Effects of input on early word learning

This research will examine aspects of Infant Directed Speech in conjunction with tests of word learning. We will include a home recording (Study I) or training period (Study II), followed by experimental testing to ascertain learning outcomes. Study I will explore the effect on word learning of (a) the frequency of occurrence of a word and (b) variability in usage. We will exploit the naturally occurring individual differences in family uses of infant names - as regards both frequency of use and variability in its forms (i.e., varied names or nicknames) - and of the well established finding that their own names are among the first word forms that infants recognize. We will begin by arranging for the home recording of 6-month-old infants in order to obtain spontaneous input speech, in the absence of any observer, to establish the amount and form of use of the infant's name. The lab test for Study I will then present to infants 'target' words such as bike or cup, which they have been found not to recognize at that age, as part of a short passage in which the word repeatedly follows either (a) the child's name or (b) a closely matched alternate name. The test portion of the experiment will involve presentation of the two target words in random alternation with two unfamiliarised words. The learning outcome is then gleaned from differences in listening time to the (a) word, heard combined with the familiar name, as compared with the (b) word, heard combined with a structurally matched unfamiliar name and the two unfamiliarised words. We expect those infants whose home recording included more uses of their name and/or fewer variants to show longer listening to the words that were combined with their name in the passage than will the remaining infants.
Secondly, we are interested in ascertaining whether or not hearing words in isolation leads to better learning outcomes than does hearing them embedded in sentences. We will provide parents of infants who have reached 11 months of age with a picture-book with the names of each image printed either as a one-word label or as the final word in a short multiword sentence, with each book containing six words presented in each way (with each word presented in isolation for half of the children, in sentences for the other half). The parents will be asked to read the book twice a day over a three-week period, without deviating from the wording we provide. After the training period the infants will be tested in the lab. Half of them will be tested with lists of words contrasting (a) the set of words trained in isolation, (b) those trained in sentences, and (c) untrained words. The other half of the infants will be tested on a 'segmentation task', which requires the identification of words in passages of continuous speech, again with passages contrasting the three word types. The learning outcome will again be measured in terms of listening times, here to words trained in isolation and in sentences and to untrained words. We expect that in both tasks infants will show a higher rate of preferential attention to the words trained in isolation in comparison with the words trained in the sentence-final condition (which will differ individually by infant), although all are expected to show longer looking to trained over untrained words.
The issues investigated here, the significance of quantity and variability of speech input to infants and of the context in which it is presented, are both important questions in the study of the effects of input on language development. Taken together with preliminary results showing cultural differences in the quantity of language used with infants as well as in the degree of use of isolated words in input speech (Keren-Portnoy et al., 2010), the proposed research can lead to more efficient and more solidly research-based intervention programs with parents of young infants, especially in cultures in which parents do not often speak to their infants.

We believe that the findings of the proposed study will be of potential interest to Speech and Language Therapists as well as to any professionals involved in parent or family training in lower SES groups. For therapists, our findings as to the most salient and memorable aspects of the input received by infants in the home, with specific evidence regarding learning outcomes, should provide useful information as to the most effective form of modeling to be used in intervention with infants at risk of language delay, whether for economic, social or genetic reasons.
For parental training, similarly, concrete findings demonstrating the impact not only of quantity of input but also of differing ways of packaging input to infants may be an important benefit. Note that such techniques as Baby Sign are often in the news lately and have become popular with parents, despite the lack of any evidence of efficacy in typically developing children of educated families [http://www.guardian.co.uk/lifeandstyle/2010/mar/14/sign-language-for-babies]; this demonstrates the widespread parental interest in ways of maximizing learning opportunities for the child. The claims made are often extreme, as in a recent report on NBC in the US suggesting that baby sign leads to 'an increase in IQ and greater emotional stability' (http://www.nbc.com/news-sports/msnbc-video/teaching-baby-sign-language-before-speech/). In reality, introducing infants to such additional communication systems is unlikely to be needed either in traditional societies, where long cultural experience supports child raising methods, often with strong community support, or in well-educated familiels in European society. However, as Heath (1983) was perhaps the first to show, lower SES groups, living in urban situations, may be disconnected from either education or tradition. For such families the findings of our studies could be of real value.