Children think before they speak

A linguistic contrast between English and Korean provides a telling test of different ideas about whether thought precedes the acquisition of language, or whether certain concepts are language-specific.

Snug fit: in situations such as this, the Korean language makes a distinction between tight and loose contact. Credit: S. OLIVER/DK IMAGES

In his autobiography, written in the fourth century AD, Saint Augustine1 described how he learned to talk: “By constantly hearing words, as they occurred in various sentences, I collected gradually for what they stood, and having broken in my mouth to these signs, I thereby gave utterance to my will.” For Augustine, thought precedes language: language is a tool with which to express one's ideas and to understand the ideas of others. This is the view of many contemporary philosophers and psychologists2,3,4, but it is not the only possibility. Many scholars would instead endorse the theory of linguistic relativity, and maintain that learning a language has a profound influence on a child's mental life. If so, then speakers of different languages might think in very different ways5,6.

On page 453 of this issue, Hespos and Spelke7 present data, from 5-month-old babies, that support Saint Augustine's view. They concentrate on a much-studied linguistic contrast. Korean, but not English, makes a distinction between ‘tight-fitting contact' and ‘loose-fitting contact'. For instance, Korean uses different verbs when describing placing a shoe in a large box, where it fits loosely, and when placing the shoe in a small box, where it is a tight fit — even young children who are just beginning to learn Korean honour this distinction when they speak6. Hespos and Spelke ask whether this distinction between two sorts of contact is universal, and exists before language-learning (in which case it should be present in babies), or whether it is the result of acquiring Korean (in which case it should be present only in children and adults who have some knowledge of that language).

They address this question by using a standard method in infant cognition. They show babies instances of a given category until they get bored (or habituated) and stop looking, and then see if the babies perk up — look for longer — at an instance from a new category. If so, it means that babies are sensitive to the categorical difference. Using this method, Hespos and Spelke find that 5-month-olds who are raised in an English-speaking community are sensitive to the Korean categories of meaning. If the babies are habituated to tight-fitting events, such as a cylinder placed within a narrow container or a ring-like object placed around a post, they will look for longer when later shown a loose-fitting event, such as a cylinder placed into a wide container (see Figs 1 and 2 of the paper, pages 453 and 454). The converse is also true. If habituated to loose-fitting events, babies will look for longer when shown a tight-fitting event. In this domain at least, the traditional view is right: thought precedes language.

Hespos and Spelke note the analogy here with phonology, in which there are also cross-linguistic differences — certain acoustic contrasts are present in some languages but not others. It is not that children become increasingly sensitive to the distinctions made in the language that they are exposed to. Instead, they start off sensitive to every distinction that human languages make; the process of learning a particular language involves becoming insensitive to those distinctions that are irrelevant, and learning what to ignore.

Phonology and meaning differ in certain important ways, however. Another striking fact about phonological development is that the early sensitivity disappears. If the child's language does not exploit a distinction, then the child loses the ability to notice it. This is one reason why it is so difficult to learn a second language. But, as Hespos and Spelke point out, even an adult English-speaker who has never heard Korean can tell the difference between a tight fit and a loose fit. This difference between phonology and meaning makes sense. Phonology is for communicating; once a language is learned, nothing is lost by jettisoning those phonological contrasts that are irrelevant. But meaningful contrasts such as loose fit and tight fit are for making sense of the world. This is nicely demonstrated by the finding that 5-month-olds can use their sensitivity in a non-linguistic context, when predicting the motions of objects.

In addition, although all phonological distinctions made by language may be innate, this cannot be true for all distinctions of meaning. Babies might understand the contrast between tight fit and loose fit, and between support and containment, but they are unlikely to comprehend the contrasting meanings of the verbs ‘leering’ and ‘glaring’, or the nouns ‘accountant’ and ‘lawyer’. This must be learned.

What is the nature of this learning? One compromise view is that there is a universal core of meaningful distinctions that all humans share, but other distinctions of meaning that people make are shaped by the forces of language; this is consistent with the theory of linguistic relativity. But it is also possible that the strong Augustinian view is correct: language learning might really be the act of learning to express ideas that already exist, either because they are unlearned (as is likely to be true of the domain studied by Hespos and Spelke) or because they have been learned though experience with the physical and social world.

The question of how language and thought are related is one of the deepest in psychology, and there are many variants of the claim of linguistic relativity that this current research does not address8,9,10. But the capacities of 5-month-olds do pose a serious challenge for certain strong versions of the view that language precedes thought, and show that, in some domains at least, children think before they speak.


  1. 1

    Saint Augustine The Confessions of Saint Augustine (Random House, New York, 1961).

    Google Scholar 

  2. 2

    Fodor, J. A. The Language of Thought (Crowell, New York, 1975).

    Google Scholar 

  3. 3

    Bloom, P. How Children Learn the Meanings of Words (MIT Press, Cambridge, Massachusetts, 2000).

    Google Scholar 

  4. 4

    Gleitman, L. R. Language Acquisition 1, 3–55 (1990).

    Article  Google Scholar 

  5. 5

    Whorf, B. L. Language, Thought, and Reality (MIT Press, Cambridge, Massachusetts, 1956).

    Google Scholar 

  6. 6

    Choi, S. & Bowerman, M. Cognition 41, 83–121 (1991).

    CAS  Article  Google Scholar 

  7. 7

    Hespos, S. J. & Spelke, E. S. Nature 430, 453–456 (2004).

    ADS  CAS  Article  Google Scholar 

  8. 8

    Xu, F. Acta Psychol. 102, 113–136 (1999).

    CAS  Article  Google Scholar 

  9. 9

    Tager-Flusberg, H. & Sullivan, K. Cognition 76, 59–89 (2000).

    CAS  Article  Google Scholar 

  10. 10

    Bloom, P. & Keil, F. Mind Lang. 16, 351–364 (2001).

    Article  Google Scholar 

Download references

Author information



Rights and permissions

Reprints and Permissions

About this article

Cite this article

Bloom, P. Children think before they speak. Nature 430, 410–411 (2004).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing