Before her AAAS Fellowship, before her PhD in linguistics, before her postdoctoral research at MIT, before she would spend her career keenly observing how children learn to talk and how adults suffer when losing the ability to speak and understand language, Lise Menn had the thrill of contradicting conventional wisdom and being right. It began when her son Danny starting making pronunciation errors in a way that accepted linguistic theory couldn't account for.
“I didn’t have any plans of working in child language at the time. I wanted to be a linguist, but I wanted to be someone who reconstructed vanished languages. I wanted the really romantic stuff. Then I started having this stuff happening in my kitchen,” remembers Menn, who is now Professor Emerita of Linguistics at the University of Colorado Boulder.
Menn has spent her career investigating the concealed neurological hustle and bustle that happens when we construct and utter meaningful sentences, and what happens when this complex process fails. A proponent of examining and recording how language is used in practice and not just in theory, Menn’s research has branched into two seemingly different but related fields: the study of how we first acquire language as children and the study of aphasia, the inability to form and understand language that can result from a brain injury, such as a stroke.
While Menn’s research has delved into how sentences get organized (syntax) as well as the process by which words mean anything at all (semantics), her scientific career began and has largely gravitated around the less popular field of phonology, the study of how intelligible words result when our lips, teeth, tongues, jaws and vocal cords all work together to create articulate speech. Her phonetic inquiries began when Danny first tried to speak.
The Scene: A Late 1960s Massachusetts Kitchen.
Menn, a former math major-turned-housewife-turned-budding linguist is watching her two young children. Two-year-old Danny is a late talker, and a strange talker. When Danny speaks monosyllabic words like “big” and “bug,” his vowels—nestled in the snug center of his utterances—are clear. His consonants, however, are muddled. “Big bug” comes out as “gig gug.” This gets Menn to thinking about the linguistics papers and textbooks her two friends, both currently at nearby Harvard University, gave her to read. As she listens and really tries to understand Danny, Menn has an insight: Danny is making a unique error, one that the linguistic research she’s been scrutinizing can’t explain. It hits her. She’s got a research-worthy topic. She decides to take a risk: She calls up a nearby professor, an expert on children’s language development. He tells her to write up what she hears and send it to him.
“I assumed he figured he’d never hear from me,” remembers Menn. “He told me to write it [the findings] up. So, I wrote it up and sent it to him. And eventually it became my first paper.”
Today, the types of phonetic errors made by Danny—who is now a fully articulate adult—are estimated to be made by about 10 percent of English-speaking children. But at the time Menn wrote her first paper, the field of childhood language acquisition, then dominated by men, was short on good observational data of how children actually learned to talk.
Drawing an analogy from the field of biology, the accepted wisdom of childhood language acquisition theorized that language in children formed in much the same way that an embryo grows in the womb. Assuming a uniform, progressive development of language from childhood to adulthood, many linguists believed that childhood language would not and could not display patterns that weren’t also exhibited in adult language in some form. Danny proved them wrong.
“You had this idea of a very smooth—to use a mathematics terms, monotonic—development [of language]—that would just become a better and better approximation of adult [language],” explains Menn. “But adult language doesn’t have anything like big bug for gig gug as a general pattern. Now had they known more embryology, they wouldn’t have made that error because people knew for example that mammalian embryos have gills.”
Rather than a smooth development, it is now understood from data collected by Menn and others that children do a great deal more problem-solving when they first attempt to talk than many linguists had previously thought. This is essentially what Danny was doing.
Faced with the head scratcher of how to get his vocal cords and everything attached to his lower jaw—his tongue, his lower lip, his lower teeth—to act in unison with what’s attached to his skull—his upper lip, upper teeth, and the roof of his mouth, Danny chose an elegant solution: simply use the same consonant sound at both the beginning and end of each word. Big bug became gig gug.
A decade after writing down her observations of Danny’s speech errors, Menn’s career would diverge from studying childhood language to the study of aphasia. In aphasia research, she would discover a similar story of inadequate theorizing based on limited data.
The Scene: Aphasia Research Center, Boston University School of Medicine
They are not happy. They have lost most of their friends. Their family members struggle to communicate with them. They are trapped inside their minds, their intelligence obscured behind garbled sentences and misunderstandings. Some were brilliant before the brain injury. But now forming and understanding even simple sentences is hard work, or impossible. They are constantly frustrated. They have lost that most human of abilities: language. They have aphasia. Menn, tape recorder in hand, has come to speak with and try to understand them and maybe learn something about how language works and how important it is to have language for feeling human.
“Most people with aphasia that I’ve met, and I’ve probably met 50 or 60 of them,” says Menn, “they are so isolated by their language disorder. They feel very keenly that people think they are stupid because they can’t talk…Their whole lives have been taken away from them.”
Yet, says Menn, there are rewards to working with aphasics in spite of the sadness. Her experience with aphasics would give Menn a deep compassion for those suffering from the disorder. She would become friends with many of the aphasics she worked with, including a woman named Shirley, who acquired aphasia after having a stroke at age 63. Together Menn and Shirley would write Shirley Says: Living with Aphasia, a first-person guide for people suffering from the disorder. Observing Shirley and others would also push Menn to again question her field’s accepted wisdom. To understand why requires a bit of history.
A decade before Menn started her linguistics career in her New England kitchen, linguist Noam Chomsky revolutionized the field by proposing a radical idea: humans possessed the ability to produce understandable language due to a kind of instinctual ability to organize meaningful sentences. This organizing principle became known as Universal Grammar.
Drawing on a philosophy of science concept called parsimony (aka. Occam’s razor), Chomsky submitted the idea that humans generate grammatical sentences almost effortlessly due to a single, economical mechanism at work in our minds. Particular languages were learned due to an innate Language Acquisition Device, which explained why children were able to acquire language from the supposedly poor examples of language usage provided to them by adults.
The ability to generate grammar also put what many thought would be the final nail in the coffin of a competing, behaviorist theory of language: the idea that we produce language as a chain of words that follow other words based on probabilities.
To dispute the notion that people were simply making probabilistic decisions when speaking, Chomsky offered the unlikely but grammatically understandable sentence: “Colorless green ideas sleep furiously.” The sentence while improbable—green isn’t likely to follow colorless in most utterances of English; and ideas isn’t likely to precede sleep—nonetheless makes a kind of sense because of how the words are organized.
“The idea was to have your theory really lean and mean,” says Menn. “It wasn’t a bad way to start. But that’s not how the brain works.” Or as Menn put it in one of her lectures, “Parsimony has no role in the design of brain networks.”
The brain is now believed to be something like a massively parallel processor, with neurons in constant cooperation and competition firing in multiple regions at once. While something like an organizing principle of grammar is undoubtedly at work, the new understanding of the brain suggests that the brain nonetheless employs statistical inferences when constructing sentences, inferences based on language samples taken from the environment.
For evidence, says Menn, consider the errors that aphasics make.
Scene: The Boston Cookie Theft image; a 1960s-era suburban Kitchen Gone Awry
While gazing out the window, a mom absent-mindedly dries a plate as her kitchen faucet fills her sink to overflowing, spilling water onto the floor. She is standing in a growing pool. Her two children, taking advantage of mom’s daze, are sneaking cookies from a cookie jar on a high cabinet shelf. One is a boy, perched on a stool. His right hand is in the jar. As he passes a cookie to his eager sister with his left hand, his stool tips. He is about to fall.
The above scene is from the Boston Cookie Theft image, a frequently used sketch of a 1960s kitchen gone wrong. The image is used to help elicit responses from linguistic research subjects.
Some individuals with aphasia have difficulty making sentences using syntactic connectors (think articles, pronouns, prepositions). When they cobble phrases together, they tend to leave out these little-but-important words, or they put them in the wrong place. Other aphasic speakers produce what sounds like normal grammar, but have difficulty finding nouns and verbs. Both forms are often studied by asking people with aphasia to describe a picture in which a number of things are happening at the same time. Hence the Boston Cookie Theft image.
Aphasic speakers who make syntactic errors can often find the words for the key elements in the chaotic kitchen—water, kids, mom, cookies. But they frequently have trouble ordering those elements in a meaningful way. Consider this example cited in Menn’s 2013 paper in WIREs Cognitive Science: “And the boy give to a cookie—the boy give to girl a cookie.” (The subject struggles with expressing who is giving the cookie to whom.)
Conversely aphasics who make semantic errors, or errors of meaning, can position most or all of the syntactic elements in order, but typically can’t find the nouns, verbs, and adjectives they need. Instead, in a result reminiscent of Lewis Carroll’s poem the “Jabberwocky,” they substitute made-up words for real ones. Consider this Boston Cookie Theft utterance cited by Menn: “and he’s falling down on the; And the water is all [rinked] out; Off the pipe; for the end; Clearing away; I’m getting all [malled]; And he’s getting all [retout]; and she skidded on the [dill].”
As Menn points out in her papers, what’s fascinating about these errors and the similar ones made by perfectly normal speakers is what they reveal about how our brains play with probability when constructing sentences.
Aphasics, researchers have discovered, are usually quicker to use and comprehend more frequently used nouns than less frequently used ones. This can be seen with plurals.
Words that tend to be found in quantities greater than one—eyes, hands, potatoes, leaves—are easier for aphasics to find in the mental lexicon then their less frequent singular forms—eye, hand, potato, leaf. The opposite is also true. Aphasics have a harder time with the plural forms of words that are more common as singulars.
Likewise, phrases made of words that are frequently found together—you’re welcome, ladies and gentlemen, have a seat—are easier for aphasics to understand. These phrases—‘collocations’ in linguistic lingo—can also be seen in errors in which common idioms and phrases are blended together. Blending errors, such as help all you want—a combination of the expressions help yourself and take all you want—are common in both aphasics and normal speakers, suggesting, says Menn, that a good deal of probabilistic problem-solving is involved in normal sentence construction.
“When we have people making mistakes that are blends, we have pretty good evidence that those expressions were battling about in that person’s head at the same time, and they were competing with one another and got blended,” says Menn.
What all this points to, says Menn, is that the decision-making process happening inside our minds when we form meaningful sentences is far more complex and less streamlined than early ideas about generative grammar led many to believe.
Toward the end of her career, Menn attempted to put her understanding of the complex phonological, syntactic and semantic processes underlying sentence construction into something like a unified model. She called her theory the “Mini-model Integrating Sequential, Categorical and Hierarchical Activation,” or MISCHA, model. She would abandon the model shortly after constructing it, concluding that it wasn’t complex enough. Yet Menn’s contribution to her field likely lies elsewhere.
In the textbook she wrote, Psycholinguistics: Introduction and Applications, in her time as Associate Editor of the journal Aphasiology, and in her own papers on aphasia and childhood language development, time and again, Menn, like many contemporary linguists, would stress and push for the publication and sharing of more and more observational data of actual language usage.
Self-deprecating and yet sanguine about her impact on her field, Menn is hopeful that the current generation of computer-savvy linguists will take her and others’ data and run with it, making their own far more complex and accurate models of the mind.
“To the extent that we are dealing with empirical data out there,” says Menn, “regardless of how removed theories right now are from being able to deal with those data, we’ll eventually converge on what’s going on.”