Cognitive Daily

Blogging on Peer-Reviewed ResearchChildren follow a consistent pattern when they acquire language. Instead of learning the most common words first, they start by learning a disproportionate number of nouns. In the youngest talkers nouns form up to 60 percent of their vocabulary, compared to just 40 percent of the vocabulary of a typical 2 and a half year-old (who now knows over 600 words).

This pattern applies in many different languages, even Mandarin and Korean, where verbs appear in more prominent positions in sentences. The phenomenon is so universal that it has led some theorists to speculate that acquisition of non-noun forms is simply beyond the cognitive ability of infants — they have to “grow into” using verbs. But there’s another possible explanation: it could be that anyone learning a language is going to learn nouns first, simply because that’s the easiest way to learn language. In this view, the early dominance of nouns has nothing to do with general cognitive ability and everything to do with the process of language learning.

One way to distinguish between the two explanations could be to look at second-language learners. An older child — or even an adult — learning a language clearly has the cognitive ability to understand all types of words. But people learning second languages usually learn them differently from infants: they receive formal instruction; they memorize lists of words. Babies are never “taught” language — they acquire it naturally by watching others speak.

A team led by Jesse Snedeker realized that one population might be able to cast some light on the issue: Children adopted internationally. More than 20,000 kids are adopted in foreign countries and brought to the U.S. each year. If they’re adopted before reaching school age, they learn language in the same way as infants: by watching others speak, rather than formal instruction. Snedeker’s team located 27 recently adopted children between two and a half to five and a half years old and tested their language abilities every three months until they had been in the U.S. for 18 months.

Parents recorded themselves playing with the kids for an hour, and checked off words their kids could produce from a list of 680. Their results were compared to a population of non-adopted American children. This graph shows the number of words they had learned after specified durations in the U.S.:


The adopted children learned words quickly, with many children who’d been in the U.S. for just 12 months possessing the vocabulary of a 30-month-old. Clearly these older children learn language faster than infants — but what about the proportion of nouns in their vocabulary? Take a look at this chart:


As you can see, the proportion of nouns in the adopted children’s vocabulary declines in nearly the identical pattern to the infant controls. Both infants and older children acquire nouns first, followed by other word types.

The researchers also analyzed sentence complexity based on the recordings of parents playing with their kids. Here are those results:


Both the infants and the older kids produced more complex sentences as their vocabulary increased. Again, the pattern of the results was nearly identical — the only difference between language acquisition for infants and older kids was the rate of learning: older kids learn faster.

So the reason that infants learn nouns faster than verbs and other parts of speech isn’t necessarily that they aren’t generally smart or mature enough: older children follow precisely the same pattern of language development.

Snedeker, J., Geren, J., Shafto, C. (2007). Starting Over: International Adoption as a Natural Experiment in Language Development. Psychological Science, 18(1), 79-87.


  1. #1 RM
    November 7, 2007

    Makes sense. If you are learning a language “organically” (as opposed from flashcard/two-column old language-new language lists, it’s much easier to attach words to objects than words to actions/descriptions. – Parent points to cup, says “Cup”. Child picks up cup, says “Cup”, gets positive feedback. Try that with “flip”. Possible, but less convenient.

    On those lines, I’d also front the hypothesis that, within the nouns, tangible nouns (cup, blanket, mom) would be picked up first, and intangible ones (happiness, anger) would be picked up later.

  2. #2 Daryl McCullough
    November 7, 2007

    Nouns are much more useful in communication than other parts of speech. It’s often the case that if you indicate the noun, the verb is usually obvious. If a mother says “Coat!” it means “Put on your coat” if you are leaving without one, and it means “Put away your coat” if you just dropped it on the floor. If a child says “Cookie!” it means to give him a cookie (or perhaps, “There is a monster behind you that is shaped like a giant, evil Oreo”).

  3. #3 Tony Jeremiah
    November 7, 2007

    Re: So the reason that infants learn nouns faster than verbs and other parts of speech isn’t necessarily that they aren’t generally smart or mature enough: older children follow precisely the same pattern of language development.


    Piaget would probably agree with this general argument and that of the paper, primarily as it concerns the transition from concrete to abstract thinking within the individual. However, I’d bet Vygotsky’s sociocultural perspective of cognitive development would offer a slightly different point of view. Namely that language development takes place in a sociocultural context, and thus aspects of language should be influenced by one’s linguistic environment.

    Stated less abstractly, children learn speech primarily by mimicking adult models. As far as I know, adult English speakers emphasize nouns when communicating with infants; while Chinese, Japanese, and Korean-speaking children acquire verbs faster because nouns tend to be omitted from adult sentences while verbs are stressed (Kim, McGregor, & Thompson, 2000; Tardif, Gelman, & Xu, 1999).

  4. #4 Dave Munger
    November 7, 2007


    This is not an area of research I’m extremely familiar with, but according to the study authors, Korean and Mandarin speakers do not learn verbs faster. They cite Gentner & Boroditsky (2001), which is a review chapter appearing in Language acquisition and conceptual development.

  5. #5 Tony Jeremiah
    November 7, 2007

    Hi Dave,

    I’m not really an expert in this area either. However, I have a general knowledge of developmental psychology as I’ve been teaching it for quite a few years. Consequently, those two references came to my mind. Admittedly, they are (also) references are from a secondary source (textbook I use). So either the textbook author did not summarize those articles correctly, or, those studies actually contradict the one here, which is not terribly uncommon for psych research or science in general.

    Personally, I think many of these studies appear contradictory, but in fact, are really just different sides of the same coin. In this instance, Here it’s basically a variation on the old nature (Piaget) vs. nurture (Vygotsky) argument concerning this topic.

    The two references mentioned above are:

    Kim, M., McGregor, K.K., & Thompson, C.K. (2000). Early lexical development in English and Korean-speaking children: Language-general and language specific patterns. Journal of Child Language, 27, 225-254.

    Tardif,T., Gelman, S.A., & Xu, F. (1999). Putting the “noun bias” in context: A comparison of English and Mandarin. Child Development, 70, 620-635.

    I have the actual psych science article you present here, and don’t see these articles in their reference section, which is an interesting observation.

  6. #6 Ed Yong
    November 7, 2007

    Is there any data on the proportion of abstract vs. concrete nouns in a child’s repertoire versus an adult’s?

  7. #7 Amanda Owen
    November 7, 2007

    Whether Korean and Mandarin speaking children learn nouns or verbs faster is currently disputed. Some people have found evidence from spontaneous language and vocabulary checklists that verbs are acquired earlier. More recently the earlier methods have been questioned and experimental data supports the opposing perspective. Study methods really seem to affect the directionality of the results.I think the question is actually still open.

    See also: Bornstein, Cote, Maital, Painter, Park, Pascual, PĂȘcheux, Ruel, Venuti, Vyt (2004). Cross-Linguistic Analysis of Vocabulary in Young Children: Spanish, Dutch, French, Hebrew, Italian, Korean, and American English. Child Development 75 (4), 1115-1139.

    Kauschke, Hae-Wook, Pae (2007). Similarities and variation in noun and verb acquisition: A crosslinguistic study of children learning German, Korean, and Turkish. Language and Cognitive Processes, Volume 22, Issue 7 pages 1 – 28.

    Perhaps a relevant question to follow up on the article (which I haven’t read) is what the children’s original language was and how closely it related to English.

  8. #8 BT
    November 7, 2007

    Evidence for vocabulary development in Mandarin/Korean primarily comes from transcripts of what children say in naturalistic observations (which tend to overrepresent the number of verbs) and from vocabulary checklists (which tend to overrepresent the number of nouns). So, I agree with Amanda that study methods affect the directionality of the results.

    In general, across cultures, learning nouns does appear to be faster and easier than learning other types of words (prepositions, verbs). However, studies have suggested that the extent to which nouns dominate is quite malleable across cultures, particularly in children’s earliest words.

    For example, it has been shown that across different methods, Mandarin-learning children tend to learn proportionately more verbs earlier than English-learning children. And that quite often a significant number of verbs already exists in Mandarin children’s first ten words. Correspondingly, as Tony mentioned, studies have shown differences on the level of language input of Mandarin/Korean caregivers (more verb-focused relative to their English counterparts).

    At the same time, RM’s point is well-taken, as even Mandarin-learning children’s early verbs tend to be words for motions and actions, which are more tangible perceptually.

    In any case, the Mandarin/Korean data has always been an interesting cross-linguistic difference that can enlighten how we understand even the basic process of word learning.

  9. #9 Miss Cellania
    November 7, 2007

    My second daughter came to us speaking Tamil at age 2 and a half. She learned most of her English from her big sister, who was only 3. From what I recall, her main communication for a year was “no”, “mine”, and “I do it myself.” Meaning, she was just like a native English~speaking toddler!

  10. #10 Tatil Sever
    November 8, 2007

    At least in English, the verb that applies to an action may sound quite differently based on present vs. past tense. Even when they are regular verbs, there all these suffixes. That may hinder learning those verbs. In English, nouns usually do not take suffixes for different settings. When they do, they are usually done to come up with a new noun that refers to some other object.

  11. #11 Peter Murray
    November 8, 2007

    @Tony Jeremiah:

    Based on their abstracts (I’ve included them below, as they’re publicly available online), the two articles you cite appear to present evidence only for the relatively higher proportion of verbs to nouns in early learners of Korean and Mandarin, respectively, as compared to early learners of English. Which is unsurprising if it is true (as Dave writes) that ‘verbs appear in more prominent positions in [Mandarin and Korean] sentences,’ and (as Kim 2000 notes) that ‘Korean-speaking caregivers presented more activity-oriented utterances, more verbs, and more salient cues to verbs than did English-speaking caregivers.’

    However, the articles do not appear to present any evidence that early learners of Mandarin or Korean acquire their respective language’s verbs faster than they learn its nouns. On the contrary, Kim et al. (2000) explicitly confirms (at least for Korean) the finding in Snedeker et al (2007) that, across languages, children tend to learn their respective languages’ nouns faster than they learn its verbs. And while it is not clear, from its abstract, how to read Tardif et al. (1999) on this score, a quick skim of the article (sorry; not publicly available, but I’ve quoted a relevant snippet below) reveals the same findings for Mandarin. So, it appears that these articles do not, in fact, contradict Snedeker et al. (2007), but rather confirm it.

    Perhaps more interesting is the support that both studies appear to give to Amanda Owen’s point about how study methods and circumstances impact study results (e.g., book-reading vs. playing with toys, how caregivers structure interactions with children, etc.).

    Kim et al (2000):
    ‘The present study examined the composition of the early productive vocabulary of eight Korean- and eight English-learning children and the morpho-syntactic, semantic, and pragmatic characteristics of their caregivers’ input in order to determine parallels between caregiver input and early lexical development. Vocabulary acquisition was followed using maternal diary and checklists for the Korean-learning children (from a mean age of 1;6 to 1;9) and for the English-learning children (from a mean age of 1;4 to 1;8). Results showed that both Korean- learning and English-learning children acquired significantly more nouns than verbs at the 50-word mark. However, Korean children learned significantly more verbs than did English-learning children. The relative ease with which Korean learners, as compared to English learners, acquired verbs parallels several differences in the linguistic and socio-pragmatic characteristics of the input addressed to them. Korean- speaking caregivers presented more activity-oriented utterances, more verbs, and more salient cues to verbs than did English-speaking caregivers. These data suggest that both general and language-specific factors shape the early lexicon.’

    Tardif et al (1999):
    ‘Recently, researchers have been debating whether children exhibit a universal “noun bias” when learning a first language. The present study compares the proportions of nouns and verbs in the early vocabularies of 24 English- and 24 Mandarin-speaking toddlers (M age = 20 months) and their mothers. Three different methods were used to measure the proportion of noun types, relative to verb types: controlled observations in three contexts (book reading, mechanical toy play, regular toy play), identical across languages; a vocabulary checklist (MacArthur Communicative Development Inventory); and mothers’ reporting of their children’s “first words.” Across all measures, Mandarin-speaking children were found to have relatively fewer nouns and more verbs than English-speaking children. However, context itself played an important role in the proportions of nouns found in children’s vocabularies, such that, regardless of the language spoken, children’s vocabularies appeared dominated by nouns when they were engaged in book reading, but not when they were playing with toys. Mothers’ speech to children showed the same language differences (relatively more verbs in Mandarin), although both Mandarin- and English-speaking mothers produced relatively more verbs than their children. In sum, whether or not language-learning toddlers demonstrate a “noun bias” depends on a variety of factors, including the methods by which their vocabularies are sampled and the contexts in which observations occur.’

    Tardif et al (1999):630-631
    ‘[both child and adult] Mandarin speakers used a relatively higher proportion of verb types in their productive speech than did [child and adult] English speakers. However, consistent with the claims made by Gentner (1982) and others, children [whether speakers of Mandarin or of English] used a higher proportion of nouns than adults.’

  12. #12 Tony Jeremiah
    November 9, 2007

    @ Peter et al.

    Great discussion.

    Another issue might be the distinction between production and comprehension vocabulary. The cited articles seem focused primarily on production vocabulary. It may be important to also investigate developmental trends in comprehension vocabulary (i.e.,how many words children can understand rather than say), given that the implicit assumption underlying vocabulary acquisition is that it is cognitive, rather than the current behavioral view (i.e., speech production) of the noun bias. It could be that for some unknown cognitive reason, nouns are easier to say than non-nouns, but, sucn non-nouns already exist within the child’s linguistic repertoire.

    If it is true that some sociolinguistic contexts emphasize verbs relative to others, and, that comprehension vocabulary develops before production vocabulary (Bornstein & Lamb, 1992), it might be important to determine if there are developmental differences in comprehension vocabulary (e.g., for English vs. Mandarin children) particularly concerning verbs.

    I’m not familiar with the specifics of the various methodological approaches. However, examining comprehension vocabulary would most likely also require a different methodological approach than those indicated here. As an example, I wonder what the vocab data would look like for nouns and verbs among English and Mandarin speakers, if instead of focusing on what children say, the test focuses on their ability to recognize nouns and verbs. This could involve a methodology such as pointing to pictures that represent nouns and verbs.

  13. #13 Amanda Owen
    November 9, 2007

    Most of the studies use a vocabulary checklist called the McCarthur-Bates Communication Development Inventory (M-B CDI or MCDI). It asks parents to report words their children understand from ages 8 months -16 months and words their children undertand or produce from 12? – 30 months. It has been re-developed and normed in a variety of languages other than English and is a large source of data used to report child vocabulary skills. It has been externally validated via diary studies and experimental data. Anyway, most of the studies (with the exception of Kim et al) are reporting both.