The Language Game
How improvisation created language and changed the world
reviewED BY Stefan Huyghe
Often, the language sciences view linguistic communication as a monologue: As if, when we are talking, we simply churn out an idea that resides in our head in the form of a signal transferred through speech and simply unpacked by the listener.
The Language Game (not to be confused with the book of the same name by Ewandro Magalhães), however, looks at the evolution of language through a new prism and concludes it to be fundamentally interactive and collaborative. The ingenious book by cognitive scientists Morten H. Christiansen and Nick Chater argues that meaning comes from language use itself. It sees language development as very similar to a game of charades. Both are all about improvisation and active collaboration as people try to get ideas across. When we communicate, we are improvising to provide cues to help each other understand what we’re trying to say.
The authors argue that, during a conversation, the listener does a huge amount of work to process the speech of their conversation partner. It’s not just the speaker producing a signal — unlike machines that passively wait to receive a message, humans use our knowledge of the world, everything we know about each other, and what was said earlier in the conversation to make sense of the information that is transmitted to us.
Language is a form of improvised communication, a theory in sharp contrast with the more general view of looking at language as a fixed code that allows us to bottle our thoughts into a stream of words to be uncorked and decoded by the listener. Language is a game of charades in which the clues are words and sentences, created, understood and sometimes not.
The captivating book opens with the arrival of Captain James Cook and his crew on Tierra del Fuego in 1769. When they met with a group of Haush hunter-gatherers on the beach, two of Cook’s sailors advanced. Soon, two Indigenous people reciprocated, throwing away the small sticks they were carrying. Cook’s men interpreted this as a signal of peaceful intentions. They were right, and soon the two groups were exchanging gifts and sharing food. Without any knowledge of each other’s language or customs, they were able to communicate through a high-stakes game of cross-cultural charades.
Language is perhaps humanity’s most astonishing capacity — one that remains poorly understood. How do noises convey meaning? Where do the complex layers of linguistic patterns come from? Why do children learn language so easily, whereas chimpanzees can fail to learn it at all? Those are the fascinating questions addressed by the scientists in this exceptional read.
The authors posit that we’ve been getting many of the answers to these questions wrong, simply because scientists have been looking at language incorrectly. They call into question that humans possess an innate language ability that is hard-wired into our brain encoded with grammatical rules. Language isn’t about rules at all. As the Tierra del Fuego story illustrates, it is about the desire to be understood by any means possible. Rather than ruled by constraints, language is fueled by improvisation and imagination.
Construction grammar in child language acquisition is a central theme in the book. Children use small chunks of language which have both a sound and a meaning. They can be individual words or morphemes — small parts of words which lend meaning and structure — or they can be attempts to put together nouns and verbs in the form of sentences and phrases. Each construction is a little package that helps connect sound and meaning. Language itself is made up of a large number of these, and they interact in complicated ways. The theory contrasts radically with Noam Chomsky’s theory that grammar has a deep abstract structure of great mathematical complexity with variations in different languages. The construction grammar approach instead follows the charade pattern: You build up a language by learning the pieces. As you continue to build more pieces, you interlock and interact more and create more pieces. In this way, language develops gradually.
Children seem to learn these little patterns one by one — they do not suddenly master a highly abstract grammatical rule, but rather, experiment with these chunks of language. Kids seem to learn the elements of language bit by bit. Second language acquisition seems to follow a similar pattern.
Through experiments, we now know that kids are excellent learners. There is a lot of information in what we say to kids that they can potentially use for learning language without having any built-in linguistic constraints such as in Universal Grammar.
Perhaps this is the most controversial finding of the authors’ research: Breaking away from the traditional Chomsky-Pinker view of how language works, Christiansen and Chater lay claim to the notion that new language is produced without actually having any Universal Grammar.
Christiansen built a computational model that takes small chunks of language and uses them for comprehension and consequent language production. He was able to model the acquisition of language by training the machine to pick up on the chunks or constructions across 29 Old World languages. Similarly, other researchers have even developed models like GPT-3, which are trained upon billions of words and can write in a passably human tone and style, providing further evidence that we can produce a silicon learner. This AI is even producing poetry at Cornell.
Looking at adult language use, we can account for at least 50% of almost everything we say, just by considering small chunks that are reused. These are little formulaic sequences that we use again and again. While we may say something that is new and fresh every once in a while, much of our everyday word usage seems to be recycled.
Despite all things that seem to be unique from language to language, we seem to have roughly the same thoughts across cultures. Languages are just one way of conveying information. Notations for mathematics are another example. If you didn’t have a decimal notation, doing math or arithmetic could prove to be a lot tougher. It’s just like finding the right words to express a thought. It’s not the language that fixes the boundaries. Thoughts, therefore, don’t seem to be the primary purpose of language.
So why do languages produce certain patterns and not others? The authors suggest it has to do with the constraints of our human body and the way our brains function. The patterns or absence of them are derived from a suite of other more general skills that are not necessarily dedicated to language.
We know writing systems have lots of constraints as well. You could use all kinds of crazy ways to design writing that could take on any form to convey information. Nevertheless, we tend to use nice natural strokes, and quite often alphabets. The claim that this is the product of some innate writing instinct falls flat, however, considering the fact that writing’s only been around for a few thousand years or so.
Morten Christiansen believes that the challenge of language acquisition is practical, rather than theoretical. Often viewed in the past as a problem of inference in which we try to piece together the abstract grammar from incomplete input, people acquire language by practicing across a myriad of social interactions and playing linguistic charades. He concludes that it might best be construed as skill learning, in a very similar way to learning other complex human skills such as riding a bicycle or playing a musical instrument.
The Language Game is a wonderful book that aims to make the academic accessible to the general reader. It explains how language works, answers why we have it, why animals communicate through different systems, and many other questions. It introduces the concept of language as charades, and the authors hope it will inspire people to pursue a career within the language sciences. If you are interested in languages, you will love reading The Language Game.
Subscribe to stay updated between magazine issues.