Look at this sequence and guess what the next number in it should be:

Most people would guess that it is ‘6,’ assuming the formula in (2a). That is a very reasonable hypothesis, but the fact of the matter is, the next number in the sequence could also be ‘126,’ if you use the formula in (2b).

Without the 6th element in the sequence, there is no way to tell which formula is correct. Even if we have the 6th element, though, this problem doesn’t go away. We could easily modify the formula in (2b) so that the 6th element is ‘6’ but the 7th element — or any other term — diverges from the formula in (2b).

This illustrates the fact that an infinite rule-governed system is unlearnable. When a system is infinite, we can never be sure that the next element in the system will follow the pattern we think we have found.

The problem of infinite systems is relevant to language because language is productive, allowing us to combine the pieces of language in unlimited ways. This means that language is also an infinite system. Anytime you make a sentence, you can take that sentence and transform it into another sentence by adding “I believe that” or “Javier says that”  or “Amita thinks that”, etc…, to the beginning, as shown in (3).

There is no limit to the amount of times you can do this, in principle, although you may eventually run out of breath or forget where you are in the sentence. But those are physical limitations. The grammar itself doesn’t have any limits. In addition to that, using embedded clauses like we did in (3) is only one way we could add to sentences infinitely. This kind of infinite building is in every part of the grammar, including in possessives (4) or with prepositional phrases (5).^[1]

This leaves us with a problem. Language is infinite, but infinite systems are unlearnable. Somehow, though, most people seem to learn language. This is known as the poverty of the stimulus argument since the language we receive in our environment (the stimulus) is insufficient for us to acquire language. In other words, it is impoverished.

Noam Chomsky’s solution to this problem is the idea that some parts of language are innate. That is, some components of language are built into human DNA. In fact, he argued that there are three components to the human capacity for language: general cognitive capacities, experience, and a language-specific genetic component. 

• General cognitive capacity. There are many parts of human cognition that can help us learn language but aren’t specific to language. These are general cognitive capacities that we already have for other skills, but that we can also use to help us learn language. They include the ability to analyze data and spot patterns, developmental constraints, and principles of efficient computation.
• Experience. We know that experience plays a role in language acquisition. If you are surrounded by Japanese speakers as a child, you will learn Japanese. But if you are surrounded by Swahili speakers, you will learn Swahili. The language in your environment is called the input. You will learn a different language—both different grammatical rules and different vocabulary—depending on what language you’re exposed to. This is where all the differences between languages come from.
• Language-specific genetic component. This genetic component is a capacity for language that we are born with, but unlike general cognitive capacities, it is specialized for language. Chomsky named it Universal Grammar, which is often abbreviated as UG. Universal Grammar should be uniform for all humans, since we have approximately the same genetics. It allows us to take our input and analyze it by instinct as an infant. The timing of language acquisition might also be constrained by genetics.

Besides the poverty of the stimulus argument, there are a few other pieces of evidence for Universal Grammar:

• Even children who are not exposed to a rich language environment tend to develop language of some kind. For example, deaf children in a hearing family who do not otherwise have access to a signed language community tend to develop a homesign system.
• There is evidence from acquisition that children need to be exposed to language within a certain timeline known as the Critical Period in order to be successful at acquiring language.
• There is a dissociation between language development and cognitive development. Some people (those with Specific Language Impairment) show deficient language skills but normal cognitive development, while others (those with Williams Syndrome) seem to show delayed cognitive development but normal language skills.
• Specific Language Impairment has even been traced to a particular gene called FOXP2!
• Certain parts of the brain are specialized for language use, such as Broca’s and Wernicke’s Areas.

In summary, according to the poverty of the stimulus argument, language is infinite, and although infinite systems should be unlearnable, we still manage to learn language. The solution to this is Universal Grammar, which is the hypothesis that there is an innate, genetically-endowed cognitive capacity that is specific to language.

On the other hand, the theory of Universal Grammar and the poverty of the stimulus argument are not accepted by all linguists. However, it is a foundational assumption in the framework of Generative Grammar, which is the approach taken in the textbook. In the next section, we will learn a bit more about the debate between those who adopt UG and who don’t.

Check yourself!

References and further resources

Comedy and satirical linguistics

🎉 Wug Life. 2 Oct 2017. This Meowth can speak human language! Tumblr. https://wuglife.tumblr.com/post/165979527843/hahaha-oh-man-this-is-beautiful-a-quick

Academic sources

🔍 Boxell, Oliver. 2016. The place of Universal Grammar in the study of language and mind: A Response to Dąbrowska (2015). Open Linguistics 2: 352–372.

🔍 Dąbrowska, Ewa. 2015. What exactly is Universal Grammar, and has anyone seen it? Frontiers in Psychology 6: 852.