How infants learn signs for objects : testing the nature of unimodal mappings

Access full-text files




Shield, Aaron Michael

Journal Title

Journal ISSN

Volume Title



This paper reports three experiments on infants’ sign learning. In order for an infant to learn a spoken word, s/he must first associate the sound pattern of that word with an object by mapping an auditory stimulus (the word) onto a visual stimulus (the referent). Moreover, presentation of the label and of the referent is likely to happen simultaneously during episodes of joint attention between mother and child. Much less is known about how infants learn signs. Unlike word learning, the learning of a sign involves a unimodal mapping between two visually presented stimuli. The signing child must look at both the referent and the signed label in order to register a sign object mapping. Instances of joint attention are thus complicated by the fact that either the child must shift his/her attention from the referent to the label or the parent must employ strategies to consolidate the object and the label within the child’s field of vision (Holzrichter & Meier, 2000). Despite these seeming complications, sign learning is not delayed vis-à-vis the acquisition of first words (Newport & Meier, 1985; Petitto, 1988). Werker, Cohen, Lloyd, Casasola and Stager (1998) sought to determine the age at which hearing infants could learn word-object pairings with minimal exposure and without contextual support. Using a “switch” paradigm, they found that 14-month-olds were able to make such mappings. Infants were exposed to pictures of unfamiliar objects while hearing a made-up word until a habituation criterion was met. In the test phase of the experiment, the associations between labels and objects were switched. The 14- month-olds looked significantly longer during the switched trial, indicating that they were surprised to hear a label/object pairing different from the one they had habituated to. Thus, those infants had successfully made a mapping between word label and object. Keeping the basic design of this study, we sought to probe the conditions under which a sign-object mapping can be made. In three studies we familiarized 14-month-old hearing infants to videotaped, non-iconic signs (the American Sign Language signs DOG and TRAIN) paired with pictures of objects. Infants were then tested on familiar and nonfamiliar sign-object pairings. The Pilot Study included 6 infants. Experiments 2 and 3 each included 32 infants but differed in the stimuli shown to the infants. Experiment 2 showed only the signer’s torso and hands. In Experiment 3, the signer’s face was also shown, and the signer pointed at the object being labeled before producing the sign. Thus, more social referencing information was provided in Experiment 3 than in Experiment 2. 14-month old girls, but not boys, looked longer at unfamiliar sign-object pairings than at familiar ones in Experiment 3, thus indicating sign learning. However, neither boys nor girls looked longer at unfamiliar sign-object pairings in Experiment 2, indicating that no association between sign and object had been made. Therefore, 14-month-old girls showed evidence of having made an association between a sign and an object when shown signs that included the signer’s face and pointing at the object. These studies help clarify under what conditions sign-object mappings take place, and how such mappings differ from spoken language word-object mappings



LCSH Subject Headings