Comments on the use of conversational features in studies of the vocalization behaviours of mothers and infants

Observations of six mother-infant dyads at infant ages of approximately 13 weeks, 20 weeks, 27 weeks and 35 weeks served as a data base to examine difficulties associated with the use of conversational features (such as the avoidance of coaction) in the analysis of mother-infant vocalization episodes. The difficulties are associated with procedures involving: (i) The pooling of data, (ii) The unit of analysis, and (iii) The selection of appropriate 'expected' values for dyadic parameters used in tests of statistical inference. To illustrate these difficulties, the use of the odds ratio statistic as a means of determining such dependence is examined and alternative approaches are suggested.