Which came first, grammatical rules or their exceptions? For decades, linguists bet on rules, but disorder and flux may turn out to be language’s most essential traits
Language is a curious mix of order and disorder. Anyone who has struggled to master English has confronted its baffling assembly of regular verb patterns (Kisha wanted), quirky sub-rules (Pablo rang) and wild exceptions (Barlow went). And English is no special case. All 7,000 of the world’s languages are characterized by elegant rules, quasi regularities and strange inconsistencies.
So which came first, the order or the disorder? Answering this question turns out to be crucial to understanding how language works.
For more than a half-century, the language sciences have proceeded from the theory that order came first. Noam Chomsky, the founder of modern linguistics, proposed in the 1980s that each child is born with a genetic blueprint (a “universal grammar”) that captures deep patterns common to all languages. Similarly, the psycholinguist Steven Pinker writes of an inborn “language instinct” that captures the rules of language.
The idea of a spontaneous order emerging from chaos may sound implausible, but in nature self-organization is ubiquitous, from snowflakes to flocks of birds.
An alternative possibility is gaining ground, however—one that holds that language starts from disorderly, ad hoc, communicative signals. Order—in the form of regular verb endings, grammar rules and much more—emerges slowly, spontaneously and always incompletely, over countless interactions and across many generations.
The idea of a spontaneous order emerging from chaos may sound implausible. But in nature it is ubiquitous, from the self-organization of snowflakes and flocks of birds to the hexagonal basalt columns of the Giant’s Causeway in Northern Ireland. The chemist and philosopher Michael Polanyi, and later the economist Friedrich Hayek, pointed to spontaneous order in human affairs by stressing that markets produce an orderly system of prices and production from the chaos of the marketplace.
Perhaps it is also the case that the order of language arises not from a hard-wired instinct within the genes and mind of each individual but from the cumulative result of social interactions among individuals. To see such a process in action, consider the remarkable phenomenon of “grammaticalization”: the curious process through which words with concrete, specific meanings transmute into components of grammar, such as pronouns, prepositions, conjunctions, verb endings and more.
Sometimes this process occurs through erosion, meaning the gradual truncation of syllables or even words: going to becomes gonna; did not becomes didn’t. Over longer periods of time, erosion can be far more dramatic. Starting with the Latin mea domina (my mistress), we progress through French ma dame or madame (for Mrs.), to English madam, ma’am, mum, and sometimes even just ’m (as in Yes’m).
In fact, the path from Old English (the language of Beowulf and the Arthurian legends), through Middle English (the language of Chaucer), to the English of today, is a story of erosion—of collapsed distinctions and lost endings. Old English nouns had a complex system of case markings (nominative, accusative, genitive, dative and instrumental) and three grammatical genders, which applied not just to nouns but to demonstratives and adjectives. All of that has since been rendered invisible.
But now we have a puzzle. If complex case markings and verb endings tend gradually to disappear, where do they come from in the first place? The answer is that they were formed by the fusion of once-independent words.
Consider the Latin construction cantare habeo: I have (something) to sing. By saying that you have something to sing, you suggest that singing will occur in the future, if it occurs at all. Over time, the meaning of constructions like this one broadened to apply to any future action, such as sleeping or laughing. But the independent verb habere remains, which creates a new way of talking about things that will happen in the future—that is, it creates a new future tense.
In the modern descendants of Latin, the forms of have are often tacked on to the infinitive form of a verb to give the future tense and, in some cases, have eroded. So, for example, to have in the first person is ai, ho and he, in French, Italian and Spanish, respectively, and the future tense of, for example, I will sing, is chanterai, cantarò, cantaré.