Elsevier

Cognition

Volume 127, Issue 3, June 2013, Pages 391-397
Cognition

The acquisition of abstract words by young infants

https://doi.org/10.1016/j.cognition.2013.02.011Get rights and content

Abstract

Young infants’ learning of words for abstract concepts like ‘all gone’ and ‘eat,’ in contrast to their learning of more concrete words like ‘apple’ and ‘shoe,’ may follow a relatively protracted developmental course. We examined whether infants know such abstract words. Parents named one of two events shown in side-by-side videos while their 6–16-month-old infants (n = 98) watched. On average, infants successfully looked at the named video by 10 months, but not earlier, and infants’ looking at the named referent increased robustly at around 14 months. Six-month-olds already understand concrete words in this task (Bergelson & Swingley, 2012). A video-corpus analysis of unscripted mother-infant interaction showed that mothers used the tested abstract words less often in the presence of their referent events than they used concrete words in the presence of their referent objects. We suggest that referential uncertainty in abstract words’ teaching conditions may explain the later acquisition of abstract than concrete words, and we discuss the possible role of changes in social-cognitive abilities over the 6–14 month period.

Highlights

► Infants 10 mo. & older show understanding of abstract words like “uh-oh” and “eat”. ► Six to nine-month-olds do not, in contrast to their performance with concrete nouns. ► At 14 months, infants show more robust abstract word-meaning knowledge. ► Corpus analyses show abstract words are used less reliably with their referents.

Introduction

To learn their native language, children must learn words. And to learn words, children must identify words in speech, and grasp what others mean when they talk. The predominant hypothesis about the course of language learning has long been that development proceeds first with speech signal analysis, and only later with discovery of word meaning. This perspective is motivated by demonstrations of precocious phonetic learning between 6 and 10 months (e.g., Jusczyk and Hohne, 1997, Kuhl et al., 1992, Polka and Werker, 1994), subsequent advances in social cognition (e.g., Carpenter, Nagell, Tomasello, Butterworth, & Moore, 1998), and finally the onset of referential communication at about 11 months, when infants first produce meaningful speech and gesture (e.g., Bates et al., 1975, Camaioni et al., 2004). According to this view, the typical 10-month-old knows the auditory forms of dozens of words, but has yet to invest them with meaning (e.g., Jusczyk, 1997, Swingley, 2005), perhaps pending a better understanding of humans as intentional agents.

The notion that development in social cognition is a prerequisite for learning words follows from the premise that the typical conditions under which infants encounter words are insufficient for infants’ making the connection between the words and their denotations using perceptual association mechanisms alone (Gleitman & Gleitman, 1992). If a parent says “I’ll go get a spoon” in the absence of a spoon, this “teaching trial” is misleading for the simple associative learner who perceives “spoon” and some spoonless applesauce, but is potentially helpful to the intention-reading child who tracks the parent’s goals until he returns with the spoon. Not all researchers agree about this premise, however, maintaining that whatever social cognitive skills infants may or may not have, words and their referents co-occur with sufficient reliability to be learnable by infants using domain-general cognitive capacities for perceptual association. Thus, there is debate about whether intention-reading skills are necessary for young children’s learning of all words (Tomasello, 2001, Waxman and Gelman, 2009), perhaps just “hard,” more abstract words (Gleitman, Cassidy, Nappa, Papafragou, & Trueswell, 2005), or no early words at all (Colunga & Smith, 2005).

One empirical approach to characterizing the mechanisms of early word learning is to test lexical knowledge in children who have only very rudimentary social cognitive skills. Indeed, young infants’ early intention-reading and joint attention skills are limited. For example, at 6–7 months, infants can follow a person’s gaze to an object, but do not appear to understand that gaze implies object-directed interests or goals (Woodward, 2003). Such infants do not yet engage in true “triadic” interactions where they knowingly share attention to an object with another person (Carpenter & Liebal, 2011); the ability to appreciate gaze as both social and goal-directed does not appear until around 9–10 months (Beier & Spelke, 2012). On the other hand, more basic goal attribution and belief computation has been shown in social cognition research around 6–7 months (Csibra, 2008, Kovacs et al., 2010).

Despite 6–7-month-olds’ apparent lack of sophistication in recognizing others’ intentions, two recent studies have shown that 6–7 month olds know some object word meanings, including words referring to body parts, e.g. hand, and foods, e.g. banana (Bergelson and Swingley, 2012, Tincoff and Jusczyk, 2012). Word understanding at this age implies either that rich intention-reading skills are not necessary for learning all words, or that such skills have been underestimated in 6-month-olds.1

Given the theoretical possibility that these object words may have been learned by infants using generic mechanisms of perceptual association and categorization, here we examined more abstract words, such as “eat,” “wet,” and “hi,” whose referents in the child’s experience are, visually speaking, more diverse from instance to instance. Moreover, while concrete words are often used in the presence of the objects they refer to (Gogate, Bahrick, & Watson, 2000), abstract words such as action verbs have denotations that are often transient by nature, and instances of such words may not be as closely linked in time to their referents (Tomasello & Kruger, 1992). Learning such words may thus be more challenging for younger infants.

We tested children ranging from 6 to 16 months. This served three goals. First, if 6–9 month olds fail with abstract words, it may indicate that learning concrete and abstract words requires different skills with different developmental courses, or that the learning conditions for abstract words are less favorable. In contrast, if infants succeed, it would suggest that even perceptually diverse categories can be learned and linked to words by children without fully-developed intention-reading skills. Second, if children start to succeed between 10 and 12 months, it will suggest that learning abstract words, unlike concrete words, emerges in parallel with important advances in social cognition (though it would not show that this developmental link is a causal one). Third, if word-understanding performance improves significantly at around 14 months, as Bergelson and Swingley (2012) found for more concrete words, it will provide further evidence for a change in language-relevant cognitive or social abilities in children, including perhaps a deeper understanding of joint attention (Carpenter & Call, in press), a better grip on the conventional nature of words (Buresh & Woodward, 2007), or improvement in appreciating the nature of the experimental task.

To better interpret developmental features of our word-understanding results, we also conducted a series of video-corpus analyses of the contexts in which parents use the concrete and abstract words tested in our studies. Coders annotated a range of interactional features in all instances of these words in 20 recording sessions from the Providence corpus (Demuth, Culbertson, & Alter, 2006). These data were supplemented with analyses of word frequencies in the Brent and Siskind (2001) corpus.

Section snippets

Participants

Three age groups were tested: 34 6–9 month-olds (M = 8.37 mo., R = 6.24–9.79 mo., 19 girls); 46 10–13 month-olds (M = 11.96 mo., R = 10.02–13.99 mo., 26 girls); 18 14–16 month-olds (M = 14.99 mo., R = 14.03–16.52 mo., 11 girls). 48 Infants were excluded due to fussiness (39), technical problems (3), failure to meet language or health criteria (2), or parental influence (4). Infants were recruited from the Philadelphia area by mail, email, phone, and in person. All were healthy, carried full-term, heard >75% English at

Corpus analyses

We examined mothers’ use of the words tested here as well as the words tested in Bergelson and Swingley (2012) in both the Brent Corpus (an audio corpus of 16 mothers interacting with their 9–15 month old infants), and in 20 videos of the Providence Video Corpus (5 mothers interacting with their young children; we selected a subset in which children ranged from 11 to 18 months). In the Brent Corpus we compared frequency counts in isolation (i.e., in one-word utterances) and overall. In the

Results from eyetracking study

To measure whether infants fixated the named event more upon hearing it named, we computed a difference in fixation proportions: how much infants looked at one video when it was the target, minus their proportion of looking to it when it was the distracter. This computation, which corrects for bias due to preferences for one video over the other (Bergelson & Swingley, 2012), yields one score for each item-pair. For instance, with the pair kiss–dance an infant’s performance was given as how much

Discussion

These findings enrich our understanding of the early stages of language acquisition, showing that by 10–13 months, but not earlier, infants linked several common abstract words to their referents. This in turn suggests that the word-learning mechanisms and social/cognitive abilities that are needed to learn abstract words under ordinary daily-life conditions are in place approximately half a year earlier than previous laboratory tests had indicated (Golinkoff, Hirsh-Pasek, Cauley, & Gordon, 1987

Conclusions

The current findings contribute to the literature on language acquisition in several ways. We showed that infants as young as 10 months old identify novel referents of common words that do not refer to concrete objects, but younger infants do not. Thus, the acquisition of abstract and concrete words differs ontogenetically, and may require skills with differing developmental trajectories. The beginnings of abstract word learning, but not concrete word learning, appear to occur in parallel with

Acknowledgements

This work was funded by an NSF Graduate Research Fellowship to EB, the NSF-IGERT Program, and NIH Grant R01-HD049681 to DS.

References (37)

  • Bloom, L. (1993). The transition from infancy to language: Acquiring the power of expression. Cambridge [England]; New...
  • R. Brooks et al.

    The development of gaze following and its relation to language

    Developmental Science

    (2005)
  • L. Camaioni et al.

    The role of declarative pointing in developing a theory of mind

    Infancy

    (2004)
  • Carpenter, M., & Call, J. (in press). How joint is the joint attention of apes and human infants? In H. S. Terrace & J....
  • M. Carpenter et al.

    Joint attention, communication, and knowing together in infancy

  • M. Carpenter et al.

    Social cognition, joint attention, and communicative competence from 9 to 15 months of age

    Monographs of the Society for Research in Child Development

    (1998)
  • E. Colunga et al.

    From the lexicon to expectations about kinds: A role for associative learning

    Psychological Review

    (2005)
  • K. Demuth et al.

    Word-minimality, epenthesis and coda licensing in the early acquisition of English

    Language and Speech

    (2006)
  • Cited by (0)

    View full text