What Is a Word?

A definition is needed to understand the nature and the origins of language.

Posted August 13, 2021 |  Reviewed by Tyler Woods


  • Linguists have yet to define words satisfactorily.
  • That ambiguity has led to the erroneous conclusion that animals can learn words.
  • A satisfactory definition distinguishes between utterances that manipulate behavior and those that transmit knowledge.

Although it is generally agreed that language differentiates humans from other animals, there’s less agreement as to why. Many scholars, notably Noam Chomsky, have argued that it’s grammar that makes humans unique: the ability to create an unlimited number of meanings by combining a finite number of words. I agree, but I would add that, because grammar could not exist without words, words are also uniquely human.


Noam Chomsky Source: Wikipedia

Curiously, there’s less agreement about the status of words. In a widely cited article, Chomsky and colleagues defined two “faculties of language”: a broad faculty that includes, among other abilities, words and concepts, and a narrow faculty that only includes grammar (Hauser et al., 2002). In that framework, they concluded that the narrow faculty is uniquely human.

Can animals learn words?

As evidence that animals use words, they cited research on the communicative abilities of chimpanzees, monkeys, dolphins, and birds. Other articles claim that monkeys use words when they warn other monkeys about the presence of different types of predators and that parrots can imitate human words. Remarkably, none of those studies defined words.

Regarding chimpanzees, Chomsky, among others, cited the ability of chimpanzees to produce words. The words in question weren’t spoken because of the limitations of an ape’s articulatory apparatus. Apes can’t produce the sounds that comprise vocal languages. Instead, they were trained to use gestures that were similar to the signs of American Sign Language, a language used by hundreds of thousands of deaf people. Chimpanzees were also trained to touch “lexigrams,” symbols composed of different colors and geometric patterns, each having its own meaning.

Elsewhere, I have argued that the sole function of an ape’s use of symbols is to obtain particular rewards. In each instance, the ape’s performance amounted to nothing more than well-trained motor acts that were used in a ritualized manner (Terrace, 1985). I have also noted other differences between the utterances of apes and human infants. The utterances of human infants are spontaneous and bi-directional. Ape utterances are neither. Most important is an ape’s inability to name objects (Terrace, 2019).

These differences are illustrated by the example of teaching an infant or an ape to utter the word “dog.” Suppose the infant looks at a dog and utters “dog.” Her parent might reply, “Nice dog,” “Big dog,” or “No, that’s a cat.”

An ape is shown a dog, or a picture of a dog. In order to obtain food or drink, the ape signs “dog,” or touches the lexigram for dog. For the ape, the sight of a dog is simply a cue for making a response to obtain a physical reward. By contrast, the human infant names the dog, for which her parent responds socially, often verbally, but not typically with a physical reward.article continues after advertisement

In this example, the ape didn’t transmit information to the listener. Indeed, there was no listener. Pushing a button on a vending machine would be just as effective. Such responses are called imperatives, uni-directional utterances that are motivated by physical rewards.

Imperatives are not limited to apes. Human infants also utter imperatives, for example, “up,” “milk,” etc. But if imperatives were the only type of utterance an infant could make, she could not be considered to have learned language. That’s because imperatives constitute a minuscule fraction of human vocabulary. The vast majority of their utterances are what linguists refer to as declaratives, all of which are conversational. A speaker comments about some feature of the environment and the listener responds. The listener either agrees with the speaker, elaborates his utterance, or questions it.

What are we to make of the observation that a chimpanzee will learn to sign “banana” to request one? Or of a chimpanzee who touches a particular lexigram in anticipation of a reward? Do those signs or lexigrams qualify as words?

To answer that question, we need to define words. Dictionary definitions are only moderately helpful for assessing a chimpanzee’s utterances. That’s because dictionaries describe a human’s use of words. For example, one dictionary defines a word as, “a single distinct meaningful element of speech.” Meaningful is defined as, “the nonlinguistic cultural correlate, reference…of a linguistic form.”

Up to a point, that definition helps by citing a nonlinguistic reference, such as a person, object, or event. But it doesn’t specify the direction of the utterance in question. It could be a declarative, part of a conversation in which there is an exchange of information between a speaker and a listener, or it could be uni-directional as, for example, a request for a reward.

A definition of a word

For these reasons, I define a word as an arbitrary symbol that is used conversationally, that is, declaratively. This definition excludes cries, whimpers, and other unlearned vocalizations that infants make. It also excludes imperatives, and is consistent with declarative pointing, a universal antecedent of words.

Declarative pointing, which is often interpreted as a non-verbal response, has been observed in all cultures. Pointing occurs a month or two before an infant begins to refer to objects verbally. No non-human primate has ever been observed to point in their natural environment. Although there are reports of pointing by apes that have been raised by humans, such pointing is imperative, entirely in the service of obtaining a physical reward.article continues after advertisement

Shortly after an infant begins to point to objects, she begins to name them. When, for example, she points to a dog, her caretaker responds “dog” and the infant imitates that utterance vocally. The advantage of the utterance “dog” over pointing is that it is a more precise form of reference. If, for example, the dog was next to a tree, and an infant pointed in that direction, we wouldn’t know if she pointed to the dog or to the tree.

Because the number of rewards one can request is small, there is an upper limit to the number of imperatives. In all of the thousands of extant human languages, however, the number of declarative words is unlimited. It is always possible to conceive of a new word to name a particular object, action, or attribute. It is that feature that allowed our ancestors to refer to objects that were not immediately present, to past and future events, and to imaginary objects. In short, the transition from animal communication to declarative words marked the beginning of verbal culture.

Language is considered to be one of the nine major transitions of evolution (Maynard Smith & Szathmáry, 1995). Without disagreeing with that claim, I would argue that language, as we know it, required two major transitions. Why two? Because language is too complex to have been selected from animal communication in one step. It is difficult enough to explain the transition of animal communication to words without the complexity of grammar.

The first transition was a shift from fixed, uni-directional emotional signals to bi-directional words. The second was the addition of grammatical rules for arranging words. It was that transition, the ability to create an indefinitely large number of meanings from a finite set of words, that Chomsky argued was the essence of language. It should be clear, however, that the second transition could not have occurred without words.

Defining a word as an arbitrary symbol that is used conversationally clarifies the first transition. That transition required a qualitatively new form of communication, one that supplemented the small repertoire of emotional signals that animals used for millions of years. Instead of only being able to manipulate another’s behavior, recent ancestors began to inform another about mutually important events. This could only be done with declarative utterances.


Hauser, M., Chomsky, N., & Fitch, W. T. (2002). The faculty of language: What is it, who has it, and how did it evolve? Science, 298, 1569-1579.

Szathmáry, E., & Smith, J. M. (1995). The major evolutionary transitions. Nature, 374(6519), 227–232.

Terrace, H. (1985). In the beginning was the name. American Psychologist, 40, 1011-1028.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s