The Whale Song Translation: A Voyage of Discovery To Neptune and Beyond (8 page)

“Looks like wavy lines to me,” said Greg.

“Is it a stream?” asked Dmitri.

“You’re both right,” replied Melanie. “It represents ‘water.’”

“By the way,” said Dmitri, “is Chinese script another of your many talents?”

“I only know a few characters. There are so many. This one-to-one correspondence between each ideogram and the thing or idea it represents requires thousands of different symbols and demands a steep learning curve to master each and every image. One virtue, however, is that written language is influenced by personal artistic interpretation. Think about the art of calligraphy.”

“You’re right,” said Dmitri. “In Japan, I was struck by the artistry of the Kanji characters on the signs and billboards.”

“I wouldn’t be surprised if the bonsai sculpting of trees and bushes is also rooted in the pictorial forms of their written language,” said Greg.

A subtle smile played at the corners of Melanie’s lips. “Your insights never disappoint.”

“Let me guess,” said Greg. “The word gram shapes in the Speakeasy program are the audio equivalent to the pictorial symbols used in written language.”

“Like the Chinese ideograms,” added Dmitri.

Melanie flicked her arm, as if banging out a rim shot. “Precisely. When we learn the written language, we memorize the patterns of the ideograms with the visual memory region of the brain. For spoken language, we store the shapes of the frequencies of individual words using the auditory memory region.”

Dmitri snapped his fingers and thumbs. “So we literally maintain separate dictionaries of both visual and spoken symbols in our mental database. While we’re reading, we instantaneously perform a search-and-match procedure for each word.”

Melanie nodded. “We’re on the same page. Likewise, when listening to a speaker’s words, our brains first draw the frequency shape of the word gram in the auditory processing center.”

“And then we scan the memorized database of target shapes for a match,” added Greg.

“And this all happens, word for word, in real time,” she said. “It’s amazing.”

Dmitri smiled and stroked his chin. “I’m still puzzled by the exact relationship between these frequencies and the shapes of the word grams.”

Melanie looked up at the wall clock. “How much time do you have?”

“Whatever it takes.” Dmitri shifted closer to her.

“Let’s review the rules for constructing the word grams,” she replied. “It’ll help if I introduce you to the representational form of written language, based on an alphabet of written symbols, such as the letters in the English alphabet.”

As an alternative to pictorial ideograms, she explained how each word can be represented by a unique string of primitive, alphabetic characters. “Similarly, each word in a spoken language is represented by a unique string of primitive units, called phonemes. Phonemes are the vowels and consonant sounds of spoken language.”

“Slow down,” said Greg. “Phonemes are the building blocks of the spoken language. So, there are layers of symbols. Words are the big thought symbols composed of primitive written alphabetic or spoken phonemic symbols.”

Melanie nodded. “Let’s get back to your original question, about the relationship between the frequencies and the symbols. Here, look at this.”

She swiped the marker pen vertically down the center of the whiteboard, made a sharp right turn, and continued horizontally. The finished figure resembled the two legs of a right triangle bereft of its hypotenuse.

“On this x,y plot,” she marked the horizontal line with an ‘x’ and the vertical line with a ‘y,’ “the x and y axes represent the range of frequencies for a pair of tones. Each x,y point on the plot represents a unique pair of tones.” She peppered the plot with a sequence of points that reminded Dmitri of a constellation of stars. “I’m plotting the points representing the first two resonant frequency tones for each of the ten primary English language vowels. The x-axis represents the range of possible frequencies of the lowest frequency tone and the y-axis is the frequency of the second tone. We can generate the different combinations of tone-pair frequencies by changing the shape of our vocal tracts.”

“Wait, something just occurred to me.” Dmitri stepped up to her and extended his arm. “May I?” Their arms lightly brushed when she handed him the marker pen. “In your plot, the points representing the tone pairs for the ten English vowels are arranged in a triangular pattern and equidistant from one other.” He drew a triangle whose perimeter circumscribed the outer boundary of the points.

A jumble of lines appeared on Melanie’s forehead. “Uh-huh?”

“It’s a modem constellation.” Dmitri sounded triumphant. “The vowels are communication symbols, similar to the optimal arrangement of symbols used by engineers in the design of communications devices. There’s a fundamental principle of communications theory called the Shannon theorem. A noisy communication link, like a phone wire or the air between two people talking, has a limitation on the maximum number of frequency-dependent symbols that can be transmitted without errors. As the noise level increases, you’d have to reduce the number of phoneme symbols to prevent the error rate from increasing.”

“And since conversations need to be error-free over a wide range of background noise levels, this explains why the English language evolved to have no more than ten primary vowels.” Melanie grinned. “That’s so cool. I never realized the evolution of language is shaped by a universal principle of communications engineering.”

“Now you know why I love my job.” Dmitri suddenly realized that the Shannon Theorem also applied to the effect of the Navy’s sonar blasts on the whales’ ability to communicate.

“Me too.” She surprised Dmitri with a playful high five. “Let’s revisit the relationship between the frequencies and the word grams. When we say a word, we generate a continuous series of tone pairs which, when plotted in x,y frequency space, look like a picture or shape similar to an ideogram in a written language. Every different spoken word therefore has a different pictorial representation.”

“The word gram shapes,” said Greg.

“Yes.”

“As with a child’s coloring book,” said Dmitri, “I can sketch a shape by connecting the series of dots along the path of each shape.” He connected Melanie’s dots into the shape of a cloud.

Melanie grabbed her own marker pen and started to scribble. “In the case of spoken language, the guiding dots are the tone-pair frequencies of the phonemes, the vowels and consonants. They are the primary frequency targets we need to connect in order to correctly pronounce words. The vowel tone pairs correspond to the points I’ve already plotted in the middle of the graph.” She started to scribble around the edges of the graph. “Now I’m adding the points corresponding to the low, medium, and high-frequency consonants around the borders. This is the plot that finally answers your earlier question, Dmitri. It’s the underlying principle for learning speech. The mastery of the phonemes is the basis for learning a language and the basis of the Speakeasy system.”

“It’s ingenious,” said Dmitri. “The Speakeasy word grams also provide the visual insight as to how our brains recognize the words of language. Every spoken word corresponds to a particular shape, or word gram, resulting from connecting the dots of the different phonemes in frequency space.”

“The language region in our brains processes the continuous sequence of phoneme frequencies,” said Greg, “searching for a match to the sequences of phoneme frequencies corresponding to each word we’ve stored in our learned database.”

“By George, you’ve got it!” Melanie laughed. “I tell my students to imagine they’ve entered a magical soundscape where they can exchange their thoughts by shape writing and shape matching.”

“Now you’re the poetic one,” said Dmitri.

“When you mentioned shape writing, I thought of something cool,” said Greg. “Speakeasy’s shape-matching concept is similar to the Swype texting app on my Android phone. It’s also based on a shape-matching algorithm literally called shape writing. Instead of typing in the text, you can drag your finger over the touch screen keypad, from letter to letter, to spell each word. The shape of the path your finger traces is compared, word for word, to a dictionary of stored shapes.”

“Spoken language is shape writing by sound,” said Dmitri, “and the keyboard is the map of phonemes stored in the frequency processing region of our brains.”

“And it’s all based upon a universal principle of communications engineering called the Shannon Theorem.” Melanie nodded rhythmically as she spoke.

“Better be careful,” said Dmitri. “You’re talking like an engineer.”

“Resistance is futile.” Greg’s voice sounded robotic. “Welcome to the Borg.”

As the trio shared a laugh, Melanie led the men back to Javier’s workstation where Erika still tutored the boy. “Thanks for helping me out, Erika. One more favor. Please set up the Speakeasy to display the word ‘sky.’”

“Sure enough.” Erika sat down next to Javier, sliding the keyboard over to her side of the desk. Her garden tattoos fluttered in an imaginary breeze when she started typing a series of commands. Greg appeared spellbound.

Melanie faced the boy, her fingers dancing. “Javier, let’s do another demo for our guests. Let’s pronounce the word ‘sky.’”

When Javier nodded, the bright red target word gram for the word “sky” appeared on the Speakeasy display. To Dmitri, its shape resembled a puffy crimson cloud.

“‘Sky’ is a sequence of four phonemes.” Melanie placed a finger upon the display. “The word gram for ‘sky’ is formed by connecting the tone-pair points for the frictional consonant ‘s’ followed by the plosive consonant ‘k’ followed by the vowel ‘ah’ followed by the vowel ‘ee.’” She traced her finger from point to point. “In addition to hitting the four target points, Javier must also stay on the correct frequency path that connects the points. Otherwise, it’ll sound like a mispronunciation. Did you realize a foreign accent is a consistent mispronunciation, a bias induced when the word gram targets of the native language are overlaid on top of the phonemes of the learned language?”

“
Oui, cheri
.” Dmitri delivered his brutish Bronx accent with a smirk.

Melanie grimaced. Her “Oy vey!” sounded as if she’d pinched her nose shut. “Go ahead, Javier.”

When Melanie double-tapped Javier’s shoulder, he grasped the edge of the desk, took a deep breath, stiffened his spine, and mumbled a first attempt. Just as Dmitri expected, the plot of the boy’s word gram was a far cry from the target shape. Nonetheless, a new idea, as yet elusive, lurked at the periphery of his awareness.

Javier’s next attempt sounded much better. During the span of three more iterations, as the boy’s pronunciation improved and the shapes traced onto the display converged to the correct version, Dmitri’s amorphous vision was coming into focus. He became aware of a powerful force emanating from the shapes on the screen. On Javier’s final attempt, the monitor erupted with colorful videogame-like special effects. At the same moment the boy cried, “Yeah!” and thrust his arms into the air, the welling realization breached the barriers of Dmitri’s consciousness.

As the symbols of the boy’s thoughts emerged from the display, Dmitri felt himself directly experience the creative process of the human mind. The medium that conveyed the symbols of language—sight, sound, touch—suddenly seemed immaterial. He felt liberated to perceive the process in its elemental form, freed from its bonds to matter and energy. This, he realized, was a new gateway for connecting with the minds of other sound-based creatures, human or otherwise.

“Are you still with us?”

Melanie’s voice roused Dmitri. He blinked three times. “Sorry. I just had a fantastic revelation about the Speakeasy.”

“Yes, it’s a remarkable breakthrough. It simulates how the auditory region of our brains interprets each word as a unique shape in frequency space, a word gram.”

“Just like each word in the Chinese language is written as a unique pattern or ideogram on the printed page,” remarked Greg. “And don’t forget, the computer plotting the word grams has to do hundreds and thousands of computations per second to analyze the continuous stream of sound for the correct frequencies.”

Dmitri smiled at the boy. “Javier’s brain also performs umpteen calculations per second in order to trigger the vocal tract adjustments to hit the phonemic target points in frequency space.”

Melanie tousled the boy’s hair. “You see, Javier, your brain is more powerful than any computer. Good work.” Her hands spoke the same words. “Go ahead and practice your verbs for a while. I’ll be back to test you in twenty minutes.”

Dmitri was struck by Melanie’s rapport with her student. She showered him with affection, and the boy’s face was luminescent as he gazed up. She turned back toward Dmitri. “It’s a miracle,” she said. “The Speakeasy’s like a voice-activated videogame. The children can’t wait for their next lesson.”

“In your own words, Melanie,” said Dmitri, “science and technology help the disadvantaged bridge the worlds of silence and sound.”

“I think I mentioned some of those things, but it’s nice of you to express them poetically.”

Dmitri felt a rush and decided to go for broke. “After yesterday’s whale watch, Greg and I met with the Director of PICES. I volunteered to analyze a recording of humpback whale songs. Your fantastic demo with Javier has given me a new inspiration. I’m intrigued by the idea of using Speakeasy to search the songs for phonemic patterns. How about it?”