Abstract
This thesis is a cross-linguistic and developmental investigation of the development of language and working memory. English, Greek and Himba (a language spoken in Namibia) differ with respect to the domain of colour, providing a fruitful environment for such an investigation. The theory of colour universals (Berlin & Kay, 1969) predicts that a language should encode a maximum of eleven colour terms (BCTs), and English is a good example of this theory. Greek, however, encodes twelve instead of the suggested maximum of eleven BCTs, splitting the blue region into two sub-regions, a light and a dark one, and respectively encoding two instead of the predicted one blue term. Himba on the other hand, encodes five BCTs. In addition, Greek BCTs are overall longer than the corresponding English terms. Children aged 3-6 years were tested on a list, a colour naming and a comprehension task. Overall, child naming reflects adult naming, and colour term acquisition progresses from the acquisition of basic colour terms to the use of modifiers and non-basic terms. At the age of three, Greek and English children name correctly a mean of almost five out of the six focal primary colours and colour term acquisition order is marked by the late acquisition of grey and brown. The differences in the way colour space is partitioned among Greek and Himba seemed to be reflected in a non-linguistic memory recognition task, as shown by individual score correlations, correlations and MDS analyses of naming and memory similarity matrices and same-name patterns of errors. On the other hand, the differences in the colour word length among Greek and English seemed to account for a word-length effect, detected both on a digit and a colour word span. This word length effect disappeared when the same task was purely visual, showing the yet inadequate development of a memory encoding strategy. Those effects of both internal and external linguistic differences on memory indicate that the predictions of the weakest form of linguistic relativity can be detected among young children.