A MATTER OF MEANING: REFLECTIONS ON FORTY YEARS OF JCL

How do children learn words and their meanings? This topic has occupied considerable space in JCL since its beginning. My own publications in the journal have been mostly on this topic,Footnote ¹ concerned with what words children learn in the early stages – and variations thereof – and how children learn and use the meanings of words in comprehension and production. These topics have also been widely covered and discussed by many prominent researchers, in books (in temporal order) such as Macnamara (Reference Macnamara1982), Nelson (Reference Nelson1985), Clark (Reference Clark1993), MacWhinney (Reference MacWhinney1999), and P. Bloom (Reference Bloom2000).

There is more than enough material in these works for new scholars to contemplate as they consider possible research directions of their own. For the problems opened up in the 1970s and 1980s have not been entirely resolved by subsequent empirical investigations. In this contribution to the JCL ‘reflection project’, I note areas that deserve further theoretical and empirical work, based on my reading of this literature. These include the questions of ‘What is meaning?’ and ‘How does the child grasp the meaning of words?’ The variability of pathways to meaningful language that different children take, and the social coalition between adult and child toward understanding the world in linguistic ways, are also open. Not least is the problem of becoming a conventional user of meaning so as to acquire new knowledge and solve cognitive developmental problems.

WHAT IS MEANING?

Different authors make varied broad claims about meaning. For example, it is sometimes stated authoritatively that, contrary to our common usage, words do not have meanings, or at least not dictionary type meanings. One must consult philosophers like Putnam (Reference Putnam and Putnam1975) to unpack this claim, which actually boils down to this: meaning is a personal concept that guides the use of words in a particular context to refer to things and events in the world. On the other hand, words must also have common meanings that mediate between the speaker and the hearer in a conversation (for example). The personal concept of each participant might differ somewhat, but the common portion shared between users enables effective communication. Given this recognition that there is a common shared meaning in use, but variations in individual meaning systems, it is the common meaning that the child must somehow at least partially grasp in learning the word. For example, the child may have a very idiosyncratic notion of what ‘breakfast’ refers to, overlapping the common use of the term very little, but enough to serve the purpose of family use.

Most of the writers cited earlier here converge on the idea that word meanings are based on concepts and that communication takes place in words that link to mental concepts for speakers and listeners. So far so good. But that children's personal concepts or meanings may differ widely in composition from the adults' is an important point that is often overlooked or dismissed outright. The more inexperienced the child with the norms of the language community, the more variation from the ‘norm’ the meaning/concept is likely to be. This was an idea that I explored very early in my career (Nelson, Reference Nelson1974) based on my study of eighteen children acquiring their first words over the course of a year from age 0;11 or 1;0 to 2;6 (Nelson, Reference Nelson1973, Reference Nelson1974).

It is widely asserted (e.g. P. Bloom, Reference Bloom2000; Hirsh-Pasek, Golinkoff & Hollich, Reference Hirsh-Pasek, Golinkoff, Hollich, Hollich, Hirsh-Pasek, Golinkoff, Brand, Brown, Chung, Hennon and Rocroi2000) that the process of word learning is one of the child's ‘mapping’ words to concepts – often ‘fast mapping’, presumed necessary in order to learn the five to eight words per day between two and six years of age that are required to account for the total vocabulary achieved by school age. But this usage of ‘mapping’ tends to obscure the point that the child's conceptual base is bound to differ from the adult's. ‘Cognitive mapping’ implies a large space with relationships among points indicated in the map structure (like a state map with cities and roads delineated or a sky map indicating stars and the positions of planets). The child's ‘space’ is likely to be quite sparsely filled in comparison to the adult's, with much potentially relevant to the language to be learned quite empty. Moreover, the first publications on ‘fast mapping’ color terms (Carey, Reference Carey, Wanner and Gleitman1982) revealed that the acquisition of a new color term varied in its reference across children and over time for the same child, whereas ‘mapping’ implies specificity of the space mapped. Word learning must in many cases provide not only the new word but the content of its common concept, or where the word can be located in the yet unexplored territory. Closely examined, in fact, the mapping metaphor not only does not help, it harms by making word learning seem easy and flawless, by attaching words to familiar concepts, rather than challenging and enlightening, leading into new fields of exploration. As Tomasello (Reference Tomasello2002) pleaded, “Could we please lose the mapping metaphor, please?”

The problem of how to reconcile the child's paucity of relevant concepts with relative ease in learning new words has been avoided largely because most of the focus of the early word learning literature has been on names for things, or at least on words that can be referenced ostensively, by pointing at what is being named. This focus is reasonable as a starting point, and many children start there with good reason. They can relate familiar objects to the words used in context because they have become acquainted with the world of actions, people, and things through their own experience. But even everyday language takes the child beyond the range of his or her own perceptual/action learning. And it is this contribution of word learning that is most significant for the child's subsequent development in language, cognition, and social–cultural development. The potential of language for change in the child's knowledge systems and cognitive structures needs to be at the forefront of our investigations into the child's meaning systems.

The meaning of a word as grasped by the infant whose cognitive space has been previously defined only through experiential memory derived from perception and action in the world (much of it very close to the ground) cannot be relied on for many aspects of the adult's meaning. Even on the level of perceptual-action knowledge, the child's base must be expected to differ importantly from the adult's. Linda Smith's research, using mobile cameras recording the child's-eye-view has provided unanticipated evidence of the wide difference between the child's view of action and the adult's. Smith is convinced from this and other work that the child's initial word meanings are derived from perceptual information (Smith, Reference Smith2013). She has long claimed that it is unnecessary (maybe illegitimate) to propose concepts as a cognitive basis for word learning within the dynamic systems theory that her research employs. Instead, she emphasizes developmental process: “The developmental links between object name learning, visual object recognition, and pretend play highlight the cascade that is developmental process – that development consists of many interacting and mutual dependencies across systems that may seem at first unrelated” (p. 626).

Like other researchers cited here, my view from the beginning (e.g. Nelson, Reference Nelson1974) was that children began with concepts, but with an emphasis on the point that their concepts were different from those of adults'. Mandler (Reference Mandler2004) took a similar position in her investigations of children's conceptual category distinctions in play with animate and inanimate figures. She found that at age 0;10 to 1;0 infants made clear distinctions between the category of animate and inanimate, but not within these broad categories. Thus her work suggested that infants' conceptual categories are not well fitted to the ‘basic level’ object terms typically acquired in the first phase of infant word learning (Mandler & McDonough, Reference Mandler and McDonough2000). Is it then the case that children learn the basic level categories (e.g. of dog and cat) from language? There is at this point no agreement on the composition of a conceptual basis in infancy (e.g. Carey, Reference Carey2009), or even whether the concept of concepts is necessary to understand either the infant mind or older minds. But researchers continue to assume that infants are mapping words to internal concepts in word learning studies.

The now widely shared intuition that a dynamic systems approach to these issues will advance our understanding of both the acquisition of meaning and the variability among children in its achievement seems well grounded. Findings from the child's-eye-view of the prelanguage world have the potential for understanding the tendency during the early word learning period by some children to ‘over-extend’ the word learned to items outside the category. For example, the child's word ‘dog’ might be (and often is) generalized to all four-legged animals, indicating a larger class than the ‘basic level’ term allows. Many early trackers of the acquisition of words by single children commented on the use of a name to refer not just to the ‘basic level’ type (‘dog’, ‘car’, ‘duck’, ‘daddy’) but to an overextended and sometimes mysterious set of other objects or events. For example, Piaget (Reference Piaget1962) commented on the child's use of his word for ‘horse’ to other animals and to a number of other, less obviously related household items. Vygotsky (Reference Vygotsky and Kozulin1986) similarly cited a child's extension of the word for a ‘duck’ seen at a pond to items at home that appeared to have little in common with a duck or its appearance. More recent examples and analyses of both over- and underextension were documented by Anglin (Reference Anglin1977), Bowerman (Reference Bowerman, Wanner and Gleitman1982), and Rescorla (Reference Rescorla1980), among others. These and ‘nonce words’ (sounds that have no meanings in the adult language) produced by some children and used in specific contexts, but not always with meanings evident to others, are evidence in favor of the assumption that the conceptual space of beginning language users differs in important ways from that of adults. However, the contemporary view is that children in general do not learn ‘weird’ words and that evidence to the contrary is ill-founded (P. Bloom, Reference Bloom, Hall and Waxman2004).

In summary, ‘meaning’ is a term with many meanings and varieties (Millikan, Reference Millikan2006). For the purpose of understanding how children learn the meanings of words, from first words to later word learning, it is common and convenient to think of meaning in terms of personal and (at least partially) shared concepts, but it is important to recognize that the child's concept may differ from the adult's in size and numbers and kinds of connection to other parts of the conceptual system, being at first derived largely from the infant's perception and action in the world. In many cases the child may have no available concept to relate the word to. Finally, to my knowledge there is no established theory about how the young child's words relate to each other in the growing system, either within the conceptual system or separately (see Nelson, Reference Nelson1996, for discussion). There is much room for exploration here.

VARIABILITY AND CHANGE

One of the first points that I noticed when I carried out my original study of eighteen children's early word learning (Nelson, Reference Nelson1973) was how different children were in their ways of entering into and proceeding within the system of making sense. I expected to find that pretty much every ‘normal’ child would proceed in more or less the same way, learning names for things, and coming out at about two years with a workable vocabulary. (A similar assumption seems to be made by many in the field today.) The eighteen children I studied, however, differed in their acquisition patterns in many different ways. Perhaps the most obvious difference was the timing of their acquisitions, with a few children accumulating fifty or more words by age 1;1 or 1;2, and one or two others not reaching that ‘milestone’ until 2;0 or 2;2. This range of a year or more among normally developing English-speaking middle-class children still seems quite remarkable, although it is seldom remarked upon in the literature. All of the children in the sample had achieved vocabulary levels within a ‘normal range’ by 2;6. None were deemed language delayed at that time. Also, all had begun to speak a few words by 1;3. The difference was not so much in the starting as in the process that followed. I would have thought that this variability would have alerted researchers to ask what lay behind the lag of the slower learners, or led to the success of the early ones. What is it about learning words that might pose special problems for some but not others? I will return to this in the next section.

The other kind of variability I found in that first study, only partially related to the first, was the difference in the composition of the vocabularies – the kinds of words learned. It is still common to assume that most of the words learned during the second year are nouns, and most of those are object names. This is held to be such reliable truth that investigation of the acquisition of other kinds of words and phrases and other ‘styles’ of acquisition has been confined to a very few researchers (Lieven, Pine & Dresner Barnes, Reference Lieven, Pine and Dresner Barnes1992). Yet, among the children I studied, a significant minority did not enter the language with a caseload of object names. Instead they began with efforts to express their needs, interests, and feelings through both single words and short condensed phrases: ‘open the door’, ‘go away’, ‘where's it?’, ‘all gone’, ‘I love you’, and the like (see also Lois Bloom, Reference Bloom1993; Peters, Reference Peters1983). I labeled these two types ‘referential’ and ‘expressive’, although no child was simply one or the other.

During the ‘middle years’ of JCL there was a considerable interest in identifying differences among children in word learning and in other ways. Lois Bloom (L. Bloom, Lightbown & Hood, Reference Bloom, Lightbown and Hood1975) identified different nominal and pronominal emphases, while Bretherton, McNew, Snyder, and Bates (Reference Bretherton, McNew, Snyder and Bates1983) focused on differences in terms of analytic and holistic styles at age 1;8. These different foci converged on common bases for distinguishing two ‘styles’ of early learning. (For a review of the work from that period, see Nelson, Reference Nelson1985, chapter 4.) There has been relatively little follow-up work on this that I am aware of since that period.

From the perspective of the field of scholarship it seems reasonable to concentrate on one dominant learning problem (e.g. learning the meanings of nouns, or how to identify objects that are referable by a given name). However, neglecting the alternative entry ways that children engage in has resulted in a misplaced assumption that noun learning is where all the action is. Ignoring the other kinds of words or phrases that appear in large numbers on the widely used MacArther check list for toddlers, filled out by parents, leaves much of the data unaccounted for, as Nelson et al. (Reference Nelson, Hampson and Kessler-Shaw1993) documented.

Children classed as expressive in my original sample and in others on the basis of noun non-dominance generally learned words at a slower rate than referential children. However, this was not always the case and did not in fact account for the wide difference in age of achieving fifty words or more (on average 1;5–1;6).

Still another variability among children in the process of learning words is the issue of under- and overextension of a word's reference or meaning (see ‘Meaning and knowledge acquisition’ below). As previously noted, the general view today appears to be that overextension is a minor phenomenon, signifying little of interest, that goes away by age 2;0 in any event. This evaluation is in line with the belief that the dominant middle ground is where the general process we are looking for will be revealed, and that children in general have reliable cognitive mechanisms for acquiring the language and do not make egregious errors in mapping. However, it is also true that a good theory should explain the exceptions to the norm as well as the norm itself. Development is a complicated process involving the coordination of different systems, thus ‘errors’ may appear from many different areas. They should be studied for what they reveal about the underlying process rather than being ignored.

THE SOCIAL COALITION SYSTEM

Most children learn their first language within the family, with parents, other caregivers, and siblings, learning words in the course of interactions during play, household routines, story reading, and other such joint activities. This home context served as the background for most of the first wave of research on early word learning, supplemented by laboratory experiments to test particular hypotheses. Most of my work and that of my students was carried out in home settings, and it varied between simple observation, observation directed at specific behaviors, and experimental variations and tasks. Today, many studies use the MacArthur checklist (Fenson et al., Reference Fenson, Dale, Reznick, Bates, Thal and Pethick1994) for mothers to report numbers and kinds of words understood and/or produced by their child as a ‘language control’ for other tasks. The development of this checklist was a valuable contribution to the field, enabling comparison across languages and cultures, as well as social classes, but it cannot substitute for direct observation of the acquisition process.

There are important trade-offs here. Jerome Bruner's (Reference Bruner1983) account, based on analysis of videotape records of mother and child interactions, clearly reveals the advantages of close attention to the interactions of adult and child in understanding the emergence of shared meanings. It undermines the attempt to place the process of acquiring meaning solely in the head of the infant, and sets the stage for understanding the process of using language as a collaborative enterprise, one requiring cooperation on the part of both participants, as Tomasello (Reference Tomasello2008) has stressed. The ‘social side’ of language emergence in the child was revealed early in the form of ‘child-directed speech’ or ‘motherese’, which was shown to vary in significant ways from adult-directed speech in terms of vocabulary, pitch, and tonality (see Snow, Reference Snow and MacWhinney1999, for a history and update on work in this area). This characterization of a common form of speech observed across many different cultures (Fernald, Reference Fernald, Barkow, Cosmides and Tooby1992) seemed to indicate a ‘built-in’ disposition for communication with the young. Yet, further studies have also revealed that not all cultures engage in ‘motherese’ and that children learn languages with different forms of social support, some more directive than others. What has become very clear, however, is that children do not learn language without the opportunity to engage in interactive speech. Simply looking at and listening to television does not work.

My 1973 study highlighting the difference between expressive and referential learning children also focused on the parent−child collaboration patterns that might produce the style difference. What I presented there were both examples of mother−child pairs seemingly in tune with regard to what language was good for, and those of mothers who took a referential interest in tutoring their young on names for things with little interest or success on the part of the child. On the other hand, there were mothers who were casual, playful, non-tutoring users and whose children reflected that in their uses.

June Hampson (Hampson & Nelson, Reference Hampson and Nelson1993) followed up these patterns with a different sample of children, whose mothers varied between nominal (naming) and non-nominal interactions with their infants. She found that the children of nominal mothers mostly were nominal (name-learning) children, while non-nominal mothers generally had expressive- learning children. Fortunately, as required by the design of her later study, some children of nominal mothers did not focus on names, but fit into the expressive category. The assumption that early word learning is mostly name learning is not entirely discounted by these findings, however, in that the apparent frequency of nominal mothers in the middle-class population of New York is much higher than that of non-nominal mothers. Hampson's study required an equal distribution of M-C pairs in four categories: N-N, N-E,E-E, and E-N. The last category (expressive mother and nominal child learner) was the hardest to fill, requiring months of hunting throughout Manhattan and some of the outer boroughs. However, the fact that it was so difficult to find examples of this kind reinforces the conclusion that word learning is a two-person collaborative game. Further reinforcement was found in the follow-up to this study that showed that among the discrepant pairs (N-E and E-N) both mothers and children grew closer in their type of vocabulary use over time. Overall, these results seem to go against P. Bloom's claim (2000, p. 84) that mother's tutoring of words in the early learning phase has no effect on the child's acquisition beyond possibly speeding up the process.

As many observers have noted, and some have studied (e.g. Benedict, Reference Benedict1979), infants generally come to comprehend the point of some words before they attempt to produce any on their own. This phase is difficult to pin down because what constitutes evidence of comprehension by babies aged 0;9 to 1;0 is not clear. Some children are apparently temperamentally disposed to rely on comprehension for a much longer period of time – sometimes to the end of the second year – than others. In those children that I originally studied these were primarily second-born of a closely spaced two-child family.Footnote ² It seems most likely that this comprehension without production ‘strategy’ is not only an effect of the child's disposition but of the social environment, as has often been suggested. Mothers are likely to spend less time interacting one-on-one with a second child, thus engaging in less of the intensive tutoring that Bruner (Reference Bruner1983) observed, while the first-born is chattering with adults and other children, providing less of a clear model than the mother might have.

The child's prelinguistic social understanding has been given extensive attention by Tomasello and his colleagues (see Tomasello, Reference Tomasello2014, for an update) and others. While important and also somewhat controversial, this is a somewhat different point that relates also to how children use their prelanguage action knowledge to support understanding of words (Akhtar, Reference Akhtar, Hall and Waxman2004; Nelson, Reference Nelson1985, Reference Nelson1996; Woodward, Reference Woodward, Hall and Waxman2004).

In summary, whereas Snow has provided a clear account of the significance of social interaction in relation to language acquisition in general, and there is now a general acceptance of this point, it is still true that very little is known about the course of word learning or grammar acquisition in relation to different contexts of social settings. What we know about children talking to adults in different settings has been well documented (e.g. Blum-Kulka & Snow, Reference Blum-Kulka and Snow2002) but there is a great deal more to be learned in this area.

MEANING AND KNOWLEDGE ACQUISITION

The child's first words enable a number of functions for communication with others. But the most dramatic change in the child's cognitive and social situation involves the ability to learn through and from language, using the repertoire of shared meanings referring to and characterizing the world in new ways, whether in conversation, story-reading, or deliberate instruction. This area is critical to the child's entry into the cultural community and into the educational system. Critical as it is, it has largely been left to linguists and educational professionals to deal with. As Catherine Snow and her colleagues have documented, vocabulary acquisition does not become less interesting after the age of two, but rather deserves intensive investigation to discover how the child manages this process in collaboration with family and teachers. As stressed previously, the process cannot simply be described as ‘mapping’ to already available concepts.

The process of acquiring language for thinking and knowing is related to the earlier processes of acquiring a first vocabulary, but involves far more complexity. This is no doubt where more verbal, professional families excel in preparing children for further educational experiences. This preparation begins with reading stories and sharing narratives of all kinds, such as talk about shared experiences, as well as providing information about exhibits when visiting the zoo, for example. Good preschool programs excel at such tasks, providing the background for further exploration of new knowledge and ideas.

OVERVIEW

This essay reflects the contributions of many researchers to the puzzles of how young children begin and proceed to build up the semantics of their language, enabling them to engage in conversations of many kinds with both adults and peers, on their way toward the formal educational process. What I have referred to here are many of the articles that I and my colleagues and students and many others have contributed to JCL and other outlets over the last forty years on the general theme of mastering the meanings of verbal communication in its first stages. Although some sources treat the acquisition of words and their meanings as almost automatic, easily mastered, I believe the process to be challenging, especially at the outset, and non-automatic, but also intriguing and enlightening to most young children. It lifts them to an entirely new level in the world of being and consciousness, socially, culturally, and mentally. It is the heart of language itself (the grammar and other parts being merely mechanics). I would urge any beginning scholars to survey this field with the understanding that there is much work to be done: what we have planted is but a beginning of uncovering the mysteries of meanings.