[Lightly revised text from Chapter 2 of my book The Tarot: Methods, Mastery, and More.]
Divination is probably as old as humankind, but it has been little recorded and even less studied. We can only speculate about how it was practiced by our ancestors, or how well it worked. But we can, perhaps, better understand the process of divining--our attempt to understand the shape and meaning of events--if we turn to other primal phenomena: language, myth, and play.
“When we contemplate our origins,” writes Richard Leakey, “we quickly come to focus on language.” World-famous for his fossil discoveries, Leakey gave great consideration to how we became who we are, and concluded that although there are many objective standards for our uniqueness as a species (such as bipedality and brain size),
in many ways it is language that makes us feel human. . . . Our thoughts, our world of imagination, our communication, our richly fashioned culture--all are woven on the loom of language.
Upon this loom, Leakey contends, was fabricated a uniquely human mental model of the world, by means of which multiple channels of sensory data are processed to meet complex practical and social challenges. 
According to one influential view (built on the work of MIT linguist Noam Chomsky), language has a “deep structure” that can be seen in the overall similarity of structural and generative rules underlying all human languages. This accounts for why children learn languages in much the same way the world over, and why all languages seem to develop and diverge in comparable patterns.
The presence in language of a deep structure seems to reflect a “language-acquisition facility” in the brain, which in turn explains how we can do anything so fabulously difficult as learning language in the first place. Physicist Heinz Pagels described how this deep structure can be glimpsed in the work of translators, such as this remarkable Russian who knows dozens of languages, Oriental as well as Western:
After he listens to someone speak, he translates the remarks into whatever language is desired--any one of dozens. How does he do it? According to him, he ‘hears’ the remarks not in any language at all, but rather as ‘a matrix of meanings’--a conceptual format of some kind that he creates. When asked to translate into a specific language, he consults the matrix and expresses that meaning into a language.
Pagels theorized that the spoken language heard by this translator was subordinate to a deeper, nonverbal logical structure that is independent of any specific language. 
The idea of an underlying structure of language--”a matrix of meanings” from which can be drawn any one of a number of “translations”--illuminates the process of divination in general, and Tarot in particular. The oldest, most basic divinatory methods all focused on the interpretation of patterns found in objects or events: the entrails of animals, smoke rising from a fire, or cloud designs in the sky.
These practices were based on the impression (fanciful or not) that natural phenomena constitute a kind of language, which expresses the deep structure of reality. Indeed, an archaic and unconscious link between language and divination is apparent in the very way we talk about divination: “I read the cards,” or “the tea leaves said . . . .”
If, indeed, there is a built-in facility of the brain that acquires/creates spoken language, there may be a variant or companion faculty that perceives natural phenomena in a similar way--a faculty that helps us discover in our surroundings certain clues to the patterns of fortune.
Tarot, however, is not a naturally occurring phenomenon. It belongs to a second category of divinatory techniques, those which are purposely created in order to provide a language we can read more easily and reliably. 
We can’t be certain when humans first developed divinatory systems, but like other practices, divination seems to have become more systematized as cultures became more complex. We do know that China had the basic form of the I Ching nearly four thousand years ago, and numerology and astrology certainly date back thousands of years. In the first millennium BC, divinatory practices were a busy business around the Mediterranean, where they continued to thrive for centuries under Christianity and Islam, in spite of orthodox denouncements.
By contrast, Tarot is the only widely used divinatory system more-or-less invented in modern times. Though the images cards have been around for about five hundred years, no esoteric and mantic properties were defined at all until the late 18th century. So perhaps it’s not surprising that Tarot is more similar than most divinatory systems to spoken language—in the sense that individual elements (the cards) have specific meanings and can be arranged according to grammar-like rules.
It’s obvious, of course, that Tarot is more amorphous than a conventional language, because there are no fixed references for its individual elements; the images can refer to virtually anything and they can be related to each other however we choose at a particular time. A Page is a child in one context, a message in another; the Emperor may be a father, a boss, an Aries man, a politician, a social structure, the force of convention, or any of several other things, depending on the cards which surround it.
But spread patterns can control and direct possible meanings, and combinations of meanings, by providing grammar, syntax, and punctuation. For example, spreads enable the reader to tell (by position) roughly what’s in the past tense, what’s in the future, what subject acts on what object, and so on.
A spread also forms “sentences” and “paragraphs” in which individual meanings can be understood by contextual reference. So if the rest of the cards in a spread have nothing to do with children or romance, the Page is more likely to be a message than a child; if, on the other hand, the Page appears with the Empress and the King of Cups, it is more likely to be a child than a message.
Although that’s a very simplified example, it’s a clear illustration of the same process that enables us to recognize one meaning for “spring” in the sentence “Tomorrow is the first day of spring,” and another meaning in the sentence “He has a spring in his step this morning.”
In many ways, Tarot “language” is actually most like the language of dream imagery, from the standpoint that (a) information is conveyed more by images than by words, and (b) the meaning of individual images is largely a function of their relationship to other images in the set. For example—if I say “I ate fish” you know exactly what the fish means; no matter whether it is a salmon or a trout, a big fish or a little one, its meaning is that it was food for me.
But if I say “I dreamed of fish,” you have no idea (and I don’t, necessarily) what the fish means. Now it becomes very important whether the fish was a salmon or a trout, big or little, swimming or sizzling on the grill. The characteristics of the particular fish and the context in which it appeared are the only clues to its significance.
Dream and language, according to French philosopher Gaston Bachelard, are brought together in reverie, the “waking dream” in which the imagination travels into the heart of reflection.  And the experience of poetic reverie is described by Bachelard in a way that applies just as well to reading Tarot images as to reading written language:
I am a dreamer of words, of written words. I think I am reading; a word stops me. I leave the page. The syllables of the word begin to move around. Stressed accents begin to invert. The word abandons its meaning like an overload which is too heavy and prevents dreaming. Then words take on other meanings as if they had the right to be young. And the words wander away, looking in the nooks and crannies of vocabulary for new company . . . .
In just such a way may the diviner find that a kind of reverie overtakes his or her imagination, and transforms a Tarot event from the prosaic kind of reading to the poetic kind.
But as marvelous as reading may be, with its capability not only to propagate information but to inspire reverie, it is dependent on language itself. Language is essentially oral-aural in character: we make and hear sounds that convey meaning. Writing greatly extends the power of language, because it fixes “sounds” so that the same words can be “heard” in the same order over and over, by many people.
But a good deal of the potential complexity of communication is sacrificed in the process, because there are no visual cues (gestures, expressions, physical references, and so on) to embellish meaning, and there is no feedback; what is written down does not change in response to our opinions, or expand in answer to our questions.
In ordinary life, reading and speaking are rather distantly related; we read to ourselves mostly, and if we read to others, we usually read word-for-word what is written. Tarot is unusual in the way it brings together a kind of written language and our spoken language. A Tarot reading involves the process of real-time translation from visual symbols to vocal narrative, so images and speech are drawn closely together; it’s as if we were reading a poem or story aloud but instead of following the words on the page, we produced free associations inspired by the words.
In this process, we can open and share consciousness in an extraordinary way.
Terrence McKenna, an important interpreter of shamanic technology, told of Amazonians who, with the help of plants that contain DMT (a type of pseudo-neurotransmitter, actually perceive sounds visually. McKenna speculated that as consciousness evolves, we may all be moving from “a language that is heard to a language that is seen, through a shift in interior processing.” 
The visible language he predicts is not—like written language—fixed in time and immune to feedback, but rather a speech apprehended equally by our ears and eyes.
McKenna’s idea is intriguing. How much more profoundly, how creatively could we communicate if our experience of language were taken in through more of our senses? Our senses, after all, offer multiple channels of input and processing in the brain, which suggests we might reach far more meaningful levels of understanding.
But whether or not such a remarkable leap of consciousness is around the corner for our species, some of us may at least catch a glimpse of this complex human potential through the visible language of Tarot.
 Leakey’s consideration of language and the development of consciousness is found in Origins Reconsidered, co-authored by Roger Lewin. This book, which revises many of the observations and conclusions contained in his earlier best-seller Origins, is wonderfully clear and open.
 Pagels’s speculations can be found in The Dreams of Reason, a fascinating exploration of complexity and information science.
 A third category is concerned with inspiration or personal experience as a means of divination, such as practiced by oracles and shamans.
 Bachelard was a philosopher of science as well as a phenomenologist of poetic consciousness, who undertook an imaginal exploration of the elemental world in The Psychoanalysis of Fire, Water and Dreams, Air and Songs, and Earth and the Reveries of Repose. The quote below is taken from The Poetics of Reverie. If you like French literature, you will love Bachelard; if you don’t, you will find his work uphill going, but generally worth the climb.
 McKenna makes this remark in an interview contained in Mavericks of the Mind. An interesting and complementary observation concerning brain structure, gender, and language forms is offered by Sam Keen in an interview in Timeless Visions, Healing Voices:
For example, when English-speaking men have a stroke in the area of the brain that controls language, they become aphasic, they can no longer speak. Whereas women with the same kind of stroke don’t become aphasic. Interestingly, Navaho and Hopi men, like women, also do not become aphasic. Their language is much more pictorial than ours; it comes out of a different part of the brain and a different view of reality.