice cream for less than seven dollars. At Smitten, in Hayes Valley, for example, they'll make your ice cream fresh when you order it, freezing it with liquid nitrogen. At other places the selling point is the the unusual flavors (or their interesting names). At Humphry Slocombe you can get foie gras, pink grapfefruit tarragon, or strawberry black olive. Bi-Rite Creamery will happily sell you honey lavender, balsamic strawberry, and salted caramel. Mitchell's specializes in Filipino and other tropical flavors, including halo halo, lucuma, ube (purple yam), and avocado. And Mr. and Mrs. Miscellaneous seems to keep running out of their latest hip flavor, orange blossom.
Well, actually, it turns out that orange blossom is not a newfangled flavor. Orange blossom is, in fact, the original ice cream flavor, appearing in the earliest recipes by the mid 1600's, the period when ice cream was invented. Ice cream was served in the Restoration court of Charles II as early as 1671, and food scholar Elizabeth David gives us what may be the English royal recipe, handwritten in Grace Countess Granville's Receipt Book by the 1680's:
The Ice Creame
Take a fine pan Like a pudding pan ½ a ¼ of a yard deep, and the bredth of a Trencher; take your Creame & sweeton it wth Sugar and 3 spoonfulls of Orrange flower water, & fill yor pan ¾ full...
By about 1696, a later edition of La Varenne's cookbook suggests using fresh orange flowers:
You must take sweet cream, and put thererto handfuls of powdered sugar, and take petals of Orange Flowers and mince them small, and put them in your Cream, and if you have no fresh Orange Flowers you must take candied, with a drop of good Orange Flower water...
And by 1700 other ice cream flavors were developed as well, including pumpkin, chocolate, and lemon, as well as a plethora of early sorbets: sour cherry, cardamom, coriander-lemon, and strawberry.
Where did these flavors come from? The use of orange flower should give you a clue: the historical roots of ice cream and sorbet, like many of our modern foods, lie in the Muslim world.
Fruit syrups, and the refereshing drinks made from mixing them with water, are called sherbet in Turkish and sharbat in Persian, from Arabic sharbah, from shariba `to drink'. These chilled (but not frozen) drinks have been popular throughoutt the Ottoman, Arab, and Persian worlds continuously since the Middle Ages. On the left is a Gül Şerbeti (rose sherbet) from a modern Turkish cookbook; and Claudia Roden talks nostalgically of the sharbat of Egypt, flavored with lemon, rose, violet, tamarind, mulberry, raisin, or liquorice.
By the 16th century Italian and French travelers had brought back words of these Turkish sherbets. In one of the earliest mentions of the word in Europe, the French naturalist Pierre Belon in 1553 described sherbets in Istanbul made of figs, plums, apricots, and raisins. Thirsty passers-by would buy a glass of syrup from wandering sellers or stands, mixed with water and chilled with ice. But by 1615 sherbets were still unavailable in Europe; here's an excerpt from a letter an Italian traveler sent home in 1615 from Constantinople, from Elizabeth David's lovely book "Harvest of the Cold Months":
scerbet, a certain composition which they make... of sugar, lemon juice, seasonings of fruit and flowers and other ingredients, something like the conserves and marmalades of Naples; when they want to drink, they put some of this composition in a jug of water...
These sherbets were the source of the fruit ices that we now call sorbets. But the Ottoman drinks (and the modern Middle Eastern ones as well) were not frozen; they were cooled with ice or snow just like modern lemonade. People had been putting ice and snow into drinks to cool them for over 4000 years, but freezing sweetened fruit juice or cream requires a much lower temperature than just ice can achieve.
So where did the idea and the technology for freezing arise? Obviously liquid nitrogen, the darling freezing technology of modernist cuisine, was not available in the 16th century.
The insight came from fireworks. In the 9th century, during the Tang dynasty, the Chinese first realized that saltpeter (potassium nitrate) could be mixed with sulpher and coal to create the explosive mixture we now call gunpowder. Gunpowder was quickly adopted by the Muslim world, where potassium nitrate was called Chinese snow in Arabic and Chinese salt in Persian.
But it was in the Arab world rather than China that the processing of purifying and refining potassium nitrate was perfected, and it was here in Damascus that it was discovered, probably by the Damascus physician Ibn Abī Uṣaybi'a, in his 1242 History of Medicine ("Uyūn al-ānbā")) (although he credits a lost work from an earlier Muslim physician, Ibn Bakhtawayh, from 1029), that saltpeter had refrigerating properties: when potassium nitrate (saltpeter) is added to water, it chills the water. Dissolving salts like potassium nitrate (KNO3) in water breaks the bonds between the ions, drawing heat from the surrounding water. This endothermic reaction, the basis of the modern cold pack shown to the right, can drop the temperature of the water enough to freeze pure water, although not low enough to freeze fruit ices or ice cream.
By the early 16th century this discovery was widely used in Muslim India to chill water for drinking. At this time most of what is today northern and central India, Pakistan, and Bangladesh, as well as parts of Afghanistan, was ruled by the Mughal emperor Akbar the Great. The Mughals were originally Turkic speakers from central Asia, and the royal line that conquered Delhi traced their descent from Genghis Khan (Mughal was the Persian word for Mongol), but had adopted the Persian language and culture. By the time of Akbar, the Persian-speaking court at Agra was a center for the arts, architecture and literature. The Ramayana and the Mahabharata were translated from Sanskrit to Persian during this period, and Akbar's keen interest in painting and architecture led to the development of styles of art that mixed Persian, Hindu, and European forms. Like many places where scientific and culinary innovation and mixing flourished (Moorish Spain, early Norman Sicily), Akbar's reign was a beacon of relative religious tolerance, in which the tax on non-Muslims was eliminated and other religions were allowed self-government. Agra (and his later court in Lahore) were steamy hot, and drinks were cooled by spinning a long-necked flask in saltpeter-water. Here's a 1596 description from the Ain I Akbari:
One sér of water is then put into a goglet of pewter, or silver, or any other such metal, and the mouth closed. Then two and a half sérs of saltpetre are thrown into a vessel, together with five sérs of water, and in this mixture the goglet is stirred about for a quarter of an hour, when the water in the goglet will become cold.
Very quickly this idea of using saltpeter to cool water was adopted in Italy. Blas Villafranca, a Spanish physician working in Rome published the idea in 1550, saying that this saltpeter bath had become the common method of cooling wine in Rome. On the left is his picture of the method, showing a bulbous flask clearly adapted from the Indian flasks above; the shape makes it easy to speed up the cooling by turning the bottle in the cold bath.
In 1589 the next step in ice cream technology was taken by the Neapolitan Giambattista Della Porta. In the 2nd edition of his "Magia Naturalis" he experimented with adding saltpeter to snow rather than to water. The result successfully froze watered wine. Della Porta's combination was a happy accident; it was not saltpeter's endothermic reaction with water that caused cooling when mixed with ice, but a completely different chemical property. Adding a solute (anything will do) lowers the freezing point of water, by interfering with the crystal structure of the ice. Adding salt or potassium chloride slowly draws water out from its chrysal mixture, and since the freezing point is lowered, turns into a salty slush. The phase shift from solid to liquid takes energy (another endothermic reaction), resulting in an even colder freezing brine that reaches -20 degrees C, easily cold enough to freeze ice cream or fruit ices.
Sometime between 1615 and 1650, the Neapolitans combined the liquid Ottoman sherbets with the newly invented saltpeter-and-ice freezing method, resulting in a new food: frozen sherbets or frozen sorbets. The idea of freezing other liquids like milks and custards soon followed. We don't have any of these early Italian recipes, the way we have early English and French recipes, but evidence for the Italian innovation comes from contemporary French ice cream makers who discussed learning their recipes from Italy. Soon afterwards the Italians also figured out that common salt worked better than saltpeter for freezing (salt is a smaller molecule than saltpeter; the smaller the molecule, the more ions from each gram of solute interferes with freezing); by 1665 the English chemist Robert Boyle said that ice and common salt was the method "much employ'd" in Italy to chill drinks and fruit.
As for the names of the flavors, mostly they are just the names of the ingredients ("chocolate", "strawberry", "orange blossom", and so on). We commonly assume that such flavor names are purely descriptive, and that factors like the sounds of the names should have no bearing on how the ice cream tastes. To paraphrase Shakespeare's Romeo and Juliet:
What's in a name? that which we call a rose sherbet
By any other name would smell as sweet;
Juliet was roughly correct; the sounds (or "phones") that make up a word don't generally tell you what the word means. By 500 BC Plato (in the Cratylus) and the Chinese linguist Xunzi of the Chinese Warring States period had figured out that the relationship between sound and meaning is usually arbitrary. A moment's thought makes it clear why this must be true: different languages have totally different sounds for the same concept, and languages only have around fifty or so phones, and obviously have a lot more ideas to express than fifty.
But it turns out that research over the last century has shown that Shakespeare was wrong; sometimes the sounds of a name do influence how people perceive ice cream. The phenomenon of sounds carrying meaning is called "sound symbolism". Sound symbolism has been most deeply studied with vowels, and in particular the difference between two classes of vowels, front vowels and back vowels, which are named depending on the position of the tongue. The vowels i (the vowel in the words cheese or bean) and ɪ (the phonetics symbol is a small capital I, pronounced as in mint or slim) are front vowels. because they are made by holding the tongue high up in the front part of the mouth. The picture to the left shows a very schematic cutaway of the head, showing the lips and teeth on the left, and the tongue high up toward the front of the mouth.
By contrast, the vowel ɑ (as in large, pod, or on) is a low back vowel; this sound is made by holding the tongue lower in the back part of the mouth; other back vowels are o (as in cold) and ɔ (as in the word pour or my mother's New York pronunciation of ought). The picture to the right shows a very schematic tongue position for these vowels; lower in general, and more toward the back of the throat.
A number of studies over the last 100 years or so have shown that front vowels in many languages tend to be used in words that refer to small, thin, light things, and back vowels in words that refer to big, fat, heavy things. It's not always true, but it's a tendency that you can see in any of the stressed vowels in words like little, teeny or itsy-bitsy (all front vowels) versus humongous or gargantuan (back vowels). Or the i vowel in Spanish chico (front vowel meaning small) versus gordo (back vowel meaning fat). Or French petit (front vowel) versus grand (back vowel).
In one marketing study, for example, Richard Klink created pairs of made-up product brand names that were identical except for having front vowels or back vowels: nidax (front vowel) verus nodax (back vowel), or detal (front vowel) versus dutal (back vowel). For a number of hypothetical products, he asked people which seemed bigger or smaller, or heavier or lighter, with questions like:
Which brand of laptop seems bigger; Detal or Dutal?
Which brand of vacuum cleaner seems heavier, Keffi or Kuffi?
Which brand of ketchup seems thicker, Nellen or Nullen?
Which brand of beer seems darker, Esab or Usab?
In each case, the participants in the study tended to choose the product named by back vowels (dutal, nodax) as the larger, heavier, thicker, darker product. Similar studies have been conducted in various other languages.
The fact that consumers think of brand names with back vowels as heavy, thick, richer products suggests that they might prefer to name ice cream with back vowels, since ice cream is a product whose whole purpose is to be heavy and rich.
Indeed, it turns out that people seem to (at least mildly) prefer ice creams that are named with back vowels. In a study in the Journal of Consumer Research Eric Yorkston and Geeta Menon had participants read a press release describing a new ice cream about to be released. Half the participants read a version where the ice cream was called "Frish" (front vowel) and the other half read a version where it was called "Frosh" (back vowel), but the press release was otherwise identical. Asked their opinions of this (still hypothetical) ice cream, the "Frosh" people rated it as smoother, creamier, and richer than the "Frish" people, and were more likely to say they would buy it. The participants were even more influenced by the vowels if they were simultanously distracted by performing some other task, suggesting that their response to the vowels was automatic, at a non-conscious level.
If people subconsciously think of ice cream names with back vowels as richer and creamier, it suggests that actual ice cream brands or flavors might also use back vowels. So I ran what Mark Liberman calls a Breakfast Experiment; a quick experiment using some easy-to-access language data. My hypothesis was that we would see more back vowels in names of actual ice cream brands or flavors. Furthermore, if front vowels indeed indicate thin, small, light , we should expect more front vowels in foods that supposed to be thin and light, like crackers.
To test the hypothesis I downloaded two lists of food names from the web. One was a list of 81 ice cream flavors that I constructed by including every flavor sold by either Haagen Dazs or Ben & Jerry's. The second was a list of 592 cracker brands from a dieting website. For each list, I counted the total number of front vowels (i, ɪ, ɛ,e,æ) and the total number of back vowels (details of the study are here). The result, shown in the table to the right, is that ice creams names indeed have more back vowels and cracker names have more front vowels.
Here are some examples of stressed back vowels in ice cream names:
Rocky Road, Jamoca Almond Fudge, Chocolate, Caramel, Cookie Dough, Coconut
And here are samples of the many cracker names with front vowels; note the extraordinary number of ɪ vowels:
Cheese Nips, Cheez It, Wheat Thins, Pretzel thins, Ritz, Krispy, Triscuit, Thin Crisps, Cheese Crisps, Chicken in a Biskit, Snack sticks, Toasted chips, Ritz bits
Of course there are exceptions: vanilla, the orange blossom of our day, has an ɪ. But most of the front vowels in ice cream flavors tend to be the names of small thin ingredients in the ice cream: (thin mint, chip, peanut brittle).
So what's going on? Why are front vowels associated with small, thin, light things, and back vowels with big, solid, heavy things?
The most widely accepted theory, called the Frequency Code, suggests that low frequencies (low pitch) and high frequencies (high pitch) are associated with particular meanings. The frequency code was developed by linguist John Ohala (my old phonetics professor!), extending work by Eugene Morton of the Smithsonian.
Morton noticed that mammals and birds tend to use low-frequency (deeper) sounds when they are aggressive or hostile, but use higher-freqeuncy (higher-pitched) sounds when frightened, appeasing, or friendly. Since larger animals naturally make deeper sounds (the roar of lions) and smaller animals naturally make high-pitched sounds (the tweet of birds), Morton's idea is that animals try to appear larger when they are competing or aggressive, but try to appear smaller and less threatening when they are trying to be friendly or appeasing.
Morton and Ohala thus suggest that humans instinctively associate the pitch of sounds with size. It turns out that front vowels like ɪ and i are higher-pitched in a particular way than back vowels ɑ and o. All vowels are composed of different frequency resonances. When the tongue is high and in the front of the mouth, it creates a small cavity in front of the tongue. Small cavities cause higher-pitched resonances (the smaller the space for vibration, the shorter the wavelength, hence the higher the frequency). One particular resonance (called the second formant) is much higher for front vowels and lower for back vowels.
Thus the frequency code suggests that front vowels are associated with small, thin, things, and back vowels with big heavy things because front vowels have higher pitched resonances, and we instinctively associate higher pitch with smaller things.
This link of high pitch with deference or friendliness may also explain the origin of the smile, which is similarly associated with appeasing or friendly behavior. The way we make a smile is by retracting the corners of the mouth. Animals like monkeys also retract the corners of their mouths to express submission (Ohala's figure (a) on the right), and use the opposite facial expression, which Ohala calls the "o-face" in which the corners of the mouth are drawn forward with the lips possibly protruding (figure (b) on the right), to indicate aggression. Retracting the corners of the mouth shrinks the size of the front cavity in the mouth, just like the vowels ɪ or i. In fact, the similarity in mouth position between smiling and the vowel i explains why we say "cheese" when we take pictures; i is the smiling vowel.
Ohala's theory is thus that smiling evolved when mammals were in competitive situations, as a way to make the voice sound more high-pitched, so as it make the smiler appear smaller and less aggressive, and hence friendlier.
Of course even if Ohala is right about the ancient evolutionary origin of the smile, smiling in humans has evolved into a means of expressing many shades of enjoyment and other emotional meanings, just as back vowels have become part of a rich and beautiful system for expressing complex meanings by combining sounds into words.
Something similarly beautiful was created as saltpeter and snow, sherbet and salt, were passed along and extended from the Chinese to the Arabs to the Mughals to the Neapolitans, to create the sweet lusciousness of ice cream. And it's a nice thought that saltpeter, applied originally to war, became the key hundreds of years later to inventing something that makes us all smile on a hot summer day.
Ice cream, anyone?