The role of novelty in word learning

The ability to use one's knowledge of an object's name to infer that a novel word might mean something else is a potentially powerful tool for vocabulary acquisition. It is particularly useful in the context where a word learner is confronted with two objects, one name-known and the other nameless, and then hears a novel label, offering the learner the opportunity to infer that the novel word refers to the previously nameless object. When and how do young infants come to understand that a novel label is a candidate name for a nameless object in such situations? Previous research does not provide a clear answer to these questions: Some researchers argue that this word learning strategy is available at the very outset of word learning whereas others have argued that it takes a long time to master - well into the child's third year. Researchers have not provided an explanation of how this ability might be mastered. The proposed research aims to tackle both of these questions by testing infants under tightly controlled laboratory conditions using a preferential looking procedure. Infants will sit on their caregivers lap in front of two pictures. One picture will depict a familiar object whose name is known to the infant (as judged by a parental questionnaire). The other object will be novel. The infant will hear a command over a loudspeaker instructing them to 'Look at the XXX' where XXX is a novel label such as 'wug'. During the presentation of the images, we will monitor the infants fixations of the two pictures using a precise eye-tracking device. We want to know whether the infant will show a preference for the novel object and form an association between the novel label and the novel object. We will test this by providing the infant with opportunities to learn two novel object-label associations in this way and then presenting them with both novel objects and one of the novel labels and measuring whether they demonstrate a looking preference for the newly-named object when they hear its label. Infants will be tested at 14 months, 19 months and 24 months in order to identify the age at which this strategy for learning new word is mastered. We will then use the youngest age group to investigate how they master this skill. The central idea behind this research programme is that infants exploit the relative novelty of objects and labels to learn new object-label associations. However, there are two ways that novelty might play a role in word learning: Firstly, the relative novelty of nameless objects in the presence of a name-known object may determine the outcome of the word learning process. To test this possibility, two nameless objects which differ in novelty from each other will be presented with a name-known object. We predict that infants will associate the novel word with the completely novel object. Secondly, the relative novelty of objects themselves may be sufficient to trigger word learning because infants pay more attention to novel objects. If they hear a novel word at the same time as viewing two nameless objects, then infants may come to associate the more novel object with the novel word. A positive outcome would indicate an important role for novelty in the process of word learning. We will use the preferential looking task to test the role of novelty for word learning in both of these situations.

The ontogenesis of mutual exclusivity (ME) is not well-understood, either in terms of its developmental trajectory or in terms of the mechanisms that underlie its use. Some researchers have argued the ME is available to the infant at the very outset of word learning whereas others have argued that it is not mastered until well into the child's third year. Likewise, there is disagreement regarding the nature of the underlying mechanisms involved: One possibility is that the infant uses socio-pragmatic knowledge to infer the meaning of novel words in a syllogistic-like manner. Another possibility is that sensitivity to novelty (novel objects go with novel labels) can provide the foundations for the mastery of ME. These issues are addressed using a combination of habituation and inter-modal preferential looking (IPL) procedures, in three sets of experiments. In the initial set of experiments, we propose a cross-sectional design with 15-, 18- and 24-month-olds to identify when infants reliably use ME to learn new words. This is achieved by training infants in a laboratory setting with two sets of name-known and name-unknown objects, thereby offering them the opportunity to learn the referents of two novel labels. Retention of object-label mappings are then tested using IPL. In the second set of experiments, infants are presented with three objects, one name-known and two name unknown. One of the name-known objects has been pre-exposed prior to training (using habituation to criterion). If novelty is critical to word learning in ME situations, we predict that infants will map novel labels to the completely novel object. In the final set of experiments, we examine whether novelty by itself is adequate for learning label-object mappings. Infants are trained with two name-unknown objects, where one has been pre-exposed and the other is completely novel. We predict that infants will map novel labels to completely novel objects.