Language can cement relationships, start wars, and allow us to say what we want in our coffee. It is one of the few psychological behaviors that separate us from our evolutionary cousins, the great apes. The quest to understand language, how we use it, and how children learn it, is age-old. In the past 50 years, however, significant progress has been made in understanding how children acquire language by utilizing the intricate interdisciplinary relationship between the linguistic, psychological, and computational study of language.
New methods to study language learning. For over 30 years, my lab has been working to probe the process of language learning. Along the way we developed two methods to probe children’s language comprehension as children’s understanding of language often precedes their ability to talk. Both methods “trick” babies into showing us what they know by relying on what they look at (their visual fixation) as their response. One method we developed in the late 1980’s (e.g., Golinkoff, Hirsh-Pasek, Cauley, & Gordon, 1987; Hirsh-Pasek & Golinkoff, 1996; Hollich, Hirsh-Pasek, & Golinkoff, 1998), the Intermodal Preferential Looking Paradigm (IPLP), is in use in laboratories around the world and relies on the video presentation of simultaneous events. Children see, for example, a boat on one side of the screen and a shoe on the other, while they hear audio that matches only one of these images (as in, “Where’s the shoe?”). We measure whether children look longer at the shoe or at the boat. If they understand the item in question (here, “shoe”), they should look longer to the shoe than the boat. Another method uses the basic premise and response of the IPLP but in “real time,” with a real person and real objects (the Interactive IPLP). We have used this method to study word learning in children as young as 10 months of age (Hollich, Hirsh-Pasek, & Golinkoff, 2000; Golinkoff & Hirsh-Pasek, 2006).
Theory development: The Emergentist Coalition Model (ECM). Our focus has been mainly on grammatical development (how children learn the syntax of their language), lexical development (how children learn their language’s vocabulary), and segmentation (how children carve up the stream of speech they hear into words). We have developed a theory called the “Emergentist Coalition Model” (Hirsh-Pasek & Golinkoff, 1996; Hollich et al., 2000) that is a truly developmental model of how language learning changes in character over the first two years of life. Using this theory as our guide, we produced a monograph on how children learn nouns, or names for objects (Hollich et al., 2000). To become a sophisticated word learner, children must learn to take a speaker’s point of view into consideration to learn the names of all kinds of things in the environment – even things that children don’t necessarily find interesting or attractive (Pruden, Hennon, Hirsh-Pasek, & Golinkoff, 2006). There are some assumptions though, that children seem to make at the beginning of word learning that they will have to relax. For example, at 12 months of age, children assume that words map to whole objects (e.g., Hollich, Golinkoff, & Hirsh-Pasek, in press) and not to their attractive parts.
How do children learn verbs? We have also made significant inroads into understanding how children learn verbs – a more difficult task than learning nouns (see Hirsh-Pasek & Golinkoff, 2006). Verbs are the architectural centerpiece of the sentence so learning verbs is tantamount to learning the grammar of one’s language. When we started our research on verbs, we were surprised at how hard they were for children to learn. Now however, we recognize a number of the factors that contribute to verb learning (e.g., Maguire, Hirsh-Pasek, & Golinkoff, 2006; Brandone, Pence, Golinkoff, Hirsh-Pasek, 2007). Our theoretical perspective and some of the research conducted in our labs were presented in our keynote address before the Boston Child Language Meeting in November, 2006 (Golinkoff & Hirsh-Pasek, 2007; Hirsh-Pasek & Golinkoff, in press).
Research on segmentation. Even before babies can learn the syntax or grammar of their language, they need to find the units that comprise their language in the stream of speech they hear. This is the problem of segmentation. Remember how difficult it is to listen to people talking in a foreign language? Where do the words and sentences begin and end? Yet babies who can’t even tie their shoes can find the units in their native language by the end of the first year of life! With Heather Bortfeld and James Morgan, we discovered that one of the “wedges” babies use to break into the language stream is their own name! When they hear their name in a passage, they can remember the word that comes after it. This also works for “mommy” or “momma” – whatever moniker their mother uses to refer to herself (Bortfeld, Morgan, Golinkoff, & Rathbun, 2005).
Babies’ perception of language-relevant aspects of events. In another line of research, we probe the concepts that underlie language (e.g., Pulverman, Hirsh-Pasek, & Golinkoff, 2006; Parish, Pruden, Hirsh-Pasek, Ma, & Golinkoff, in press). For example, if the languages of the world refer to how an action is performed in their verbs (e.g., running versus walking), and the path along which the action takes place (e.g., in a circle versus up and down), babies need to be able to perceive these elements in events. To ask whether and when babies can formulate dynamic language-relevant concepts that will surface in language in verbs and prepositions, we present babies with nonlinguistic events and use visual fixation once again as a “trick” to get babies to show us what they know. We have found that babies can discriminate between different kinds of actions and paths in events (e.g., Pulverman & Golinkoff, 2004) as well as form concepts of the actions and paths (e.g., Pruden, 2006; Song et al., 2007). Now with funding by NIH and NSF we continue to ask how babies perceive the events that language will label and how they can even find the events in the never-ending happenings in the world around them.
Implications of the Emergentist Coalition Model for language disorder. As the ECM is a model that emphasizes the availability of multiple cues to word learning, its tenets have implications for understanding and possibly remediating various types of language disorders. For example, we have used our Emergentist Coalition Model to guide us in research on how children on the autistic spectrum learn words (Parish, Hennon, Hirsh-Pasek, Golinkoff, & Tager-Flusberg, in press). Autistic children who are more sensitive to what a speaker is naming have larger vocabularies than autistic children who have difficulty inferring a speaker’s intent. These findings tell us that children can learn words without being attuned to a speaker’s intent – a problem for many autistic children. These children can learn words by using perceptual cues, such as learning the names for objects they find attractive. Our plan is to further expand the utility of the ECM for language disorder as it suggests that there may be multiple paths to the same goal.Books and monographs on language development. While much of our research on language acquisition appears in professional journals in article form, we have also written and edited books. The ECM, the new method we developed, and a number of experiments were first presented in “The Origins of Grammar: Evidence from Early Language Comprehension” (1996 – MIT Press). ; Elaboration of the ECM for word learning, the extension of the new method to real-time interaction (Interactive IPLP), and 12 experiments appeared in our SRCD Monograph (Hollich et al., 2000). “Action Meets Word: How Children Learn Verbs” (Oxford University Press) is a new edited collection on verb learning (Hirsh-Pasek & Golinkoff, 2006), written by a superb group of international scholars. Much of the research of our field on language acquisition is found in our popular press book, “How Babies Talk: The Magic and Mystery of Language in the First Three Years of Life” (Dutton/Penguin, 1999- translated into French, Spanish, Italian, Chinese – mainland and Taiwan).