Understanding how children acquire language is one of great challenges for the social sciences with critical implications for education and for intervention in atypically developing children. Vocabulary learning is a core part of language development and is characterised as a particularly hard problem: How do children know that the sounds people make with their mouths are ‘words’ and that they are names for objects, actions or properties?
The assumption that words are only arbitrarily linked to objects and actions in the world (i.e., there is nothing in the sound of ‘cat’ that brings to the mind’s eye what a cat looks like, how it moves etc) makes the task of learning words especially hard: how can the correct object be found in a visually cluttered world (when the object is one among many present), or worse, when the object is absent from the immediate environment?
Dr Pamela Perniss and colleagues on the project argue that language is, in addition to being indisputably arbitrary, also fundamentally iconic (i.e. maintaining transparent links between spoken form and meaning) and that arbitrariness and iconicity coexist to aid learning in different ways. In particular, iconicity would provide powerful cues to support learning, especially when communication concerns objects and events not visually present, by bringing some of their sensory-motor properties to the mind’s eye (e.g. when a caregiver elongates the vowel in "taaall” to refer to a very tall person).
Language has potential for iconicity in different channels of expression. First, iconicity can be found in the phonology of words, e.g. in onomatopoeia such as ‘meow’ or ‘drip’. Moreover, co-speech gestures, hand movements made by speakers can evoke the shape or movement of objects being talked about, and prosody (as in the ‘tall’ example) both have a high potential for iconic mappings. The project asks: (1) Do caregivers use iconicity (in phonology, gestures, and/or prosody) in their input to children? (2) Does iconicity aid children’s word learning? (3) Is iconicity especially useful when talking about objects which are absent in immediate environment? These questions are addressed in a complementary series of studies using naturalistic observation, semi-naturalistic methods, and standard experiments that assess iconicity in the speech, gestures, and prosody of caregivers and children aged 2-3 years. The researchers focus on this age as it is a time of remarkable vocabulary growth in which communication is often about things not present in the immediate environment (happened before, or even imaginary), and at which children understand and produce iconic gestures.
There are three main strands of investigation. In Strand 1, we take advantage of a unique corpus of caregiver-child interactions (collected by collaborator Prof Susan Goldin-Meadow, University of Chicago) in order to assess the presence of iconic cues in caregivers’ language input and their correlation with vocabulary measures taken later in life (at school entry). In Strand 2, we complement this data with collecting a semi-naturalistic audio-visual corpus of caregiver-child interactions in which objects being talked about are present or not, and known to the child or novel (in order to assess learning of new label-referent pairings). This corpus will be annotated along similar lines as the naturalistic corpus to assess the presence of iconicity in caregivers’ communication. Finally, in Strand 3 we ask whether iconic cues in the input are causally linked to learning by experimentally manipulating the presence and kind of iconic cues (and the channel in which they occur) in learning novel words.
As this research has the potential for important impact on learning and development in the early foundational years, we will actively involve educators and developmental speech-and-language therapists by setting up a researcher-practitioner network to disseminate and disc
uss implications of research within the community of practitioners.
The Role of Iconicity in Language Learning is a 3-year ESRC project, conducted by Dr Pamela Perniss at the University of Brighton together with Prof Gabriella Vigliocco (PI), Dr Liz Wonnacott and Dr Chloe Marshall at UCL. The total funding is £655,000, of which £250,000 is at Brighton with a Research Officer post. Dr Pamela Perniss is the lead on the corpus-based work (naturalistic and semi-naturalistic data) in strands 1 and 2 of the project.