Kim Plunkett

During the second year of life, infants develop a preference to attach novel labels to novel objects. This behavior is commonly known as “mutual exclusivity” (Markman, 1989). In an intermodal preferential looking experiment with 19.5-and 22.5-month-olds, stimulus repetition was ...

There is a lack of stability in language difficulties across early childhood: most late talkers (LTs) resolve their difficulties by pre-school; and a significant number of children who were not LTs subsequently manifest language difficulties. Greater reliability in predicting individual outcomes is needed, which might be achieved by waiting until later in development when language is more stable. At 18 months, productive vocabulary scores on the Oxford Communicative Developmental Inventory were used to classify children as LTs or average talkers (ATs). Thirty matched-pairs of LTs and ATs were followed up at school-age (average age 7 years), when language and literacy outcomes were assessed. For 18 children, intermediate testing at age 4 had classified them as showing typical development (TD) or specific language impairment (SLI). After correcting for multiple comparisons, there were no significant differences between the LTs and ATs on any outcome measure, and the LTs were performing in the average range. However, there were large-sized effects on all outcomes when comparing the TD and SLI groups. LT status on its own is not determinative of language and literacy difficulties. It would therefore not be appropriate to use expressive vocabulary measures alone to screen for language difficulties at 18 months. However, children with language impairment at age 4 are at risk of enduring difficulties.

Recent behavioural studies with toddlers have demonstrated that simply viewing a picture in silence triggers a cascade of linguistic processing which activates a representation of the picture's name (Mani and Plunkett, 2010, 2011). Electrophysiological studies have also shown that viewing a picture modulates the auditory evoked potentials (AEPs) triggered by later speech, from early in the second year of life (Duta et al., 2012; Friedrich and Friederici, 2005; Mani et al., 2011) further supporting the notion that picture viewing gives rise to a representation of the picture's name against which later speech can be matched. However, little is known about how and when the implicit name arises during picture viewing, or about the electrophysiological activity which supports this linguistic process. We report differences in the visual evoked potentials (VEPs) of fourteen-month-old infants who saw photographs of animals and objects, some of which were name-known (lexicalized), while waiting for an auditory label to be presented. During silent picture viewing, lateralized neural activity was selectively triggered by lexicalized items, as compared to nameless items. Lexicalized items generated a short-lasting negative-going deflection over frontal, left centro-temporal, and left occipital regions shortly after the picture appeared (126-225ms). A positive deflection was also observed over the right hemisphere (particularly centro-temporal regions) in a later, longer-lasting window (421-720ms). The lateralization of these differences in the VEP suggests the possible involvement of linguistic processes during picture viewing, and may reflect activity involved in the implicit activation of the picture's name.

A three-layer back-propagation network is used to implement a pattern association task in which four types of mapping are learned. These mappings, which are considered analogous to those which characterize the relationship between the stem and past tense forms of English verbs, include arbitrary mappings, identity mappings, vowel changes, and additions of a suffix. The degree of correspondence between parallel distributed processing (PDP) models which learn mappings of this sort (e.g., Rumelhart & McClelland, 1986, 1987) and children's acquisition of inflectional morphology has recently been at issue in discussions of the applicability of PDP models to the study of human cognition and language (Pinker & Mehler, 1989; Bever, in press). In this paper, we explore the capacity of a network to learn these types of mappings, focusing on three major issues. First, we compare the performance of a single-layered perceptron similar to the one used by Rumelhart and McClelland with a multi-layered perceptron. The results suggest that it is unlikely that a single-layered perceptron is capable of finding an adequate solution to the problem of mapping stems and past tense forms in input configurations that are sufficiently analogous to English. Second, we explore the input conditions which determine learning in these networks. Several factors that characterize linguistic input are investigated: (a) the nature of the mapping performed by the network (arbitrary, suffixation, identity, and vowel change); (b) the competition effects that arise when the task demands simultaneous learning of distinct mapping types; (c) the role of the type and token frequency of verb stems; and (d) the influence of phonological subregularities in the irregular verbs. Each of these factors is shown to have selective consequences on both successful and erroneous performance in the network. Third, we outline several types of systems which could result in U-shaped acquisition, and discuss the ways in which learning in multi-layered networks can be seen to capture several characteristics of U-shaped learning in children. In general, these models provide information about the role of input in determining the kinds of errors that a network will produce, including the conditions under which rule-like behavior and U-shaped learning will and will not emerge. The results from all simulations are discussed in light of behavioral data on children's acquisition of the past tense and the validity of drawing conclusions about the acquisition of language from models of this sort.

The traditional account of the acquisition of Englishverb morphology supposes that a dual mechanismarchitecture underlies the transition from earlyrote learning processes (in which past tense formsof verbs are correctly produced) to the systematictreatment of verbs (in which irregular verbs areprone to error). A connectionist account supposesthat this transition can occur in a single mechanism(in the form of a neural network)

Developing Object Permanence: A Connectionist Model Denis Mareschal Experimental Psychology Oxford University South Parks Road Oxford OX 13UD deni s@ psy. ox. ac. uk Abstract Kim Plunkett Experimental Psychology Oxford University South Parks Road Oxford OX 13UD ...

The impact of labelling on infant visual categorisation has yielded contradictory outcomes. Some findings indicate a beneficial role while others point to interference effects in the presence of labels. The locus of these divergent outcomes is largely unclear. We explore the hypothesis that the timing of the label is of crucial importance, proposing that synchronous presentation of words and objects induces a higher processing load than asynchronous presentation (image onset before labelling). A novelty preference experiment with 12-month-olds reveals that synchronous presentation leads to a diminished preference for a novel object on test in comparison to asynchronous labelling, suggesting a detrimental impact on category learning. However, analyses of infants' gaze patterns to object parts reveal that even synchronous labels do not hinder learning completely. We conclude that synchronous labels interfere with the familiarisation process, but this process involves shifts in fam...

Strong associations between infant vocabulary and school-age language and literacy skills would have important practical and theoretical implications: Preschool assessment of vocabulary skills could be used to identify children at risk of reading and language difficulties, and vocabulary could be viewed as a cognitive foundation for reading. However, evidence to date suggests predictive ability from infant vocabulary to later language and literacy is low. This study provides an investigation into, and interpretation of, the magnitude of such infant to school-age relationships. Three hundred British infants whose vocabularies were assessed by parent report in the 2nd year of life (between 16 and 24 months) were followed up on average 5 years later (ages ranged from 4 to 9 years), when their vocabulary, phonological and reading skills were measured. Structural equation modelling of age-regressed scores was used to assess the strength of longitudinal relationships. Infant vocabulary (a latent factor of receptive and expressive vocabulary) was a statistically significant predictor of later vocabulary, phonological awareness, reading accuracy and reading comprehension (accounting for between 4% and 18% of variance). Family risk for language or literacy difficulties explained additional variance in reading (approximately 10%) but not language outcomes. Significant longitudinal relationships between preliteracy vocabulary knowledge and subsequent reading support the theory that vocabulary is a cognitive foundation of both reading accuracy and reading comprehension. Importantly however, the stability of vocabulary skills from infancy to later childhood is too low to be sufficiently predictive of language outcomes at an individual level - a finding that fits well with the observation that the majority of 'late talkers' resolve their early language difficulties. For reading outcomes, prediction of future difficulties is likely to be improved when considering family history of language/literacy difficulties alongside infant vocabulary levels.