ISS · Infant speech segmentation and the role of the emergent lexicon
FP6 — Marie Curie Actions (Human Resources and Mobility)
- Duration
- 2005-02-21 → 2006-02-20
- EU contribution
- €114,664
- Participants
- 1
- Scheme
- IIF
Lines connect the coordinator with its partners.
Results in brief
Final Activity Report Summary - ISS (Infant speech segmentation and the role of the emergent lexicon)
The two major objectives of this study were to determine 1. whether 11-month-old infants segmented familiar words from connected speech; and 2. whether early favoured babbling patterns or Vocal motor schemes (VMS) had an effect on speech perception. In order to achieve the first project objective included 24 11-month-old and 12-month-old infants which were tested using the head turn preference paradigm to contrast passages containing trochaic words, i.e. words with strong-weak stress, likely to be familiar to infants, such as tummy and tickle, with rare words, e.g. tangy, piffle. The words were inserted near the beginning or the end of sentences spoken by a female speaker in a manner typical of infant-directed speech. The carrier sentences were identical for the familiar and the rare words. Looking times were recorded for orientation towards the two types of passages. Infants of both ages showed evidence of segmenting previously known words from passages, even though the effect was stronger at 12 than at 11 months of age. The relatively weak effect at 11 months, despite earlier evidence of 7.5-month-old infants' success in segmenting words trained in the laboratory, suggested a need for more ecologically valid tests in order to trace the course of naturalistic word learning capacities. In order to achieve the second objective, 24 infants were followed longitudinally to determine each infant's favoured babbling patterns or VMS. 14 of these infants were tested in the perception phase of the study. The results of these head turn experiments were unequivocal with respect to the relation between the number of VMS and the infant looking preferences. Infants with one VMS responded with longer looks to passages with inserted nonsense words which had the VMS that they produced, while infants with two VMS displayed longer looking times to the passages that contained the consonant that they were not yet consistently producing. Since the transition from one to two VMS typically occurred as a matter of weeks at most, the infants' categorical attention shift away from consonants that they produced towards those that they were about to produce was remarkable. Contrary to the accepted notion that perception always preceded production, this result strongly argued for production initiating a perceptual shift towards 'noticing' contrasts that paved the way for further phonetic development.
Data: CORDIS, © European Union
Project objective
The goal of the project is to bring to Europe for one year an innovative researcher in mid-career who has established a strong collaborative link with one researcher in the UK (the project host). The researcher stands to gain momentum and visibility in his career as well as to forge valuable new contacts with members of the European research community.His presence will also serve as a further stimulus to the vibrant international team of young researchers currently involved in developing a model of infant word learning in a School of Psychology of high distinction in a less favoured region. The Fellow will receive advanced training in phonetic transcription and the assessment of infant vocalisations while running a complementary experimental study of infant speech perception using the three-sample technique he designed for the Head Turn procedure. The project will focus on issues of early word representation and vocal production in speech segmentation during the period when infants begin to make the transition into language.The study is designed to test two specific hypotheses regarding infants' segmentation abilities. First, children will be tested on their ability to use the familiarity with words that they have gained outside the experimental situation to segment short passages of running speech. Second, recordings of spontaneous vocal productions will be made over a period of months to identify each child's most common patterns. Stimuli will then be developed to test infant attention to passages that d o or do not include these patterns. The findings are expected to determine whether an infant's newly emergent lexicon serves as an entry point for the segmentation of the speech stream.Two-way knowledge transfer between Europe and the US will be achieve d by the project, and the profile of the EU will gain prominence through the publication of the findings of the study and the model that is to be supported and elaborated over the period of the fellowship.
Original text from CORDIS.
Participants
- UNIVERSITY OF WALES - BANGOR · BANGOR (GWYNEDD)CoordinatorUnited Kingdom
Links
Data: CORDIS, © European Union