Copyright © The British Psychological Society Reproduction in any form (including the internet) is prohibited without prior permission from the Society Acquisition of relations between the conceptual and linguistic dimensions of linearization in descriptive text composition in grades five to nine: A comparison with oral production Monik Favart* and Jean-Michel Passerault Centre de Recherche Sur la Cognition et l’Apprentissage, CNRS, Universite ´ de Poitiers, Poitiers, France The present study looked at how children establish a relationship between the conceptual and linguistic dimensions of linearization in descriptive text composition. Written productions were compared with oral ones. French-speaking participants, drawn from the fifth, seventh, and ninth grades, produced both a written and an oral description of a picture divided up into five clusters of specific units. This material made it possible to distinguish between the macro-structural and micro-structural levels of the resulting productions. The conceptual dimension of linearization was levelled out: participants were only selected if their texts reflected the hierarchical structure of the referent, successively describing the five clusters. The linguistic dimension of linearization was tested by comparing the use made of connectives in both modalities and analysing punctuation marks and speech pauses according to the modality. We established a specific classification for the analysis of spatial connectives, distinguishing between absolute and relative connectives. As expected, the proportion of texts reflecting the hierarchy of the referent increased through grade levels. From fifth grade onwards, and in both modalities, this hierarchy was marked by appropriate linguistic devices: at the macro-structural level, absolute spatial connectives were mainly used, as well as strong punctuation marks in writing and longer pauses in speech. At the micro- structural level, relations were mainly marked by relative spatial connectives. A grade level effect was observed, mainly in the written texts. However, results showed that, in writing as well as in speech, the management of the linguistic component of discourse linearization was, to a considerable degree, determined by the conceptual one. When producing a discourse, the writer has to deal with the critical issue of linearization, which covers both the conceptual choices that have to be made when ordering ideas (Levelt, 1981) and the linguistic specification of the relationship between * Correspondence should be addressed to Dr Monik Favart, Centre de Recherche Sur la Cognition et l’Apprentissage, Universite ´ de Poitiers, Poitiers 86000, France (e-mail: monik.favart@univ-poitiers.fr). The British Psychological Society 107 British Journal of Educational Psychology (2009), 79, 107–130 q 2009 The British Psychological Society www.bpsjournals.co.uk DOI:10.1348/000709908X289981