Tussen Vier en Acht

Some preliminary results from a normative study on spontaneous speech from children between four and eight years are presented. The study is part of a project to develop a standardized and normalized method of analysis of spontaneous speech samples to be used in the assessment of developmental language disorders. From four groups of 30 4-, 5-, 6-, and 7-year olds respectively, spontaneous speech samples have been collected from adult-child conversations. The groups are heterogeneous but comparable to each other with respect to social background. The samples have been segmented into -units. To reduce interviewer-effects, a distinction was made between elliptical answers (answers to previous questions, without finite verb) and free utterances (all other utterances). 50 consecutive free utterances per child have been analyzed on several variables with regard to length (9), fluency (22), grammatical, semantic and pragmatic accept-ability (18) , and syntactic and semantic complexity (43) . The resulting frequency curves (mean frequencies per 50 free utterances per age group) have been analyzed with respect to the following questions: (1) do the groups differ in the relative frequencies for these variables? (2) if so, can these differences be fitted into a model of gradual increase or decrease? (3) or, if there are sudden decreases and/or increases in frequency, how can they be explained? Differences between the age groups were found on many of the variables, and several of these differences indeed do not fit the gradual model. Sudden dips and jumps in the frequency curves do occur. Examples of non-monotonic frequency curves are given for fluency (fig.3 and 4.), prono-minalization (fig.5), coordination and past tense (fig.7), and for adverbial phrases of time (fig.8). Some explanations are suggested for these non-gradual developments. The stability of these frequency patterns is now being tested on a second group of 120 children. If developmental frequency curves prove to be fairly stable, intensive frequency pattern analysis on both cross-sectional and longitudinal data, taking into account situational variability, might be a way to trace the more subtle developments in the later stages of language acquisition.